20:05:05 <DavidWHodgins> #startmeeting
20:05:05 <Inigo_Montoya> Meeting started Thu Mar  4 20:05:05 2021 UTC.  The chair is DavidWHodgins. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:05:05 <Inigo_Montoya> Useful Commands: #action #agreed #help #info #idea #link #topic.
20:05:22 <DavidWHodgins> #chair MageiaTJ wilcal
20:05:22 <Inigo_Montoya> Current chairs: DavidWHodgins MageiaTJ wilcal
20:05:29 <DavidWHodgins> #topic * Who's new? - If you are then come and say hello
20:05:52 <DavidWHodgins> Anyone here who hasn't been to a qa team irc meeting before?
20:06:14 <DavidWHodgins> * Testing Mageia 8 - Release & Updates
20:06:32 <DavidWHodgins> I've been helping with https://bugs.mageia.org/show_bug.cgi?id=28485
20:06:34 <[mbot> [ 28485 – mythtv: upgrade MGA7->MGA8 fails on file conflicts ]
20:06:52 <wilcal> I'm do'n fine. M8 is very useful and starting to move my other platforms to it
20:07:06 <DavidWHodgins> Will test the mythtv fix shortly
20:07:16 <MageiaTJ> Oh, good. Every time I try to do something with mythtv I get a headache.
20:07:27 <wilcal> mythtv has been very problematic over the years
20:07:28 <DavidWHodgins> :-)
20:07:38 <auroud_85[m]> :-)
20:07:59 <wilcal> I find the Raspberry Pi with Kodi to be much more useful
20:07:59 <auroud_85[m]> Many bugs fixed today thanks all of your testing.
20:08:03 <DavidWHodgins> All I'm testing is that it upgrades from m7 to m8 cleanly. No tv cards here, so can't test that it actually works.
20:08:41 <wilcal> BTW that technology is about to be completely upended
20:08:50 <MageiaTJ> I have one of those air2pc cards, but could never get it to work well.
20:08:52 <DavidWHodgins> wilcal: Are you running Mageia on a rpi4?
20:08:59 <neoclust> don't hesitate to ping me if needed, i am around
20:09:19 <wilcal> no but there is a project to do that
20:09:23 <DavidWHodgins> I have it running, but no sound over the hdmi cable
20:09:43 <DavidWHodgins> Everything else works
20:10:25 <wilcal> As the new ATSC 3.0 rolls out there's going to be LAN based tuner webservers that you can watch TV with a browser.
20:10:41 <DavidWHodgins> Is anyone aware of any other bugs about upgrading from m7 to m8 other then bug 28485 that involves file conflicts?
20:10:43 <[mbot> Bug: ['mythtv: upgrade MGA7->MGA8 fails on file conflicts', 'NEW', 'David GEIGER'] https://bugs.mageia.org/show_bug.cgi?id=28485
20:11:14 <auroud_85[m]> <DavidWHodgins "Is anyone aware of any other bug"> llvm but with devel package
20:11:17 <wilcal> The most popular one serves 4 seperate and indepandent video streams to watch. All at the same time
20:11:37 <auroud_85[m]> Not on my computer, don't remember Bug #
20:11:55 <MageiaTJ> Just the boomaga one, don't remember the number.
20:12:11 <auroud_85[m]> https://bugs.mageia.org/show_bug.cgi?id=28007
20:12:14 <[mbot> [ 28007 – llvm-11.0 mga8 packages conflicts with mga7 llvm-devel packages when updating MGA7 to MGA8 ]
20:12:14 <wilcal> Ask me again about that next week as I roll M8 in it's entirety into my boxes
20:12:18 <auroud_85[m]> Found it
20:12:25 <MageiaTJ> The script doesn't remove the mga7 version, so both wind up installed.
20:12:44 <wilcal> The one laptop I have used for months seems to be just fine with all the aps I use
20:13:01 <DavidWHodgins> I'm thinking about the timing of enabling updgrading using mgaapplet without having to specifiy --testing
20:13:17 <neoclust> DavidWHodgins: btw i will buy  Rpi4 to help testing :)
20:13:44 <MageiaTJ> bug 28430
20:13:46 <[mbot> Bug: ['When upgrading to mga8 from mga7, the boomaga update fails to remove the mga7 package', 'NEW', 'Mageia tools maintainers'] https://bugs.mageia.org/show_bug.cgi?id=28430
20:13:47 <auroud_85[m]> Tj boomaga is https://bugs.mageia.org/show_bug.cgi?id=28430
20:13:49 <[mbot> [ 28430 – When upgrading to mga8 from mga7, the boomaga update fails to remove the mga7 package ]
20:14:03 <wilcal> I use a Pi 400 for lots of support of my M8 testing
20:14:20 <auroud_85[m]> <DavidWHodgins "I'm thinking about the timing of"> Yes you're right
20:14:26 <auroud_85[m]> Time to come.
20:14:28 <joeghi> wilcal: Pi400 works?
