2015-02-04 17:44:23 -06:00
|
|
|
package terraform
|
|
|
|
|
2015-02-08 19:20:46 -06:00
|
|
|
import (
|
|
|
|
"fmt"
|
|
|
|
|
2019-09-09 17:58:44 -05:00
|
|
|
"github.com/hashicorp/hcl/v2"
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
"github.com/hashicorp/terraform/addrs"
|
|
|
|
"github.com/hashicorp/terraform/configs"
|
2018-08-16 10:40:08 -05:00
|
|
|
"github.com/hashicorp/terraform/configs/configschema"
|
|
|
|
"github.com/hashicorp/terraform/providers"
|
2018-08-17 11:42:07 -05:00
|
|
|
"github.com/hashicorp/terraform/provisioners"
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
"github.com/hashicorp/terraform/tfdiags"
|
|
|
|
"github.com/zclconf/go-cty/cty"
|
2018-06-22 18:54:27 -05:00
|
|
|
"github.com/zclconf/go-cty/cty/convert"
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
"github.com/zclconf/go-cty/cty/gocty"
|
2015-02-08 19:20:46 -06:00
|
|
|
)
|
|
|
|
|
|
|
|
// EvalValidateCount is an EvalNode implementation that validates
|
|
|
|
// the count of a resource.
|
|
|
|
type EvalValidateCount struct {
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
Resource *configs.Resource
|
2015-02-08 19:20:46 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
// TODO: test
|
2015-02-14 00:58:41 -06:00
|
|
|
func (n *EvalValidateCount) Eval(ctx EvalContext) (interface{}, error) {
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
var diags tfdiags.Diagnostics
|
2015-02-08 19:20:46 -06:00
|
|
|
var count int
|
|
|
|
var err error
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
|
|
|
|
val, valDiags := ctx.EvaluateExpr(n.Resource.Count, cty.Number, nil)
|
|
|
|
diags = diags.Append(valDiags)
|
|
|
|
if valDiags.HasErrors() {
|
2015-02-08 19:20:46 -06:00
|
|
|
goto RETURN
|
|
|
|
}
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
if val.IsNull() || !val.IsKnown() {
|
|
|
|
goto RETURN
|
2015-02-08 19:20:46 -06:00
|
|
|
}
|
|
|
|
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
err = gocty.FromCtyValue(val, &count)
|
|
|
|
if err != nil {
|
|
|
|
// The EvaluateExpr call above already guaranteed us a number value,
|
|
|
|
// so if we end up here then we have something that is out of range
|
|
|
|
// for an int, and the error message will include a description of
|
|
|
|
// the valid range.
|
|
|
|
rawVal := val.AsBigFloat()
|
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: "Invalid count value",
|
|
|
|
Detail: fmt.Sprintf("The number %s is not a valid count value: %s.", rawVal, err),
|
|
|
|
Subject: n.Resource.Count.Range().Ptr(),
|
|
|
|
})
|
|
|
|
} else if count < 0 {
|
|
|
|
rawVal := val.AsBigFloat()
|
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: "Invalid count value",
|
|
|
|
Detail: fmt.Sprintf("The number %s is not a valid count value: count must not be negative.", rawVal),
|
|
|
|
Subject: n.Resource.Count.Range().Ptr(),
|
|
|
|
})
|
2015-02-08 19:20:46 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
RETURN:
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
return nil, diags.NonFatalErr()
|
2015-02-08 19:20:46 -06:00
|
|
|
}
|
|
|
|
|
2020-09-14 07:43:14 -05:00
|
|
|
// EvalValidateProvisioner validates the configuration of a provisioner
|
|
|
|
// belonging to a resource. The provisioner config is expected to contain the
|
|
|
|
// merged connection configurations.
