Bug 128172 - [Tests] some tests in fast/events/touch test non-replicable platform-specific behavior
Summary: [Tests] some tests in fast/events/touch test non-replicable platform-specific...
Status: NEW
Alias: None
Product: WebKit
Classification: Unclassified
Component: Tools / Tests (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Nobody
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2014-02-04 03:44 PST by Carlos Garnacho
Modified: 2018-11-29 07:33 PST (History)
3 users (show)

See Also:


Attachments
gdb backtrace for crash on test fast/events/touch/touch-target-limited.html on wk2 gtk (r165814) (8.90 KB, text/plain)
2014-03-18 11:17 PDT, Carlos Alberto Lopez Perez
no flags Details
Patch (4.60 KB, text/plain)
2014-03-18 13:40 PDT, Carlos Alberto Lopez Perez
clopez: review-
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Carlos Garnacho 2014-02-04 03:44:29 PST
In the workings of a patch for bug #98931 to add touch support for GTK+, multiple tests had to be skipped because they tested platform-specific behavior in several ways:
- The EventSenderProxy object is expected to coalesce multiple operations under one single event. Under X11/GTK+, each touch update will emit an independent event that can't be coalesced, as there is nothing that groups touch IDs at that layer, so that brings 1) a different number of events than many tests expect and 2) different expectations if the effect of two touch updates are expected to happen at once.
- On some other tests, touch IDs were directly tested, even though platforms may generate those differently, as it happens with X11

So in order to be useful more generally I'd propose to change those tests to test the coherence of the events gotten, rather than a fixed stream of those. The events currently tested would be rather considered "milestones", and intermediate events that still maintain coherence would be safely skipped, that way those tests would also be useful to test GTK+ behavior. As for touch IDs, comparing those from previous events is fine, checking fixed numbers is not...
Comment 1 Carlos Alberto Lopez Perez 2014-03-18 11:16:03 PDT
The test fast/events/touch/touch-target-limited.html crashes on wk2-gtk.

Program terminated with signal 11, Segmentation fault.
#0  gdk_event_free (event=0x4f73bd2b) at gdkevents.c:654
654	  if (event->any.window)


Thread 1 (Thread 0x7f318f79ca00 (LWP 17147)):
#0  gdk_event_free (event=0x4f73bd2b) at gdkevents.c:654
#1  0x00007f319dcfb042 in WebCore::GtkTouchContextHelper::handleEvent(_GdkEvent*) () from /stuff/webkit/webkit/WebKitBuild/Release/.libs/libwebkit2gtk-3.0.so.25
#2  0x00007f319c7eaece in webkitWebViewBaseTouchEvent(_GtkWidget*, _GdkEventTouch*) () from /stuff/webkit/webkit/WebKitBuild/Release/.libs/libwebkit2gtk-3.0.so.25
[....]


I'm attaching the full crashlog
Comment 2 Carlos Alberto Lopez Perez 2014-03-18 11:17:55 PDT
Created attachment 227079 [details]
gdb backtrace for crash on test fast/events/touch/touch-target-limited.html on wk2 gtk (r165814)
Comment 3 Carlos Alberto Lopez Perez 2014-03-18 13:40:12 PDT
Created attachment 227101 [details]
Patch
Comment 4 Carlos Alberto Lopez Perez 2014-03-18 13:54:12 PDT
Comment on attachment 227101 [details]
Patch

Ups. I messed up with the webkit-patch upload script. This is a gardening patch. I just created a new bug for it #130421

Sorry for the noise.
Comment 5 Alicia Boya García 2017-12-20 09:32:28 PST
fast/events/touch/touch-target.html has started crashing too on GTK since r225977-225987.
Comment 6 Miguel Gomez 2018-11-29 07:33:12 PST
fast/events/touch/basic-single-touch-events.html

besides fail and crash it can also timeout now.