No problem! I went ahead and linked the fixed ticket to the update here.

The new gtk3-3.24.25-2.fc33 build also backports some more scrollbar sizing tweaks from upstream if anyone wants to try it out.

@qoheniac, Thanks -- I backported fixes for this to gtk3-3.24.25-2.fc33, , can you see if it helps?

People, this is not a voting contest who likes small scrollbars and who likes larger ones. The update has already gone stable and additional karma doesn't affect it at all any more.

If you want to argue for a different scrollbar size I'd suggest engaging with upstream at

Thanks everybody who tested this update!

@decathorpe, That's a deliberate change.

Thanks -- I went ahead and added 1927052 to the list of fixed bugs now.

Thanks for testing! If you haven't done so already, can you file the account setup bug in upstream GNOME gitlab issue tracker? I don't think we have anyone looking at geary code issues on Fedora side.

@martinpitt, This version actually fixes the regression you pointed out. You are testing 85.0-5, but this is 85.0-8

@clnetbox, Hm, the "Select All" behaviour change appears to be a deliberate:

Do you want to try to bring it up in upstream issue tracker?

@jridky Hm, I have a hard time believing that gtk2 can break GNOME session like this because GNOME doesn't use gtk2 at all any more. Are you sure the issue you are seeing is caused by this update?

OK, sounds like a good plan :) Thanks!

@atim Is the crash above reproducible? Can you file a ticket upstream at to get this diagnosed, please?

I'll disable autopush on this update in the mean time.

This update has been unpushed.


gnome-software still seems to work fine, screenshots load etc.

OK, sounds like it's working then :) Clicking on the title bar is something else -- it's not "full screen", it's "maximized window" and esc isn't supposed to exit that. Thanks for all the feedback and testing!

It seems to work just fine in my testing: click "f" to open the full screen view, and then "esc" to close, which then goes back to where it was before.

@jpbn Is that a regression in this version?