|
2015-02-09 13:15:54 -06:00
|
|
|
type EvalValidateProvisioner struct {
|
2019-07-29 02:18:33 -05:00
|
|
|
ResourceAddr addrs.Resource
|
|
|
|
Provisioner *provisioners.Interface
|
|
|
|
Schema **configschema.Block
|
|
|
|
Config *configs.Provisioner
|
|
|
|
ResourceHasCount bool
|
|
|
|
ResourceHasForEach bool
|
2015-02-09 13:15:54 -06:00
|
|
|
}
|
|
|
|
|
2020-09-14 07:43:14 -05:00
|
|
|
func (n *EvalValidateProvisioner) Validate(ctx EvalContext) error {
|
2015-02-14 00:58:41 -06:00
|
|
|
provisioner := *n.Provisioner
|
|
|
|
config := *n.Config
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
schema := *n.Schema
|
|
|
|
|
|
|
|
var diags tfdiags.Diagnostics
|
|
|
|
|
2020-09-14 07:43:14 -05:00
|
|
|
// Validate the provisioner's own config first
|
|
|
|
configVal, _, configDiags := n.evaluateBlock(ctx, config.Config, schema)
|
|
|
|
diags = diags.Append(configDiags)
|
|
|
|
if configDiags.HasErrors() {
|
|
|
|
return diags.Err()
|
|
|
|
}
|
2018-08-17 11:42:07 -05:00
|
|
|
|
2020-09-14 07:43:14 -05:00
|
|
|
if configVal == cty.NilVal {
|
|
|
|
// Should never happen for a well-behaved EvaluateBlock implementation
|
|
|
|
return fmt.Errorf("EvaluateBlock returned nil value")
|
2017-04-05 17:34:59 -05:00
|
|
|
}
|
|
|
|
|
2020-09-14 07:43:14 -05:00
|
|
|
req := provisioners.ValidateProvisionerConfigRequest{
|
|
|
|
Config: configVal,
|
2017-04-05 17:34:59 -05:00
|
|
|
}
|
|
|
|
|
2020-09-14 07:43:14 -05:00
|
|
|
resp := provisioner.ValidateProvisionerConfig(req)
|
|
|
|
diags = diags.Append(resp.Diagnostics)
|
|
|
|
|
|
|
|
// Now validate the connection config, which contains the merged bodies
|
|
|
|
// of the resource and provisioner connection blocks.
|
|
|
|
connDiags := n.validateConnConfig(ctx, config.Connection, n.ResourceAddr)
|
|
|
|
diags = diags.Append(connDiags)
|
|
|
|
|
|
|
|
return diags.NonFatalErr()
|
2015-02-09 13:15:54 -06:00
|
|
|
}
|
|
|
|
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
func (n *EvalValidateProvisioner) validateConnConfig(ctx EvalContext, config *configs.Connection, self addrs.Referenceable) tfdiags.Diagnostics {
|
2017-04-05 17:34:59 -05:00
|
|
|
// We can't comprehensively validate the connection config since its
|
|
|
|
// final structure is decided by the communicator and we can't instantiate
|
|
|
|
// that until we have a complete instance state. However, we *can* catch
|
|
|
|
// configuration keys that are not valid for *any* communicator, catching
|
|
|
|
// typos early rather than waiting until we actually try to run one of
|
|
|
|
// the resource's provisioners.
|
|
|
|
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
var diags tfdiags.Diagnostics
|
2017-04-05 17:34:59 -05:00
|
|
|
|
2018-05-11 16:58:39 -05:00
|
|
|
if config == nil || config.Config == nil {
|
|
|
|
// No block to validate
|
|
|
|
return diags
|
|
|
|
}
|
|
|
|
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
// We evaluate here just by evaluating the block and returning any
|
|
|
|
// diagnostics we get, since evaluation alone is enough to check for
|
|
|
|
// extraneous arguments and incorrectly-typed arguments.
|
2018-06-22 18:54:27 -05:00
|
|
|
_, _, configDiags := n.evaluateBlock(ctx, config.Config, connectionBlockSupersetSchema)
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
diags = diags.Append(configDiags)
|
2017-04-05 17:34:59 -05:00
|
|
|
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
return diags
|
|
|
|
}
|
2017-04-05 17:34:59 -05:00
|
|
|
|
2018-06-22 18:54:27 -05:00
|
|
|
func (n *EvalValidateProvisioner) evaluateBlock(ctx EvalContext, body hcl.Body, schema *configschema.Block) (cty.Value, hcl.Body, tfdiags.Diagnostics) {
|
|
|
|
keyData := EvalDataForNoInstanceKey
|
|
|
|
selfAddr := n.ResourceAddr.Instance(addrs.NoKey)
|
|
|
|
|
|
|
|
if n.ResourceHasCount {
|
|
|
|
// For a resource that has count, we allow count.index but don't
|
|
|
|
// know at this stage what it will return.
