GNOME Bugzilla – Bug 706756
GtkAssistant behaviour changed and now throws 'Page flow is broken'
Last modified: 2013-08-31 02:29:14 UTC
Created attachment 253062 [details] Video exposing the problem When doing the color calibration process, You can't go further the step when you're giving a name to the color profile. There is an error in the console. (gnome-control-center:22014): Gtk-CRITICAL **: Page flow is broken. You may want to end it with a page of type GTK_ASSISTANT_PAGE_CONFIRM or GTK_ASSISTANT_PAGE_SUMMARY
Hmm, I can reproduce here with jhbuilt versions of everything. I think this is a GTK regression, as with 3.8.x it works fine with no warnings. If you use colord from git master, you can test this without sensor hardware. Just add --create-dummy-sensor to the colord commandline. If you kill colord to restart it, be sure to restart g-s-d to get the correct devices added. As a random possibly related note, the sidebar is no longer populated in the assistant with newer versions of GTK and the first page of the assistant is no longer marked as initially active even though it's set in the ui file. You can see both buglets in the video posted by Baptiste.
I'll reassign this to GTK+ -- but if the color panel is doing something funky I guess I'll have to use the custom page type and do all the button handling myself again, although this isn't exactly fun.
Can't reproduce this. colord --create-dummy-device doesn't work here. Other assistant examples work fine, locally.
Looks like this commit broke the a11ytests: Is this an expected change? diff --git a/testsuite/a11y/assistant.txt b/testsuite/a11y/assistant.txt index ef78ff4..53bb37b 100644 --- a/testsuite/a11y/assistant.txt +++ b/testsuite/a11y/assistant.txt @@ -9,7 +9,7 @@ window1 button1 "push button" index: 0 - name: Button 1 + name: Page 1 state: enabled focusable sensitive showing visible toolkit: gtk <AtkComponent> @@ -24,7 +24,7 @@ window1 button2 "push button" index: 0 - name: Button 2 + name: Page 2 state: enabled focusable sensitive visible toolkit: gtk <AtkComponent> Is that an expected change?
no, I'll have to investigate how this is happening.
seems to be the correct output, after all