843926 - Update ATK headers

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[alexander :surkov (:asurkov)]

Our version of ATK headers was added in 2006 and doesn't seems it was ever updated.

Current version is 2.7.91 per https://mail.gnome.org/archives/gnome-announce-list/2013-February/msg00019.html.

1) Can we update to latest version, i.e. are ATK headers backward compatible?
2) We have folder structure atk1-0/atk which assumes we can support more than one atk header set. Is it necessary/have usecase? Can we be fine with single atk header set?
3) Do we need special permission from Gecko stuff (license and all) to make an update?

Comment 1

[Trevor Saunders (:tbsaunde)]

(In reply to alexander :surkov from comment #0)
> Our version of ATK headers was added in 2006 and doesn't seems it was ever
> updated.

we changed it some in bug 480317

> Current version is 2.7.91 per
> https://mail.gnome.org/archives/gnome-announce-list/2013-February/msg00019.
> html.
> 
> 1) Can we update to latest version, i.e. are ATK headers backward compatible?

maybe, I'd just do atkrelationtype.h for now since its easier to check.

> 2) We have folder structure atk1-0/atk which assumes we can support more
> than one atk header set. Is it necessary/have usecase? Can we be fine with
> single atk header set?

I don't see a point, but it doesn't seem worth to change it.

> 3) Do we need special permission from Gecko stuff (license and all) to make
> an update?

I don't think so, I don't think we bothered for bug 480317.

Comment 2

[alexander :surkov (:asurkov)]

(In reply to Trevor Saunders (:tbsaunde) from comment #1)
> > 1) Can we update to latest version, i.e. are ATK headers backward compatible?
> 
> maybe, I'd just do atkrelationtype.h for now since its easier to check.

I wanted to just copy everything to stay updated

Comment 3

[alexander :surkov (:asurkov)]

(In reply to Trevor Saunders (:tbsaunde) from comment #1)

> > 2) We have folder structure atk1-0/atk which assumes we can support more
> > than one atk header set. Is it necessary/have usecase? Can we be fine with
> > single atk header set?
> 
> I don't see a point, but it doesn't seem worth to change it.

technically it's not atk1-0 anymore :)

Comment 4

[alexander :surkov (:asurkov)]

Attachment: patch

update to atk 2.10 (http://ftp.gnome.org/pub/GNOME/sources/atk/2.10/)

Comment 5

[alexander :surkov (:asurkov)]

Comment on attachment 812812 [details] [diff] [review]
patch

this one gives errors:

../../../other-licenses/atk-1.0/atk/atkutil.h:59:16: error: declaration for parameter 'AtkEventListener' but no such parameter
In file included from ../../../other-licenses/atk-1.0/atk/atkcomponent.h:28:0,
                 from ../../../other-licenses/atk-1.0/atk/atk.h:27,
                 from ../../dist/system_wrappers/atk/atk.h:3,
                 from ../../../widget/gtk/maiRedundantObjectFactory.c:8:
../../../other-licenses/atk-1.0/atk/atkutil.h:43:35: error: declaration for parameter 'AtkKeyEventStruct' but no such parameter
../../../other-licenses/atk-1.0/atk/atkutil.h:42:30: error: declaration for parameter 'AtkUtilClass' but no such parameter
../../../other-licenses/atk-1.0/atk/atkutil.h:41:25: error: declaration for parameter 'AtkUtil' but no such parameter
In file included from ../../../other-licenses/atk-1.0/atk/atk.h:26:0,
                 from ../../dist/system_wrappers/atk/atk.h:3,
                 from ../../../widget/gtk/maiRedundantObjectFactory.c:8:
../../../other-licenses/atk-1.0/atk/atkaction.h:71:7: error: declaration for parameter 'atk_action_get_type' but no such parameter
../../../other-licenses/atk-1.0/atk/atkaction.h:49:32: error: declaration for parameter 'AtkActionIface' but no such parameter
../../../other-licenses/atk-1.0/atk/atkaction.h:47:27: error: parameter 'AtkAction' has incomplete type
../../../other-licenses/atk-1.0/atk/atkaction.h:47:27: error: declaration for parameter 'AtkAction' but no such parameter
In file included from ../../../other-licenses/atk-1.0/atk/atk.h:25:0,
                 from ../../dist/system_wrappers/atk/atk.h:3,
                 from ../../../widget/gtk/maiRedundantObjectFactory.c:8:
../../../other-licenses/atk-1.0/atk/atkobject.h:640:14: error: declaration for parameter 'atk_object_get_object_locale' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:639:9: error: declaration for parameter 'atk_role_register' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:638:14: error: declaration for parameter 'atk_role_get_localized_name' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:635:10: error: declaration for parameter 'atk_object_remove_relationship' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:632:10: error: declaration for parameter 'atk_object_add_relationship' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:628:9: error: declaration for parameter 'atk_role_for_name' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:627:14: error: declaration for parameter 'atk_role_get_name' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:624:6: error: declaration for parameter 'atk_object_initialize' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:621:6: error: declaration for parameter 'atk_object_notify_state_change' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:618:6: error: declaration for parameter 'atk_object_remove_property_change_handler' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:616:7: error: declaration for parameter 'atk_object_connect_property_change_handler' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:612:6: error: declaration for parameter 'atk_object_set_role' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:610:6: error: declaration for parameter 'atk_object_set_parent' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:608:6: error: declaration for parameter 'atk_object_set_description' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:606:6: error: declaration for parameter 'atk_object_set_name' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:605:6: error: declaration for parameter 'atk_object_get_index_in_parent' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:604:14: error: declaration for parameter 'atk_object_ref_state_set' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:603:18: error: declaration for parameter 'atk_object_get_attributes' but no such parameter
../../../other-licenses/atk-1.0/atk/atkobject.h:599:10: error: declaration for parameter 'atk_object_get_layer' but no such parameter
../../../widget/gtk/maiRedundantObjectFactory.c:97:1: error: expected '{' at end of input
../../../widget/gtk/maiRedundantObjectFactory.c:97:1: error: control reaches end of non-void function [-Werror=return-type]
cc1: all warnings being treated as errors
make[5]: *** [maiRedundantObjectFactory.o] Error 1
make[5]: *** Waiting for unfinished jobs....