|
|
|
|
keyData = InstanceKeyEvalData{
|
|
|
|
CountIndex: cty.UnknownVal(cty.Number),
|
|
|
|
}
|
|
|
|
|
|
|
|
// "self" can't point to an unknown key, but we'll force it to be
|
|
|
|
// key 0 here, which should return an unknown value of the
|
|
|
|
// expected type since none of these elements are known at this
|
|
|
|
// point anyway.
|
|
|
|
selfAddr = n.ResourceAddr.Instance(addrs.IntKey(0))
|
2019-07-29 02:18:33 -05:00
|
|
|
} else if n.ResourceHasForEach {
|
|
|
|
// For a resource that has for_each, we allow each.value and each.key
|
|
|
|
// but don't know at this stage what it will return.
|
|
|
|
keyData = InstanceKeyEvalData{
|
|
|
|
EachKey: cty.UnknownVal(cty.String),
|
|
|
|
EachValue: cty.DynamicVal,
|
|
|
|
}
|
|
|
|
|
|
|
|
// "self" can't point to an unknown key, but we'll force it to be
|
|
|
|
// key "" here, which should return an unknown value of the
|
|
|
|
// expected type since none of these elements are known at
|
|
|
|
// this point anyway.
|
|
|
|
selfAddr = n.ResourceAddr.Instance(addrs.StringKey(""))
|
2018-06-22 18:54:27 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
return ctx.EvaluateBlock(body, schema, selfAddr, keyData)
|
|
|
|
}
|
|
|
|
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
// connectionBlockSupersetSchema is a schema representing the superset of all
|
|
|
|
// possible arguments for "connection" blocks across all supported connection
|
|
|
|
// types.
|
|
|
|
//
|
|
|
|
// This currently lives here because we've not yet updated our communicator
|
|
|
|
// subsystem to be aware of schema itself. Once that is done, we can remove
|
|
|
|
// this and use a type-specific schema from the communicator to validate
|
|
|
|
// exactly what is expected for a given connection type.
|
|
|
|
var connectionBlockSupersetSchema = &configschema.Block{
|
|
|
|
Attributes: map[string]*configschema.Attribute{
|
2018-05-10 19:18:17 -05:00
|
|
|
// NOTE: "type" is not included here because it's treated special
|
|
|
|
// by the config loader and stored away in a separate field.
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
|
|
|
|
// Common attributes for both connection types
|
2018-12-19 14:20:23 -06:00
|
|
|
"host": {
|
|
|
|
Type: cty.String,
|
|
|
|
Required: true,
|
|
|
|
},
|
2018-05-29 14:00:16 -05:00
|
|
|
"type": {
|
|
|
|
Type: cty.String,
|
|
|
|
Optional: true,
|
|
|
|
},
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
"user": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"password": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"port": {
|
2018-05-29 14:00:16 -05:00
|
|
|
Type: cty.String,
|
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"timeout": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"script_path": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
2017-04-05 17:34:59 -05:00
|
|
|
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
// For type=ssh only (enforced in ssh communicator)
|
|
|
|
"private_key": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
2018-09-18 19:36:39 -05:00
|
|
|
"certificate": {
|
|
|
|
Type: cty.String,
|
|
|
|
Optional: true,
|
|
|
|
},
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
"host_key": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"agent": {
|
|
|
|
Type: cty.Bool,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"agent_identity": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"bastion_host": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"bastion_host_key": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"bastion_port": {
|
|
|
|
Type: cty.Number,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"bastion_user": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"bastion_password": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"bastion_private_key": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
2019-07-21 01:32:48 -05:00
|
|
|
"bastion_certificate": {
|
|
|
|
Type: cty.String,
|
|
|
|
Optional: true,
|
|
|
|
},
|
2017-04-05 17:34:59 -05:00
|
|
|
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
// For type=winrm only (enforced in winrm communicator)
|
|
|
|
"https": {
|
|
|
|
Type: cty.Bool,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"insecure": {
|
|
|
|
Type: cty.Bool,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"cacert": {
|
|
|
|
Type: cty.String,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
"use_ntlm": {
|
|
|
|
Type: cty.Bool,
|
2018-05-29 14:00:16 -05:00
|
|
|
Optional: true,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
},
|
|
|
|
},
|
2017-04-05 17:34:59 -05:00
|
|
|
}
|
|
|
|
|
2019-02-21 20:10:31 -06:00
|
|
|
// connectionBlockSupersetSchema is a schema representing the superset of all
|
|
|
|
// possible arguments for "connection" blocks across all supported connection
|
|
|
|
// types.