20:15:26 <wilcal> Extremely well. It's the fastest Pi out
20:15:34 <wilcal> But not yet with Mageia ARM
20:16:07 <wilcal> I suspect it would work with IceWM or maybe even Xfce
20:17:27 <Guygoye_> with LxQt too
20:17:32 <wilcal> First big slug of M8 updates went through here this morning. No problems. Made me nervious
20:18:28 <wilcal> FWIW both my Pi 4's do an NFS mount from my Mageia 7.1 server
20:18:51 <DavidWHodgins> The longer we wait before enabling mgaapplet for everyone, the more updates there will be that were not present when we tested, so the sooner the better
20:19:02 <wilcal> Which makes them extremely useful
20:20:01 <auroud_85[m]> I do think we should enable it
20:20:02 <DavidWHodgins> For now, please keep working on upgrade tests
20:20:31 <DavidWHodgins> auroud_85[m]: We have to fix at least bug 28430 first
20:20:33 <[mbot> Bug: ['When upgrading to mga8 from mga7, the boomaga update fails to remove the mga7 package', 'NEW', 'Mageia tools maintainers'] https://bugs.mageia.org/show_bug.cgi?id=28430
20:20:45 <auroud_85[m]> There is no major Bug Report on desktops packages nor some server.
20:20:51 <DavidWHodgins> Oops. Meant bug 28485
20:20:53 <[mbot> Bug: ['mythtv: upgrade MGA7->MGA8 fails on file conflicts', 'NEW', 'David GEIGER'] https://bugs.mageia.org/show_bug.cgi?id=28485
20:21:00 <auroud_85[m]> The samba scriptlet not fixed
20:21:18 <auroud_85[m]> Yes.
20:22:36 <DavidWHodgins> File conflicts are really bad because it isn't just the bad package that doesn't get upgraded, it's all of the packages in the same transaction. At leaest that's my understanding
20:23:44 <DavidWHodgins> Depending on what else is in the same transaction it can have a cascade effect, and leave the system non-functional
20:23:56 <auroud_85[m]> Yes correct
20:24:56 <DavidWHodgins> In the case of the mythtv bug, it left rpm itself broken
20:25:24 <DavidWHodgins> Fixable, but not from within that system
20:25:34 <DavidWHodgins> Requires booting working media
20:27:22 <DavidWHodgins> So anyone who hasn't done upgrade tests yet, please do so, ensuring the mirror is up-to-date before starting, and make sure it's not your only system. Having a working live iso is strongly recommended
20:28:08 <wilcal> Should it work in a Vbox client?
20:28:17 <auroud_85[m]> Yep, we should at least communicate that mgaapplet will be activated as soon as https://bugs.mageia.org/show_bug.cgi?id=28393 is fixed
20:28:19 <[mbot> [ 28393 – Tracker: Upgrade Mageia 7 to Mageia 8 ]
20:28:37 <DavidWHodgins> wilcal: Yes. If you have enough space you can test it in a snapshot so it's easy to backou
20:28:53 <wilcal> Clone a Vbox M7.1 client then mgaapplet that
20:29:44 <DavidWHodgins> auroud_85[m]: I'll be splitting it into another report so mythtv can be pushed and another one opened for opencv
20:30:52 <auroud_85[m]> opencv should be opened yes
20:33:03 <DavidWHodgins> #topic m7/m8 updates
20:33:53 <DavidWHodgins> As usually happens, when cauldron reopens post release we get swamped with all of the updates that have accumulated since release freeze
20:34:15 <MageiaTJ> List is shorter than it was a couple of days ago.
20:34:16 <wilcal> cross fingers for tomorrow
20:34:55 <DavidWHodgins> Start at the top of the list for each release in https://madb.mageia.org/tools/updates and focus on the security updates first
20:35:00 <[mbot> [ Mageia App Db - Current Update candidates ]
20:35:31 <DavidWHodgins> For m7, they are all security. :-(
20:35:54 <wilcal> lots of dups between M7 and M8
20:36:18 <DavidWHodgins> Yes. Those must be tested on both releases before they can be validated
20:36:30 <wilcal> good ole roundcubemail is in there
20:37:39 <MageiaTJ> And Tomcat
20:38:12 <wilcal> When is EOL for M7.1?
20:38:54 <auroud_85[m]> <wilcal "When is EOL for M7.1?"> End of may 2021
20:38:56 <DavidWHodgins> 3 months from m8 release
20:39:22 <DavidWHodgins> So May 26th, 2021
20:39:34 <MageiaTJ> python-cryptography - didn't we just do that for M7?
20:40:49 <DavidWHodgins> It's probably a different issue. I haven't looked at it yet
20:41:22 <DavidWHodgins> Same test procedure though, so should be easy to retest
20:41:37 <auroud_85[m]> Note that https://bugs.mageia.org/show_bug.cgi?id=28392 glib2 is a security one for M8, and a M7 update will eventually did if we found patchs.
20:41:39 <[mbot> [ 28392 – glib2.0 new security issues fixed upstream in 2.66.6 ]
20:42:09 <auroud_85[m]> The update for M7 is Bug 28520
20:42:22 <auroud_85[m]> Not assigned to QA.
20:44:24 <DavidWHodgins> It looks like, Thanks to Thomas holding off the update, the impact is minor for us.
