After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 653293 - AtkText: document new signals
AtkText: document new signals
Status: RESOLVED FIXED
Product: atk
Classification: Platform
Component: atk
unspecified
Other Linux
: Normal normal
: ---
Assigned To: ATK maintainer(s)
ATK maintainer(s)
: 653292 (view as bug list)
Depends on:
Blocks: 638537
 
 
Reported: 2011-06-23 23:14 UTC by Matthias Clasen
Modified: 2014-02-04 14:50 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
patch (2.83 KB, patch)
2011-06-23 23:14 UTC, Matthias Clasen
none Details | Review
adding support to details on text-insert and text-remove. Document it. (1.90 KB, patch)
2014-02-03 12:46 UTC, Alejandro Piñeiro Iglesias (IRC: infapi00)
none Details | Review

Description Matthias Clasen 2011-06-23 23:14:48 UTC
Created attachment 190546 [details] [review]
patch

...and the old ones too.
Comment 1 Joanmarie Diggs (IRC: joanie) 2011-06-23 23:18:20 UTC
*** Bug 653292 has been marked as a duplicate of this bug. ***
Comment 2 Matthias Clasen 2011-06-23 23:19:17 UTC
One thing I forgot to add: the docs should also state that 'empty' emissions with length == 0 should be suppressed.
Comment 3 Matthias Clasen 2011-06-23 23:21:57 UTC
Joanie points out that the @text parameter for text_insert/text_remove needs to be renamed to @object or so, to avoid a clash with the other @text.
Comment 4 Matthias Clasen 2011-06-23 23:30:50 UTC
I'll also point out that the new signals don't need to be G_SIGNAL_DETAILED, since they don't take any detail strings.
Comment 5 Alejandro Piñeiro Iglesias (IRC: infapi00) 2013-12-11 13:06:15 UTC
Ups, I missed this when I revamped ATK documentation with bug 684665.

(In reply to comment #4)
> I'll also point out that the new signals don't need to be G_SIGNAL_DETAILED,
> since they don't take any detail strings.

In any case I changed this with a recent commit.

Closing bug. Thanks for the patch, and sorry for missing it.
Comment 6 Colin Walters 2013-12-11 15:13:39 UTC
This change broke the gtk+/a11y tests:

gtk+/a11ychildren.test: (/usr/libexec/gtk+/installed-tests/children:7687): GLib-GObject-WARNING **: ../../gobject/gsignal.c:3110: signal id '286' does not support detail (363)
Comment 7 Alejandro Piñeiro Iglesias (IRC: infapi00) 2013-12-11 15:30:31 UTC
(In reply to comment #6)
> This change broke the gtk+/a11y tests:
> 
> gtk+/a11ychildren.test: (/usr/libexec/gtk+/installed-tests/children:7687):
> GLib-GObject-WARNING **: ../../gobject/gsignal.c:3110: signal id '286' does not
> support detail (363)

My bad, I removed G_SIGNAL_DETAILED on the new signals as pointed by Matthias Clasen, but I wrongly removed it too on the old one.

Fixed with this commit:
https://git.gnome.org/browse/atk/commit/?id=693ff82e87471a2268fde236a0c97befe7d970c6

Thanks for pointing it.
Comment 8 Giovanni Campagna 2014-02-03 09:58:29 UTC
Removing G_SIGNAL_DETAILED from text-insert also broke Firefox, because mozilla's a11y implementation emits the signal with detail.
I agree this should be fixed in firefox, but not filling up everyone's log with GObject warnings is also a good idea...
Comment 9 Alejandro Piñeiro Iglesias (IRC: infapi00) 2014-02-03 11:26:58 UTC
(In reply to comment #8)
> Removing G_SIGNAL_DETAILED from text-insert also broke Firefox, because
> mozilla's a11y implementation emits the signal with detail.

After a quick research.

Firefox is adding the "::system" suffix here. So this bug is related to bug 639466 (initially assigned to Li Yuan, but that is not working on atk anymore). One of the problems on that bug was that was not possible to add two levels of details through at-spi. Now that with the new signals "insert" and "delete" detail is not needed, now it would be possible.

So I will reopen this bug, and discuss a little with Joanmarie and Mike Gorse. Anyway right now I'm biased to the following solution:
  * Add again the flag G_SIGNAL_DETAILED for those events.
  * Document that one possible detail for those events is "::system"

Note that document that "::system" on this signals will not automatically solve bug 639466, as that bug was wider, not only about this two signals (but will be a good starting point).

> I agree this should be fixed in firefox, but not filling up everyone's log with
> GObject warnings is also a good idea...

Taking into account my previous comment, it is likely that we will solve this bug on ATK. Thanks for reporting this issue.
Comment 10 Alejandro Piñeiro Iglesias (IRC: infapi00) 2014-02-03 12:46:24 UTC
Created attachment 267944 [details] [review]
adding support to details on text-insert and text-remove. Document it.

After an IRC chat, implementation of what I suggested on comment 9. Posted in order to review documentation wording.
Comment 11 Alejandro Piñeiro Iglesias (IRC: infapi00) 2014-02-03 13:00:25 UTC
(In reply to comment #10)
> Created an attachment (id=267944) [details] [review]
> adding support to details on text-insert and text-remove. Document it.
> 
> After an IRC chat, implementation of what I suggested on comment 9. Posted in
> order to review documentation wording.

Patch committted with some documentation nits (thanks Joanmarie for the review). Closing bug.
Comment 12 Alejandro Piñeiro Iglesias (IRC: infapi00) 2014-02-04 14:50:53 UTC
(In reply to comment #8)
> Removing G_SIGNAL_DETAILED from text-insert also broke Firefox, because
> mozilla's a11y implementation emits the signal with detail.
> I agree this should be fixed in firefox, but not filling up everyone's log with
> GObject warnings is also a good idea...

FWIW, the fix for that problem was included on atk 2.11.6 release, just released today. Apologies for any inconvenience.