|
|
|
|
//
|
|
|
|
// This currently lives here because we've not yet updated our communicator
|
|
|
|
// subsystem to be aware of schema itself. It's exported only for use in the
|
|
|
|
// configs/configupgrade package and should not be used from anywhere else.
|
|
|
|
// The caller may not modify any part of the returned schema data structure.
|
|
|
|
func ConnectionBlockSupersetSchema() *configschema.Block {
|
|
|
|
return connectionBlockSupersetSchema
|
|
|
|
}
|
|
|
|
|
2020-09-14 07:43:14 -05:00
|
|
|
// EvalValidateResource validates the configuration of a resource.
|
2015-02-04 17:44:23 -06:00
|
|
|
type EvalValidateResource struct {
|
2018-06-22 18:54:27 -05:00
|
|
|
Addr addrs.Resource
|
2018-08-16 10:40:08 -05:00
|
|
|
Provider *providers.Interface
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
ProviderSchema **ProviderSchema
|
|
|
|
Config *configs.Resource
|
2020-03-05 18:53:24 -06:00
|
|
|
ProviderMetas map[addrs.Provider]*configs.ProviderMeta
|
2016-06-30 18:22:20 -05:00
|
|
|
|
|
|
|
// IgnoreWarnings means that warnings will not be passed through. This allows
|
|
|
|
// "just-in-time" passes of validation to continue execution through warnings.
|
|
|
|
IgnoreWarnings bool
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
|
|
|
|
// ConfigVal, if non-nil, will be updated with the value resulting from
|
|
|
|
// evaluating the given configuration body. Since validation is performed
|
|
|
|
// very early, this value is likely to contain lots of unknown values,
|
|
|
|
// but its type will conform to the schema of the resource type associated
|
|
|
|
// with the resource instance being validated.
|
|
|
|
ConfigVal *cty.Value
|
2015-02-04 17:44:23 -06:00
|
|
|
}
|
|
|
|
|
2020-09-14 07:43:14 -05:00
|
|
|
func (n *EvalValidateResource) Validate(ctx EvalContext) error {
|
2018-05-04 21:56:40 -05:00
|
|
|
if n.ProviderSchema == nil || *n.ProviderSchema == nil {
|
2020-09-14 07:43:14 -05:00
|
|
|
return fmt.Errorf("EvalValidateResource has nil schema for %s", n.Addr)
|
2018-05-04 21:43:28 -05:00
|
|
|
}
|
|
|
|
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
var diags tfdiags.Diagnostics
|
2015-02-14 00:58:41 -06:00
|
|
|
provider := *n.Provider
|
|
|
|
cfg := *n.Config
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
schema := *n.ProviderSchema
|
|
|
|
mode := cfg.Mode
|
|
|
|
|
2018-06-22 18:54:27 -05:00
|
|
|
keyData := EvalDataForNoInstanceKey
|
2020-04-08 16:09:17 -05:00
|
|
|
|
|
|
|
switch {
|
|
|
|
case n.Config.Count != nil:
|
2018-06-22 18:54:27 -05:00
|
|
|
// If the config block has count, we'll evaluate with an unknown
|
|
|
|
// number as count.index so we can still type check even though
|
|
|
|
// we won't expand count until the plan phase.
