exited on abort in editing_host_page_replace_photo()

[Sebastien Bacher]

Binary package hint: shotwell

Shotwell 0.9 in natty just crashed while I was playing with it under valgrind, switching between ad event view and a photo and back and toggle video displaying in the events grid

[Sebastien Bacher (seb128)]

valgrind error:

"==26119== at 0x63CAE71: raise (raise.c:64)
==26119== by 0x63CE34D: abort (abort.c:92)
==26119== by 0x61D239F: g_assertion_message (gtestutils.c:1358)
==26119== by 0x61D297C: g_assertion_message_expr (gtestutils.c:1369)
==26119== by 0x8162243: editing_host_page_replace_photo (PhotoPage.c:3929)
==26119== by 0x816CC44: library_photo_page_display_for_collection (PhotoPage.c:9355)
==26119== by 0x8207AA9: library_window_switch_to_photo_page (LibraryWindow.c:5117)
==26119== by 0x8142BD7: collection_page_real_on_item_activated (CollectionPage.c:2667)
==26119== by 0x8178B44: checkerboard_page_real_on_left_released (Page.c:5398)
==26119== by 0x5A11963: _gtk_marshal_BOOLEAN__BOXED (gtkmarshalers.c:86)
==26119== by 0x611C371: g_closure_invoke (gclosure.c:767)
==26119== by 0x612F047: signal_emit_unlocked_R (gsignal.c:3252)
==26119== by 0x61378D6: g_signal_emit_valist (gsignal.c:2993)
==26119== by 0x6137CC1: g_signal_emit (gsignal.c:3040)
==26119== by 0x5B46555: gtk_widget_event_internal (gtkwidget.c:4988)
==26119== by 0x5A0FAEC: gtk_propagate_event (gtkmain.c:2493)
==26119== by 0x5A0FE4E: gtk_main_do_event (gtkmain.c:1683)
==26119== by 0x5D5CA09: gdk_event_dispatch (gdkevents-x11.c:2377)
==26119== by 0x61AAAA7: g_main_context_dispatch (gmain.c:2440)
==26119== by 0x61AB26F: g_main_context_iterate.clone.5 (gmain.c:3091)
==26119== by 0x61AB92A: g_main_loop_run (gmain.c:3299)
==26119== by 0x5A0EB98: gtk_main (gtkmain.c:1255)
==26119== by 0x8297003: application_start (Application.c:141)
==26119== by 0x813931E: library_exec (main.c:1092)
==26119== by 0x813A6FE: _vala_main (main.c:1553)
==26119== by 0x813A7FB: main (main.c:1569)"

[Sebastien Bacher (seb128)]

the assert which failed is view != NULL

[Adam Dingle (adam-yorba)]

Ticketed upstream at http://trac.yorba.org/ticket/3438 .

[Sebastien Bacher (seb128)]

Thanks, setting to low for now since that could be a race and valgrind really slow down things, it might not happen easily in a normal run

[Sebastien Bacher (seb128)]

So I noticed you closed the upstream bug but valgrind showed a problem there, wouldn't it make sense to keep the bug open until someone figure how that can happen rather than closing it because it's not trivial to trigger?

[Adam Dingle (adam-yorba)]

Sebastien,

At Yorba, if try to reproduce a bug and are unsuccessful, we generally close it, especially if only one user has reported it. If we kept such bugs open, our bug database would grow larger and larger with lots of bugs which we can do nothing about.

I do think, however, that we should start running Shotwell in valgrind as part of our testing before every release. We haven't done that much so far and I think it might reveal bugs. If we do that, perhaps we'll find a way to see this bug again.

[Sebastien Bacher (seb128)]

Ok, I understand the rational it's clearly a challenge to deal with bug reports which happen sometimes to some users, we tend toclean old bugs that didn't get any recent activity but not bugs that have a debugger log showing issue in the current version since usually those reflect a real bug. Let's see if other duplicates about that issue are coming