When loading commodities erroneously made in namespace template,
gnc_commodity_table_insert changes the namespace to User. This
change wasn't being committed, requiring that it be changed every
time.
The obvious immediate commit didn't work because the SQL backend
suppresses commits during load for performance, so create a list
of changed commodities to commit later, then do so.
Thanks to Geert for noticing the problem and to Derek for suggesting
the list workaround.
The file location is now visible in the status bar
while hovering over the recent file list of the File menu.
The currently open file is the top-most file in this list so its
location can be seen by hovering over it with the mouse.
* Bug 711440 - Tab labels have different background colour than containing gui element
* Bug 434462 - register color don't work correct with system theme color
* Add scrub function for (business) lots to eliminate unneeded lot links
* Rewrite invoice payment logic to use lot links only when absolutely needed
* Bug 684719 - Man pages for gnc-fq-* perl scripts
- most payments will behave is it did in 2.4.x
- when two or more documents are involved in the payment
a lot link between those document lots will be created
When used in a strict sense it will only merge splits
that are marked as subsplits of each other.
In relaxed mode it will merge any split two splits that
are both in the same lot and the same transaction.
This function is currently only used in the owner tree code and this change
will cause payments not properly linked to invoices to be ignored when computing
owner balances.
As discussed in
https://lists.gnucash.org/pipermail/gnucash-devel/2014-July/037868.html
as an exception the direct access to the kvp is enabled again in
import-export/aqb/gnc-ab-kvp.c because the way aqbanking transfer templates
are stored cannot be mapped directly on the properties mechanism.
With this implementation, at least the read access to the templates
works again, which means the aqbanking transfers can be used again (they
crashed immediately up until now).
However, writing a changed template to the kvp seems to not work - the
changed values seem not to be saved. This would be a bug, but not as severe
as the previous one. The unittest verifies both reading and writing,
though, and there also writing works fine. Well, this can be worked
on later.