|
|
|
|
keyData = InstanceKeyEvalData{
|
|
|
|
CountIndex: cty.UnknownVal(cty.Number),
|
|
|
|
}
|
|
|
|
|
|
|
|
// Basic type-checking of the count argument. More complete validation
|
|
|
|
// of this will happen when we DynamicExpand during the plan walk.
|
|
|
|
countDiags := n.validateCount(ctx, n.Config.Count)
|
|
|
|
diags = diags.Append(countDiags)
|
|
|
|
|
2020-04-08 16:09:17 -05:00
|
|
|
case n.Config.ForEach != nil:
|
2019-06-12 10:07:32 -05:00
|
|
|
keyData = InstanceKeyEvalData{
|
|
|
|
EachKey: cty.UnknownVal(cty.String),
|
|
|
|
EachValue: cty.UnknownVal(cty.DynamicPseudoType),
|
|
|
|
}
|
|
|
|
|
|
|
|
// Evaluate the for_each expression here so we can expose the diagnostics
|
2019-07-25 10:51:55 -05:00
|
|
|
forEachDiags := n.validateForEach(ctx, n.Config.ForEach)
|
2019-06-12 10:07:32 -05:00
|
|
|
diags = diags.Append(forEachDiags)
|
|
|
|
}
|
|
|
|
|
2020-06-16 11:38:05 -05:00
|
|
|
diags = diags.Append(validateDependsOn(ctx, n.Config.DependsOn))
|
2018-07-02 12:23:41 -05:00
|
|
|
|
2020-03-05 18:53:24 -06:00
|
|
|
// Validate the provider_meta block for the provider this resource
|
|
|
|
// belongs to, if there is one.
|
|
|
|
//
|
|
|
|
// Note: this will return an error for every resource a provider
|
|
|
|
// uses in a module, if the provider_meta for that module is
|
|
|
|
// incorrect. The only way to solve this that we've foudn is to
|
|
|
|
// insert a new ProviderMeta graph node in the graph, and make all
|
|
|
|
// that provider's resources in the module depend on the node. That's
|
|
|
|
// an awful heavy hammer to swing for this feature, which should be
|
|
|
|
// used only in limited cases with heavy coordination with the
|
|
|
|
// Terraform team, so we're going to defer that solution for a future
|
|
|
|
// enhancement to this functionality.
|
|
|
|
/*
|
|
|
|
if n.ProviderMetas != nil {
|
|
|
|
if m, ok := n.ProviderMetas[n.ProviderAddr.ProviderConfig.Type]; ok && m != nil {
|
|
|
|
// if the provider doesn't support this feature, throw an error
|
|
|
|
if (*n.ProviderSchema).ProviderMeta == nil {
|
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: fmt.Sprintf("Provider %s doesn't support provider_meta", cfg.ProviderConfigAddr()),
|
|
|
|
Detail: fmt.Sprintf("The resource %s belongs to a provider that doesn't support provider_meta blocks", n.Addr),
|
|
|
|
Subject: &m.ProviderRange,
|
|
|
|
})
|
|
|
|
} else {
|
|
|
|
_, _, metaDiags := ctx.EvaluateBlock(m.Config, (*n.ProviderSchema).ProviderMeta, nil, EvalDataForNoInstanceKey)
|
|
|
|
diags = diags.Append(metaDiags)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
*/
|
|
|
|
// BUG(paddy): we're not validating provider_meta blocks on EvalValidate right now
|
|
|
|
// because the ProviderAddr for the resource isn't available on the EvalValidate
|
|
|
|
// struct.
|
|
|
|
|
2016-05-01 21:01:48 -05:00
|
|
|
// Provider entry point varies depending on resource mode, because
|
|
|
|
// managed resources and data resources are two distinct concepts
|
|
|
|
// in the provider abstraction.
