Replace it with new async functions:
e_mail_store_go_offline()
e_mail_store_go_offline_finish()
e_mail_store_go_online()
e_mail_store_go_online_finish()
With unintrusive error dialogs gone, we can cut some unnecessary bits
out of EActivity.
I'm also adding a new enum property called "state", which is one of:
E_ACTIVITY_RUNNING
E_ACTIVITY_WAITING
E_ACTIVITY_CANCELLED
E_ACTIVITY_COMPLETED
The state of an activity must be explicitly changed. In particular,
when the user cancels an activity the state should be set only after
confirming the operation has been cancelled and not when cancellation
is requested (e.g. after receiving a G_IO_ERROR_CANCELLED, not when
the GCancellable emits "cancelled"). EActivityBar and EActivityProxy
widgets have been updated to make this distinction clearer in the UI.
E_ACTIVITY_WAITING will be used when activities have to be queued and
dispatched in sequence, which I haven't written yet.
And generate GTypes for each of them in e-mail-enumtypes.[ch].
Also, the glib-gen.mak script forced me to add a <mail/e-mail.h>
top-level header, which really isn't a bad idea anyway.
TODO: We should do this for calendar and addressbook too.
This marks the end of unintrusive error dialogs, which were too
unintrusive. We now show errors directly in the main window using
the EAlert / EAlertSink framework.
Global variables in shared libraries are a bad idea. EMailBackend now
owns the MailSession instance, which is actually now EMailSession.
Move the blocking utility functions in mail-tools.c to e-mail-session.c
and add asynchronous variants. Same approach as Camel.
Replace EMailReader.get_shell_backend() with EMailReader.get_backend(),
which returns an EMailBackend. Easier access to the EMailSession.
This pushes the get_data_dir() API down to the right level. At present, it is
still implemented by querying the shell backend for the data dir / config dir.
But this should eventually be reversed (when mail is split off to EDS) so that
the mail daemon is the one responsible for the storage locations and the shell
backend queries the daemon for these values.
Previously I was just using G_TYPE_POINTER. Use the boxed camel object type
from e-util.h instead. When camel-gobject lands, we'll use G_TYPE_OBJECT
instead.
Yes, this signal is kind of an ugly monster. I'm not sure how to improve this
significantly. But this commit removes the last EMFolderTreeModel and EShell
dependencies from MailFolderCache, which is a big step towards splitting off
the backend.
https://bugzilla.gnome.org/show_bug.cgi?id=604627
Emit a signal when we have an updated unread count for a folder rather than
pushing the update directly to a particular treemodel. This doesn't yet remove
the dependency on EMFolderTreeModel, but it's a first step.
https://bugzilla.gnome.org/show_bug.cgi?id=604627
Instead of pushing the updates to the right places, the folder cache simply
emits the appropriate signals and other objects are responsible for listening
and handling them appropriately. This allows us to cut down the dependencies of
MailFolderCache significantly, which is a huge step towards allowing us to split
it off for the backend.
Another nice thing about this is that it allows us to trim a lot of 'public' api
from the filter, vfolder, and config classes that were only used by the cache.
Now that stuff can all be internal since they're pulling changes rather than
having the changes pushed.
The last remaining problematic dependency in MailFolderCache is
EmFolderTreeModel. That is next on the chopping block.
https://bugzilla.gnome.org/show_bug.cgi?id=604627
This will allow us to decouple ourselves from some of the current dependencies,
such as the folder treemodel, the shell, etc. This just defines the signals,
the next step is to refactor things and actually make other classes use them.
We need one additional signal yet related to indicating the new unread emails,
but that one will require a little more thought I think.
https://bugzilla.gnome.org/show_bug.cgi?id=604627
Matthew Barnes