Created attachment 623452 [details] main menu in anaconda Description of problem: In F18 TC2 Beta, in KDE live, nothing is visible in the main menu. Main menu is functional, spokes are all right, but main menu is totally blank. See attached screenshot. How reproducible: Always. Steps to Reproduce: 1. Boot into F18 Beta TC2 KDE live 2. start anaconda Actual results: Nothing shows in the main menu. Expected results: Icons and text in main menu should be visible. Additional info: Proposing as beta blockerbug per alfa criterion: "The installer must be able to install each of the release blocking desktops, as well as the minimal package set, with each supported installation method"
Created attachment 624022 [details] Screenshot of the problem. I got a screenshot of this. It happens in the latest nightly (currently 10/8) too. Switching to the GTK Raleigh theme make the menu items visible again.
Hi, Installed anaconda-18.14-1.fc18.x86_64 and got the same invisible result. With the 'default' graphics nouveau driver and the 'basic graphics driver' (vesa?) Starting /usr/bin/liveinst at the command prompt gives the following messages: "localuser:root being added to access control list Starting installer, one moment... anaconda 18.14 for Fedora 18 (pre-release) started. (anaconda:2770): Gtk-WARNING **: Overriding tab label for notebook (anaconda:2770): Gtk-WARNING **: Unknown property: GtkGrid.n-rows (anaconda:2770): Gtk-WARNING **: Unknown property: GtkGrid.n-columns (anaconda:2770): Gtk-WARNING **: Unknown property: GtkGrid.n-columns (anaconda:2770): Gtk-WARNING **: Overriding tab label for notebook (anaconda:2770): Gtk-WARNING **: No object called: mountPointCompletion" Martin Kho
Those messages don't have anything to do with this, though that last one is definitely a bug I should fix. When did you first notice this happening?
Hi Chris, At least I've added a little :-) For me, it started happening in Beta TC2, never saw it in the alpha fase. I didn't try Beta TC1. Martin Kho
Hi, FYI: Just checked why I didn't try Beta TC1 live KDE. Well there were no Beta TC1 Lives :-) Martin Kho
... the last visible was Alpha TC6. Martin Kho
It was visible in Alpha TC5, I was unable to start anaconda in Alpha TC6 and it was invisible in Beta TC1.
Seems to be some sort of gtk3/oxygen-gtk3 bug or regression. gtk3-3.5.10-1.fc18 + oxygen-gtk3-1.1.0-2.fc18 as shipped with f18-Alpha, things work ok. gtk3-3.6.0-1.fc18 + oxygen-gtk3-1.1.1-1.fc18 (currently included in f18-beta-tc2) exhibits the behavior described here. I'll do more testing of gtk3 vs oxygen-gtk3 now in an attempt to bisect where things went bad.
sorry, reassigning back to anaconda mostly so this is easier to find and until I have more evidence where to better triage it.
OK, bisected the problem down to the upgrade from gtk3-3.5.12-2.fc18 (the last gtk3 version to have worked wrt to this bug) to gtk3-3.5.14-1.fc18 (the first gtk3 version where this bug is present). It may or may not be relevant that i noticed with < 3.5.14, that I could see anaconda window initially appear as a window prior to it going full screen; and with >= 3.5.14, it starts immediately full screen. triaging to gtk3 for comment.
I also see it in f18 Beta TC3 Live KDE x86_64 DVD
For the record, this wasn't broken in Alpha (some people seem unclear on that). We didn't epic fail there. Discussed at 2012-10-10 blocker review meeting: http://meetbot.fedoraproject.org/fedora-qa/2012-10-10/f18beta-blocker-review-3.2012-10-10-16.05.log.txt . Accepted as a blocker per criterion "The installer must boot (if appropriate) and run on all primary architectures, with all system firmware types that are common on those architectures, from default live image, DVD, and boot.iso install media when written to an optical disc and when written to a USB stick with at least one of the officially supported methods" - the criterion is a bit of a fudge as it's _running_, but really, this one's pretty clear and we don't need to get too tied up in procedure.
same in f18 Beta TC3 Live KDE x86_64 with desktop effects off (alt shift f12)
same in f18 Beta TC4 Live KDE x86_64
menus buttons positions are reversed in KDE anaconda -I know that this is KDE default but is confusing
It's an oxygen-gtk feature to adjust the button order in dialogs. Is it changing the order of buttons which aren't OK/Cancel or Yes/No buttons? Then Anaconda is abusing the API (it should use normal pushbuttons, not dialog buttons, for actions which aren't of the usual confirm and cancel type; dialog buttons all have a role, if the button action doesn't actually fit that role, it's used improperly) and it's a bug in Anaconda.
I spent some time debugging this today, and there are multiple bugs involved; the bulk of the bug is [1] for gtk3, but fixing that reveals some rendering glitches in oxygen-gtk3, for which I filed a patch at [2]. I think in the meantime though, the best choice is to disable the fancy hack oxygen-gtk3 uses to draw the inner shadows until these bugs are fixed - there's a build configuration provided for it. [1] https://bugzilla.gnome.org/show_bug.cgi?id=686265 [2] https://bugs.kde.org/show_bug.cgi?id=308533
Created attachment 628523 [details] disable inner shadows hack
oxygen-gtk3-1.1.1-2.fc18 has been submitted as an update for Fedora 18. https://admin.fedoraproject.org/updates/oxygen-gtk3-1.1.1-2.fc18
Discussed at 2012-10-17 blocker review meeting: http://meetbot.fedoraproject.org/fedora-qa/2012-10-17/f18beta-blocker-review-4.2012-10-17-16.00.log.txt . It should be possible to test the fix for this with TC5, or we could build a special KDE live for testing. Either way, we just need to test this, verify the fix, and up-karma the update.
I built a couple of live images to test this fix, one with 1.1.1-1, one with 1.1.1-2, no other differences. The 1.1.1-1 image shows the bug, the 1.1.1-2 image works, so this is verified as fixed.
oxygen-gtk3-1.1.1-2.fc18 has been pushed to the Fedora 18 stable repository. If problems still persist, please make note of it in this bug report.
Confirmed, fixed in TC6.
It's "fixed" by a cheap workaround. :-( I'd really like to see the actual issue fixed sooner rather than later…
What actual issue do you think is not fixed ?
(In reply to comment #24) > It's "fixed" by a cheap workaround. :-( I'd really like to see the actual > issue fixed sooner rather than later… If you read comment #17 you can find out I wrote patches for the actual issue(s), and they have been all committed in the relevant upstreams.
I know, and thanks for that! But I don't think those fixes are in Fedora yet. The oxygen-gtk3 update which got pushed only disables inner shadows.