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
switch mode {
|
|
|
|
case addrs.ManagedResourceMode:
|
2018-11-27 17:30:18 -06:00
|
|
|
schema, _ := schema.SchemaForResourceType(mode, cfg.Type)
|
|
|
|
if schema == nil {
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: "Invalid resource type",
|
|
|
|
Detail: fmt.Sprintf("The provider %s does not support resource type %q.", cfg.ProviderConfigAddr(), cfg.Type),
|
|
|
|
Subject: &cfg.TypeRange,
|
|
|
|
})
|
2020-09-14 07:43:14 -05:00
|
|
|
return diags.Err()
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
}
|
|
|
|
|
2018-06-22 15:04:19 -05:00
|
|
|
configVal, _, valDiags := ctx.EvaluateBlock(cfg.Config, schema, nil, keyData)
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
diags = diags.Append(valDiags)
|
|
|
|
if valDiags.HasErrors() {
|
2020-09-14 07:43:14 -05:00
|
|
|
return diags.Err()
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
}
|
|
|
|
|
2018-12-14 19:14:17 -06:00
|
|
|
if cfg.Managed != nil { // can be nil only in tests with poorly-configured mocks
|
|
|
|
for _, traversal := range cfg.Managed.IgnoreChanges {
|
2020-09-25 15:59:58 -05:00
|
|
|
// This will error out if the traversal contains an invalid
|
|
|
|
// index step. That is OK if we want users to be able to ignore
|
|
|
|
// a key that is no longer specified in the config.
|
2018-12-14 19:14:17 -06:00
|
|
|
moreDiags := schema.StaticValidateTraversal(traversal)
|
|
|
|
diags = diags.Append(moreDiags)
|
2020-09-25 15:59:58 -05:00
|
|
|
if diags.HasErrors() {
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
|
|
|
|
// first check to see if this assigned in the config
|
|
|
|
v, _ := traversal.TraverseRel(configVal)
|
|
|
|
if !v.IsNull() {
|
|
|
|
// it's assigned, so we can also assume it's not computed-only
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
|
|
|
|
// We can't ignore changes that don't exist in the configuration.
|
|
|
|
// We're not checking specifically if the traversal resolves to
|
|
|
|
// a computed-only value, but we can hint to the user that it
|
|
|
|
// might also be the case.
|
|
|
|
sourceRange := traversal.SourceRange()
|
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: "Cannot ignore argument not set in the configuration",
|
|
|
|
Detail: fmt.Sprintf("The ignore_changes argument is not set in the configuration.\n" +
|
|
|
|
"The ignore_changes mechanism only applies to changes " +
|
|
|
|
"within the configuration, and must be used with " +
|
|
|
|
"arguments set in the configuration and not computed by " +
|
|
|
|
"the provider.",
|
|
|
|
),
|
|
|
|
Subject: &sourceRange,
|
|
|
|
})
|
|
|
|
return diags.Err()
|
2018-12-14 19:14:17 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-08-16 10:40:08 -05:00
|
|
|
req := providers.ValidateResourceTypeConfigRequest{
|
|
|
|
TypeName: cfg.Type,
|
|
|
|
Config: configVal,
|
|
|
|
}
|
|
|
|
|
|
|
|
resp := provider.ValidateResourceTypeConfig(req)
|
2018-08-24 19:46:19 -05:00
|
|
|
diags = diags.Append(resp.Diagnostics.InConfigBody(cfg.Config))
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
|
|
|
|
if n.ConfigVal != nil {
|
|
|
|
*n.ConfigVal = configVal
|
|
|
|
}
|
|
|
|
|
|
|
|
case addrs.DataResourceMode:
|
2018-11-27 17:30:18 -06:00
|
|
|
schema, _ := schema.SchemaForResourceType(mode, cfg.Type)
|
|
|
|
if schema == nil {
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: "Invalid data source",
|
|
|
|
Detail: fmt.Sprintf("The provider %s does not support data source %q.", cfg.ProviderConfigAddr(), cfg.Type),
|
|
|
|
Subject: &cfg.TypeRange,
|
|
|
|
})
|
2020-09-14 07:43:14 -05:00
|
|
|
return diags.Err()
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
}
|
|
|
|
|
2018-06-22 15:04:19 -05:00
|
|
|
configVal, _, valDiags := ctx.EvaluateBlock(cfg.Config, schema, nil, keyData)
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
diags = diags.Append(valDiags)
|
|
|
|
if valDiags.HasErrors() {
|
2020-09-14 07:43:14 -05:00
|
|
|
return diags.Err()
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
}
|
2015-02-09 11:50:20 -06:00
|
|
|
|
2018-08-16 10:40:08 -05:00
|
|
|
req := providers.ValidateDataSourceConfigRequest{
|
|
|
|
TypeName: cfg.Type,
|
|
|
|
Config: configVal,
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
}
|
2015-02-08 19:58:02 -06:00
|
|
|
|
2018-08-16 10:40:08 -05:00
|
|
|
resp := provider.ValidateDataSourceConfig(req)
|
2018-08-24 19:46:19 -05:00
|
|
|
diags = diags.Append(resp.Diagnostics.InConfigBody(cfg.Config))
|
2015-02-08 19:58:02 -06:00
|
|
|
}
|
|
|
|
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
if n.IgnoreWarnings {
|
|
|
|
// If we _only_ have warnings then we'll return nil.