ideas?

Comment 6

[Alejandro Piñeiro]

Joanie just pointed me this bug. Some comments below:(In reply to alexander :surkov from comment #0)
> Our version of ATK headers was added in 2006 and doesn't seems it was ever
> updated.
> 
> Current version is 2.7.91 per
> https://mail.gnome.org/archives/gnome-announce-list/2013-February/msg00019.
> html.
> 
> 1) Can we update to latest version, i.e. are ATK headers backward compatible?

They should. Obviously since then some methods got deprecated, but we tried to be API and ABI compatible, event with the switch to 2.0 (that was made only to follow numbers with GNOME3). If you detect any backward incompatibility, is an error.

(In reply to alexander :surkov from comment #5)
> Comment on attachment 812812 [details] [diff] [review]
> patch
> 
> this one gives errors:

Those errors seems odd to me. They are complaining about symbols that have then there for ages.

> ../../../other-licenses/atk-1.0/atk/atkutil.h:59:16: error: declaration for
> parameter 'AtkEventListener' but no such parameter
> In file included from
> ../../../other-licenses/atk-1.0/atk/atkcomponent.h:28:0,
>                  from ../../../other-licenses/atk-1.0/atk/atk.h:27,
>                  from ../../dist/system_wrappers/atk/atk.h:3,
>                  from ../../../widget/gtk/maiRedundantObjectFactory.c:8:

On that line there are two atk.h files. One at /other-licenses/atk-1.0/atk/atk.h and another one at dist/system_wrappers/atk/atk.h. Is that normal? One wild guess is that there are two different versions of the same headers.
 
> ideas?

Not a lot, as I said those errors are extrange, as all the symbols that are not found are old (so already defined at 2006), like atk_object_set_role.

Comment 7

[alexander :surkov (:asurkov)]

(In reply to Alejandro Piñeiro from comment #6)
> Joanie just pointed me this bug.

Thanks!

> Those errors seems odd to me. They are complaining about symbols that have
> then there for ages.
> 
> > ../../../other-licenses/atk-1.0/atk/atkutil.h:59:16: error: declaration for
> > parameter 'AtkEventListener' but no such parameter
> > In file included from
> > ../../../other-licenses/atk-1.0/atk/atkcomponent.h:28:0,
> >                  from ../../../other-licenses/atk-1.0/atk/atk.h:27,
> >                  from ../../dist/system_wrappers/atk/atk.h:3,
> >                  from ../../../widget/gtk/maiRedundantObjectFactory.c:8:
> 
> On that line there are two atk.h files. One at
> /other-licenses/atk-1.0/atk/atk.h and another one at
> dist/system_wrappers/atk/atk.h. Is that normal? One wild guess is that there
> are two different versions of the same headers.

right, seems plausible.

Trev, can you check locally that? ideas what dist/system_wrappers/atk/atk.h is?

Comment 8

[alexander :surkov (:asurkov)]

Attachment: update atkrelationtype.h

let's update atkrelationtype.h only I need for bug 923289 until errors are resolved

Comment 9

[Trevor Saunders (:tbsaunde)]

Comment on attachment 813324 [details] [diff] [review]
update atkrelationtype.h

I'm not really sure why we need to keep these headers in the tree, but I don't see how doing this hurts anything.

Comment 10

[alexander :surkov (:asurkov)]

Comment on attachment 813324 [details] [diff] [review]
update atkrelationtype.h

landed http://hg.mozilla.org/integration/mozilla-inbound/rev/4e60867ca4ab

Comment 11

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/4e60867ca4ab

Comment 12

[BugBot [:suhaib / :marco/ :calixte]]

The bug assignee is inactive on Bugzilla, so the assignee is being reset.

Comment 13

[Suhaib Mujahid [:suhaib]]

Sorry, there was a problem with the detection of inactive users. I'm reverting the change.

Comment 14

[BugBot [:suhaib / :marco/ :calixte]]

Redirect a needinfo that is pending on an inactive user to the triage owner.
:Jamie, since the bug has recent activity, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 15

[James Teh [:Jamie]]

This bug did what it needed to do at the time. If we need to fully update ATK headers, we can deal with that in a new bug. See also https://phabricator.services.mozilla.com/D105216.