In 1bd8a075, a hidden site setting was added that causes Email::Styles
to treat its input as a complete document in all cases.
This commit enables that setting by default.
Some tests were removed that were broken by this change. They tested the
behaviour of applying email styles to empty strings. They weren't useful
because:
* Sending empty email is not something we ever intend to do,
* They were testing incidental behaviour - there are lots of
valid ways to process the empty string,
* Their intent wasn't clear from their descriptions,
`Nokogiri::HTML.fragment` is a huge hack (a comment in the source code
admits this). The current behavior of `Email::Styles` is to try to
emulate `fragment` using nokogumbo, but it misses some edge cases. In
particular, meta tags in a email template don't make it through to the
final email.
Instead of treating the provided HTML as an indeterminate fragment, this
commit makes `Email::Styles` treat the HTML as a complete document. This
means that the generated HTML for an email will now always contain top
level structure (a doctype, html, head and body tags).
This new behavior is behind a hidden site setting for now and defaults
off.
This is so that, on a multisite cluster, when we handle a CDN request,
the hostname that is requested corresponds to one of the sites -
specifically the default site.
Latest version of Rails contains compatibility fixes for Ruby 2.7 and some
minor security fixes we would like to have
It also broke some of the multisite tests.
Rails tries to use the same connection for reading from a replica as writing
to the leader during tests, because, with everything happening in a
transaction, changes to the DB wouldn't otherwise be reflected in the
replica connection.
The difference now is that Rails tries to do this for connections opened
after the test has started which affected rails multisite connections.
The upshot of this is that, as things stand, you are likely to
experience problems if you try to connect to a different multisite DB in
a test when the `current_db` is not 'default'.
The process_post job uses CookedPostProcessor which also uses a
DistributedMutex. There's no good reason for the timeout of the outer
lock to be smaller than the timeout of the inner lock.
In development, if the ApplicationController is reloaded, then, previous
to this commit we were emitting an instance of the previous RenderEmpty
class, but rescuing from the reloaded instance.
Looking up RenderEmpty by its fully qualified name fixes this.
Two behaviors in the mail gem collide:
1. Attachments are added as extra parts at the top level,
2. When there are both text and html parts, the content type is set to
'multipart/alternative'.
Since attachments aren't alternative renderings, for emails that contain
attachments and both html and text parts, some coercing is necessary.
On startup, (including when starting a rails console) we manipule a
collection of plugin files. Writing these files is done in multiple
observable steps, which presents opportunities for race conditions and
causes temporary corruption.
This commit uses the write, fsync and rename trick to atomically
overwrite these files instead, but reads them first to avoid unnecessary
writes.
c457d3bf was a previous attempt to fix the same problem.
Accounting for fractional seconds, a distributed mutex can be held for
almost a full second longer than its validity.
For example: if we grab the lock at 10.5 seconds passed the epoch with a
validity of 5 seconds, the lock would be released at 16 seconds passed
the epoch. However, in this case assuming that all other processing
takes a negligible amount of time, the key would be expired at 15.5
seconds passed the epoch.
Using expireat, the key is now expired exactly when the lock is released.
The trouble with having:
/tags/:tag_id/...
and:
/tags/intersection/*tag_ids
for example, is: what happens if you want a tag called intersection?
Under this new scheme. Routes referring to a single tag are unambiguous
because they are prefixed with:
/tag/:tag_id
Routes referring to the collection of tags still start with:
/tags/
This commit just adds the new routes. It doesn't remove the old ones or
cause the new ones to be used.
The tag-drop component uses the passed in categories to calculate paths,
but only the last category is relevant, since, from a category we can
calculate its ancestors.
This commit corresponds to d84c34ad which applies the same changes to
the server-side. This changes the category routes, except for the routes
that contain tags.
When we receive a list of categories, we should store them so that we
are able to query them in the browser without a page refresh.
This removes a previous fix for the same issue that was much less
general.
The plan is to switch over all cases where filterMode is set, but, to
prevent breaking plugins and themes, the category and noSubcategories
need to be inferred from any given filterMode before that can happen.
According to eviltrout, commas and pipes were the delimiters of a
Discourse specific microformat, but this is no longer the case and
hasn't been for some time.
These are the changes to tags routing that correspond to the category
routing changes of d84c34ad. The new scheme is:
/tags/c/*slug_path/:id/none/:tag_id/ENDPOINT
/tags/c/*slug_path/:id/none/:tag_id
/tags/c/*slug_path/:id/:tag_id/ENDPOINT
/tags/c/*slug_path/:id/:tag_id
Also:
Move includes call higher which makes it possible to run all of the
intermediate queries for easier debugging.
Add tests for TagsController with categories in the path.
The routes for categories are changing. The scheme that I intend to move
us to is:
/c/*slug_path/(:id)/ENDPOINT
/c/*slug_path/(:id)
This commit adds support for the new scheme to the server side without
dropping support for existing URLs. It is necessary to support existing
URLs for two reasons:
* This commit does not change any client side routing code,
* Posts that contain category hashtags that refer to a root category
are baked into URLs that do not fit this new scheme, (/c/[id]-[slug])
If we are searching for categories by their slugs, it doesn't make sense
to include subcategories since a slug, by itself, does not necessarily
uniquely identify a subcategory.
Similarly, the empty string as a slug is not a good category identifier.
Daniel Waterworth