|
|
|
|
if diags.HasErrors() {
|
2020-09-14 07:43:14 -05:00
|
|
|
return diags.NonFatalErr()
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
}
|
2020-09-14 07:43:14 -05:00
|
|
|
return nil
|
terraform: ugly huge change to weave in new HCL2-oriented types
Due to how deeply the configuration types go into Terraform Core, there
isn't a great way to switch out to HCL2 gradually. As a consequence, this
huge commit gets us from the old state to a _compilable_ new state, but
does not yet attempt to fix any tests and has a number of known missing
parts and bugs. We will continue to iterate on this in forthcoming
commits, heading back towards passing tests and making Terraform
fully-functional again.
The three main goals here are:
- Use the configuration models from the "configs" package instead of the
older models in the "config" package, which is now deprecated and
preserved only to help us write our migration tool.
- Do expression inspection and evaluation using the functionality of the
new "lang" package, instead of the Interpolator type and related
functionality in the main "terraform" package.
- Represent addresses of various objects using types in the addrs package,
rather than hand-constructed strings. This is not critical to support
the above, but was a big help during the implementation of these other
points since it made it much more explicit what kind of address is
expected in each context.
Since our new packages are built to accommodate some future planned
features that are not yet implemented (e.g. the "for_each" argument on
resources, "count"/"for_each" on modules), and since there's still a fair
amount of functionality still using old-style APIs, there is a moderate
amount of shimming here to connect new assumptions with old, hopefully in
a way that makes it easier to find and eliminate these shims later.
I apologize in advance to the person who inevitably just found this huge
commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
|
|
|
} else {
|
|
|
|
// We'll return an error if there are any diagnostics at all, even if
|
|
|
|
// some of them are warnings.
|
2020-09-14 07:43:14 -05:00
|
|
|
return diags.NonFatalErr()
|
2015-02-08 19:58:02 -06:00
|
|
|
}
|
2015-02-04 17:44:23 -06:00
|
|
|
}
|
2018-06-22 18:54:27 -05:00
|
|
|
|
|
|
|
func (n *EvalValidateResource) validateCount(ctx EvalContext, expr hcl.Expression) tfdiags.Diagnostics {
|
|
|
|
if expr == nil {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
|
|
|
var diags tfdiags.Diagnostics
|
|
|
|
|
|
|
|
countVal, countDiags := ctx.EvaluateExpr(expr, cty.Number, nil)
|
|
|
|
diags = diags.Append(countDiags)
|
|
|
|
if diags.HasErrors() {
|
|
|
|
return diags
|
|
|
|
}
|
|
|
|
|
|
|
|
if countVal.IsNull() {
|
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: "Invalid count argument",
|
|
|
|
Detail: `The given "count" argument value is null. An integer is required.`,
|
|
|
|
Subject: expr.Range().Ptr(),
|
|
|
|
})
|
|
|
|
return diags
|
|
|
|
}
|
|
|
|
|
|
|
|
var err error
|
|
|
|
countVal, err = convert.Convert(countVal, cty.Number)
|
|
|
|
if err != nil {
|
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: "Invalid count argument",
|
|
|
|
Detail: fmt.Sprintf(`The given "count" argument value is unsuitable: %s.`, err),
|
|
|
|
Subject: expr.Range().Ptr(),
|
|
|
|
})
|
|
|
|
return diags
|
|
|
|
}
|
|
|
|
|
|
|
|
// If the value isn't known then that's the best we can do for now, but
|
|
|
|
// we'll check more thoroughly during the plan walk.
