Bug 655664 – Qt 5.9.6 stabilisation

• Status: RESOLVED FIXED
• Product: Gentoo Linux

[Description Andreas Sturmlechner 2018-05-13 16:17:35 UTC]

Any objections?

[Comment 1 Andreas Sturmlechner 2018-06-16 13:17:29 UTC]

ppc will need dev-qt/qtserialport stable for >=sci-astronomy/stellarium-0.16.0-r1.

We should probably raise the target to 5.9.6.

[Comment 2 Andreas Sturmlechner 2018-07-03 15:40:36 UTC]

Frank, do you feel like testing this? :)

[Comment 3 Andreas Sturmlechner 2018-07-04 18:52:37 UTC]

@pesa, I think this is in good shape already and we won't have to wait full 30 days for an LTS point release to stabilisation. Since 5.11.1 was unmasked this will probably not get that much additional testing from now on.

[Comment 4 Davide Pesavento (RETIRED) 2018-07-04 19:19:45 UTC]

What about bug 654744? should we restrict ffmpeg to =3* in 5.9.6?

Apart from that, I agree, let's proceed.

[Comment 5 Andreas Sturmlechner 2018-07-04 19:30:52 UTC]

(In reply to Davide Pesavento from comment #4)
> What about bug 654744? should we restrict ffmpeg to =3* in 5.9.6?
I doubt we are anywhere near unmasking of ffmpeg-4, but we could do that ofc.

> Apart from that, I agree, let's proceed.
I'd like to address bug 639140 as well, I think only qtsql should get a more fine-grained subslot.

[Comment 6 Frank Krömmelbein 2018-07-04 22:43:44 UTC]

(In reply to Andreas Sturmlechner from comment #2)
> Frank, do you feel like testing this? :)

The normal update ran smoothly on 2 different installations.
I use this version on my main system since this was available without problems.

But here are the first 2 points:

1) dev-qt/qtwebkit-5.9.1 has already stable keywords:
Keywords:    5.9.1:5/5.9: amd64 ppc64 x86 ~arm ~arm64 ~ppc
The only other available version is this pre release version:
Keywords:    5.212.0_pre20180120:5/5.212: ~amd64 ~arm ~arm64 ~ppc ~ppc64 ~x86


2) dev-qt/qt3d-5.9.4 i think that is a typo in the list and should be dev-qt/qt3d-5.9.6


I start tomorrow morning the tatt run, but i skip the "beast" qtwebengine for the moment >2h compile time per rebuild.
But the analysis I can do at the weekend at the earliest.

[Comment 7 Davide Pesavento (RETIRED) 2018-07-05 05:02:01 UTC]

(In reply to Frank Krömmelbein from comment #6)
> 1) dev-qt/qtwebkit-5.9.1 has already stable keywords:
Thanks, dropped from the list.

> 2) dev-qt/qt3d-5.9.4 i think that is a typo in the list and should be
> dev-qt/qt3d-5.9.6
Definitely a typo, but why are we stabilizing it in the first place? it has no stable keywords at the moment. Is there anything that needs it?

Also added dev-qt/qt-docs-5.9.6_p201806072055 to the list.

[Comment 8 Frank Krömmelbein 2018-07-05 06:43:21 UTC]

(In reply to Davide Pesavento from comment #7)

> > 2) dev-qt/qt3d-5.9.4 i think that is a typo in the list and should be
> > dev-qt/qt3d-5.9.6
> Definitely a typo, but why are we stabilizing it in the first place? it has
> no stable keywords at the moment. Is there anything that needs it?

It seems because of https://bugs.gentoo.org/652046#c0
> This will be required by sci-geosciences/qgis[3d] with stabilisation of 3.0.1.

But then this bug here should depend on:
media-libs/assimp-4.0.1: test failure: utVersion.aiGetVersionRevisionTest
https://bugs.gentoo.org/659122

[Comment 9 Andreas Sturmlechner 2018-07-05 06:47:40 UTC]

Ah, yes. We can drop qt3d from here and simply bump it in existing bug 652046, whenever deps are ready.

[Comment 10 Frank Krömmelbein 2018-07-09 13:07:16 UTC]

I'm sorry, it took a little longer than expected.
But i have good news, no real problems found.


A few restrictions to my previous statement:

- The only (to be expected) problems arise in cases when a useflag was disabled, which was forced enabled in IUSE for e.g. useflag libinput in package qtwayland.

- I still skipped qtwebengine, estimated almost 2 days emerge time for all useflag variations.

- Also skipped anything related with +systemd and +pulsaudio useflags enabled.

- One more point, the test here with tatt is a bit problematic because I use a desktop profile (desktop/plasma) with many activated useflags + my additional global useflags.
The result is that now all tests with for e.g.-PNG fail or will be skipped (REQUIRED_USE not satisfied (probably)).
I have here now for each package where such a problem occurred, exemplary ONE combination re-tested with  USE=-useflag emerge -1 --nodeps package
So I could not test all combinations in the end.

[Comment 11 Andreas Sturmlechner 2018-07-12 10:46:52 UTC]

Thanks for testing Frank! Arches, please stabilise.

[Comment 12 Mikle Kolyada (RETIRED) 2018-07-13 13:42:21 UTC]

amd64 stable

[Comment 13 Thomas Deutschmann (RETIRED) 2018-07-15 14:25:19 UTC]

x86 stable

[Comment 14 Mikle Kolyada (RETIRED) 2018-08-22 01:22:28 UTC]

ppc64 stable

[Comment 15 Andreas Sturmlechner 2018-08-28 17:37:58 UTC]

ping ppc

[Comment 16 ernsteiswuerfel 2018-09-04 22:20:53 UTC]

Created attachment 545946 [details]
tatt test results (ppc)

Looking good on ppc.

The failed builds of qtgui were caused by USE-flag blockers. I was able to resolve them manually, where the combinations build flawlessy. Blocked packages were due to libressl.

[Comment 17 Andreas Sturmlechner 2018-09-08 07:38:00 UTC]

It appears ppc was done, thanks!