http://live.gnome.org/GnomeGoals/RemoveLibGladeUseGtkBuilder

*** Bug 590820 has been marked as a duplicate of this bug. ***

I've changed severity to normal because libglade is deprecated since GNOME 2.27

Again: gnome-pilot is not official part of GNOME. Hence enhancement and up to the maintainers to migrate or not, though we highly recommend it.

Created attachment 142304 [details] [review]
Converts the applet from libglade to GtkBuilder

Created attachment 142305 [details] [review]
Converts the capplet from libglade to GtkBuilder

Work in progress for converting the capplet from libglade to GtkBuilder.  The major part of the work is converting from a Druid to a GtkAssistant.  Others are welcome to update this patch.

Hi Robert,
Thanks for the patch.  Can you give a bit more detail as to:
  1.  the level of completeness
  2.  whether you are still working on this patch or handing it over

Matt
(I'd started work on this, but it looks like you got a bit further.  I was most of the way through the removal of GtkOptionMenu and friends.  It's not easy to track changes though .glade files, is it? :)

Hi Matt,

The first patch is complete, the second about 50% done.  I do still plan to work on it but it is low priority so it may take some time. (I will check back here for progress before continuing).

I don't have an appropriate device to test so review/fixes to the patch would help.

Yeah, Glade can be a bit of a beast to work with :)

Comment on attachment 142304 [details] [review]
Converts the applet from libglade to GtkBuilder

Found a couple of issues after an initial review:
* NULL GtkIter passed in activate_pilot_menu
* The pilots combo labels grow each time you launch the 'restore' window from the applet.

Created attachment 142457 [details] [review]
fix two issues in patch 0001, conversion of applet to GtkBuilder

Patch  obsoletes the applet patch (#142304).
This patch still needs work (I don't think the devices combo is 
populating, for example).

Committed applet patch with another minor modification (set active entry in devices combo).
commit: ce7c75c3f231cc9c9fcfd491536c6f7b76511555

Just a note: I've made progress with the capplet patch.  Hopefully I'll complete it and check it in this week, but I just thought I'd make a note here in case Robert or others were planning to work on this.

Created attachment 143324 [details] [review]
Update to previous patch

Here's my current patch in progress, in case anyone wants to review.  It has a working GtkAssistant now, but the capplet is not done yet.

It all works as much as I can test without a device to test.  The only problem I found was if you cancel the assistant it tries to access invalid widgets. I think you need to change in gpilotd-control-applet.c:

            if (gpcap != NULL)
                gnome_pilot_capplet_update (GNOME_PILOT_CAPPLET (gpcap));

to:

            if (assistant_finished)
                gnome_pilot_capplet_update (GNOME_PILOT_CAPPLET (gpcap));

As a side note, I don't know how complex the synchronisation protocol is but a test program that could simulate a device would make this a lot easier to work on.

Hi Robert, thanks for taking a look.  Without a test device, I can imagine it was a bit frustrating.  I'm not aware of simulated devices -- there used to be a software simulator (Palm OS Emulator -- aka 'poser'), for which you needed to retrieve the ROM from an actual PalmOS device.  That was pretty cool, but I think it withered.  I'm not aware of any sync client simulator for any of the palm tools (pilot-link, kpilot, jpilot, or gnome-pilot), unfortunately.

The errors you see when canceling are because I haven't yet ported the "capplet" part of the code, just the druid-->assistant.  I'm not able to get much time for this at the moment, but am chipping away at it.

I've committed this stuff now, in a commit that also removes most of the libgnomeui dependence.  Testing welcomed.