|
|
|
|
if !countVal.IsKnown() {
|
|
|
|
return diags
|
|
|
|
}
|
|
|
|
|
|
|
|
// If we _do_ know the value, then we can do a few more checks here.
|
|
|
|
var count int
|
|
|
|
err = gocty.FromCtyValue(countVal, &count)
|
|
|
|
if err != nil {
|
|
|
|
// Isn't a whole number, etc.
|
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: "Invalid count argument",
|
|
|
|
Detail: fmt.Sprintf(`The given "count" argument value is unsuitable: %s.`, err),
|
|
|
|
Subject: expr.Range().Ptr(),
|
|
|
|
})
|
|
|
|
return diags
|
|
|
|
}
|
|
|
|
|
|
|
|
if count < 0 {
|
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: "Invalid count argument",
|
|
|
|
Detail: `The given "count" argument value is unsuitable: count cannot be negative.`,
|
|
|
|
Subject: expr.Range().Ptr(),
|
|
|
|
})
|
|
|
|
return diags
|
|
|
|
}
|
|
|
|
|
|
|
|
return diags
|
|
|
|
}
|
2019-07-25 10:51:55 -05:00
|
|
|
|
|
|
|
func (n *EvalValidateResource) validateForEach(ctx EvalContext, expr hcl.Expression) (diags tfdiags.Diagnostics) {
|
2020-04-08 16:09:17 -05:00
|
|
|
val, forEachDiags := evaluateForEachExpressionValue(expr, ctx)
|
2019-07-25 10:51:55 -05:00
|
|
|
// If the value isn't known then that's the best we can do for now, but
|
|
|
|
// we'll check more thoroughly during the plan walk
|
2020-04-08 16:09:17 -05:00
|
|
|
if !val.IsKnown() {
|
2019-07-25 10:51:55 -05:00
|
|
|
return diags
|
|
|
|
}
|
|
|
|
|
|
|
|
if forEachDiags.HasErrors() {
|
|
|
|
diags = diags.Append(forEachDiags)
|
|
|
|
}
|
|
|
|
|
|
|
|
return diags
|
|
|
|
}
|
2020-06-16 11:38:05 -05:00
|
|
|
|
|
|
|
func validateDependsOn(ctx EvalContext, dependsOn []hcl.Traversal) (diags tfdiags.Diagnostics) {
|
|
|
|
for _, traversal := range dependsOn {
|
|
|
|
ref, refDiags := addrs.ParseRef(traversal)
|
|
|
|
diags = diags.Append(refDiags)
|
|
|
|
if !refDiags.HasErrors() && len(ref.Remaining) != 0 {
|
|
|
|
diags = diags.Append(&hcl.Diagnostic{
|
|
|
|
Severity: hcl.DiagError,
|
|
|
|
Summary: "Invalid depends_on reference",
|
|
|
|
Detail: "References in depends_on must be to a whole object (resource, etc), not to an attribute of an object.",
|
|
|
|
Subject: ref.Remaining.SourceRange().Ptr(),
|
|
|
|
})
|
|
|
|
}
|
|
|
|
|
|
|
|
// The ref must also refer to something that exists. To test that,
|
|
|
|
// we'll just eval it and count on the fact that our evaluator will
|
|
|
|
// detect references to non-existent objects.
|
|
|
|
if !diags.HasErrors() {
|
|
|
|
scope := ctx.EvaluationScope(nil, EvalDataForNoInstanceKey)
|
|
|
|
if scope != nil { // sometimes nil in tests, due to incomplete mocks
|
|
|
|
_, refDiags = scope.EvalReference(ref, cty.DynamicPseudoType)
|
|
|
|
diags = diags.Append(refDiags)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return diags
|
|
|
|
}
|