em_folder_tree_get_selected_folder() currently blocks, and I'm about to
remove the function. But there's a couple places where we just need to
block for now, else it would require a significant rewrite. So execute
the em_folder_tree_get_selected_folder() logic manually in those places.
This will force Evolution to reconnect to the service using the current
settings. However this is not a complete solution. If the new settings
now point to a completely different mail account, we leave behind cached
messages and database tables from the previous account such that you end
up with some weird hybrid of the previous account and current account.
I guess for now the answer is "don't do that", but we should try to
handle that more gracefully in the future -- more for architectural
correctness than it being a common real world use case.
Emitted after all changes are successfully committed over D-Bus.
Can connect to this signal to perform followup actions, such as
disconnecting the corresponding CamelService.
This function removed a CamelStore from EMFolderTreeModel and then added
it again in response to the corresponding ESource emitting a "changed"
signal (presumably from editing a mail account and saving changes).
In the past this was the only way for changes to the account URI string
to take effect without restarting Evolution. But we have CamelSettings
now and don't need this hack anymore. Time for it to die.
Save the version after the startup wizard has had a chance to
run. If the user chooses to restore data and settings from a
backup, Evolution will restart and the restored data may need
to be migrated.
If we save the version before the restart, then Evolution will
think it has already migrated data and settings to the current
version and the restored data may not be handled properly.
Looks like some ancient development environment script.
I actually use something very similar for my own development
environment, but it doesn't belong in a version control system.
Not the next stable version. If migration needs to occur multiple times
during a development cycle for different reasons, we'll need an accurate
last-used-version stamp.
AFAICT, this key does nothing useful and only confuses me every time I
read the EShell migration code.
The "version" key records the most recently used Evolution version.
That's all we need for migration. And since downgrading Evolution is
not supported, we can assume this value will only increase over time.
This came about from a mailing list thread entitled:
Sending photos from digKam: "Skipping suspicious attachment"
Apparently KDE still uses dot-folders quite heavily, despite having
helped draft the XDG Base Directory Specification.
Openning the Preferences dialog for the first time causes many
property bindings to be created. GLib emits a lot of ::notify signals
during this, one of them being about image-loading-policy.
We now correctly check, whether the change was to allow loading
images, or not (and we don't display them in that case).
Finaly, e_mail_display_reload() will always schedule the action via
g_idle_add() so that we can merge plenty of _reload() calls caused by
the afformentioned ::notify emitions into single call when all the noise
is over.
Some contacts have photos stored locally as files. Until now
EContactEditor displayed thumbnail of contact photo in EImageChooser
only if the photo was embedded in the vCard as data. This patch
adds support for dispalying thumbnails of photos stored in local files
to indicate that the contact has a picture set.
EContactPhoto URIs are already escaped, but WebKit escapes it again, so images
are not displayed sometimes. Unescaping the URI before writing it to the HTML
fixes to problem.
Fix regression from commit 99a875ed which has broken displaying
of suppressed HTML parts as attachments when the HTML part is embedded
in a multipart/* container.
