[Webkit-unassigned] [Bug 72830] [Gtk] Regression: text-inserted events lack text inserted and current line

Fri Nov 25 08:18:11 PST 2011

https://bugs.webkit.org/show_bug.cgi?id=72830

Mario Sanchez Prada <msanchez at igalia.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #116624|                            |review?
               Flag|                            |

--- Comment #1 from Mario Sanchez Prada <msanchez at igalia.com>  2011-11-25 08:18:11 PST ---
Created an attachment (id=116624)
 --> (https://bugs.webkit.org/attachment.cgi?id=116624&action=review)
Patch proposal + unit test

Attaching now a patch proposal that basically updates the code to using the new 'text-insert' and 'text-remove' signals (instead of the old, and now deprecated, 'text-changed' one), which allows us to emit the text being actually modified (inserted or removed) along with the signal, instead of just emitting the 'offset' and 'count' values and then relaying on the ATK bridge to get the right information at the right time, which is something we know does not always work fine (and one of the main reasons behind the deprecation of the 'text-changed' signal, AFAIK).

This patch also changes the signature of the nodeTextChangeNotification() and  nodeTextChangePlatformNotification() in AXObjectCache, to allow specifying the text  being modified from the text editing commands (as we're gonna send the modified text, there's no point in not providing this information to the AXObjectCache if we already know it, right?), and so I needed to update also the signatures in the platform-specific versions of that class for the mac, win and chromium ports too, so they will keep building. Fortunately, this change shouldn't impact those ports at all since they are basically ignoring this functionality and providing an empty implementation for them.

Last, I haven't included a new test with this patch, but just updated the a11y API test we already had in place to use the new signals instead of the old ones. Also, I made the most of this change to implement more careful and detailed checks in that unit test, since it's obvious that the old version of it was not good enough, as it didn't detected that things broke heavily recently (and thus the reason for this bug to be reported).

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.