20:45:01 <auroud_85[m]> Yes.
20:46:21 <Benmc> sorry - customer: <auroud_85[m]> I do think we should enable it. absolutely
20:46:59 <DavidWHodgins> As quickly as we can, once all known file conflicts have been fixed, yes.
20:48:34 <DavidWHodgins> #topic 32bit i586 and SSE/SSE2 instructions slipping
20:48:48 <wilcal> Question about M8 Plasma desktop. In M7 the mouse scrool wheel you could move from desktop to desktop. In M8 I can't seem to get that to work. Bug or do I not understand something
20:49:10 <joeghi> for the i586 slipping there are a few packages to fix
20:49:34 <DavidWHodgins> Just fyi, that's about some i586 packages that were compiled with options that resulted in packages that will not work on i586 systems that don't have sse ability in the cpu
20:49:43 <auroud_85[m]> <wilcal "Question about M8 Plasma desktop"> Only on the plasmoid in Panel, not on desktop. Upstream change.
20:49:46 <DavidWHodgins> Mostly rust programs
20:49:51 <wilcal> Vbox on M8 is working very well for me for both M7 & M8
20:50:06 <joeghi> I collected the fixes here: https://copr.fedorainfracloud.org/coprs/ghibo/mga8-nosse/
20:50:06 <[mbot> [ ghibo/mga8-nosse Copr ]
20:50:15 <joeghi> not sure only by firefox
20:50:18 <wilcal> That's too bad. That was a nice feature
20:51:57 <auroud_85[m]> <DavidWHodgins "Just fyi, that's about some i586"> Sadly on x86, that will be more and more frequent.
20:52:03 <DavidWHodgins> joeghi: They have to be fixed in the m8 build system to be available to all users
20:52:23 <joselp> Sorry
20:52:25 <joeghi> DavidWHodgins: yep. I know, that were prerelease to check.
20:52:36 <DavidWHodgins> Ok
20:52:37 <joselp> I was at dinner
20:52:47 <joeghi> DavidWHodgins: there was a thread on the MLs about that last week.
20:52:59 <wilcal> Open restaurants. What a luxury
20:53:32 <Benmc> I will say nothing
20:53:59 <joeghi> DavidWHodgins: maybe there could be other, not in the list. We should open IMHO a single bug report where to report the packages spotted with sse slipping
20:54:29 <DavidWHodgins> So everyone, please do as much upgrade testing as you can, but don't forget regular updates testing too. If you can test a given update on both Releases at the same time, that really speeds things up.
20:54:34 <joselp> no, no dinner in my house
20:55:37 <DavidWHodgins> joselp: The developers and sysadmins need to modify the build system to check for that and reject such packages with a clear error message
20:56:38 <DavidWHodgins> If the packages don't pass the build system, they can't slip through. :-)
20:57:11 <joeghi> DavidWHodgins: what do you mean?
20:57:45 <joeghi> DavidWHodgins: there is no way to check whether the code is loaded later at runtime, e.g. dlopened
20:57:50 <DavidWHodgins> Add a step to the build system to scan the generated i586 bin packages for sse2 instructions
20:58:07 <joeghi> DavidWHodgins:  it could provide list of "suspects"
20:59:09 <joeghi> DavidWHodgins: but if you have code that it's executed at runtime only after it has verified the system has sse2 then code checking would match them, even if it's not strictly using it on all the platforms.
20:59:44 <DavidWHodgins> Code downloaded and run at runtime is not under Mageia control.
21:00:03 <joeghi> DavidWHodgins: I mean loaded at runtime, not downloaded.
21:00:50 <joeghi> DavidWHodgins: e.g. I've a function "if (has_sse2) then do this, else do that".
21:01:14 <joeghi> DavidWHodgins:  you would match the code for SSE2, but the system won't generate illegal instructions on i586.
21:01:37 <DavidWHodgins> I understand. Code that takes advantage of sse if it's available, but works without it if not available. I guess those must be manually checked
21:01:39 <joeghi> DavidWHodgins: at most it could be a check added to rpmlint as warning.
21:02:05 <joeghi> DavidWHodgins: the code checking would take both, but could be false positive.
21:02:17 <DavidWHodgins> And things like virtualbox which is built for i586 but not expected to work
21:02:45 <joeghi> DavidWHodgins: virtualbox doesn't limit the CPU instructions, so you can have AVX2 even on i586.
21:02:56 <DavidWHodgins> We could simply post a list of packages known not to work on i586, with a seperate list for i586 without sse
21:02:57 <joeghi> DavidWHodgins:  you need qemu specifying -cpu pentium or ahtlon
21:03:38 <DavidWHodgins> The i586 list already includes virtualbox, xen, and qemu. We do not test those on i586 at all.
21:03:43 <joeghi> DavidWHodgins: the ones I spotted were fixed. It remain firefox under some circumstances, which I'm not sure.
21:03:47 <DavidWHodgins> i586 host that is.
21:03:54 <joeghi> DavidWHodgins: you mean virtualbox itself to run on 32bit?
21:04:02 <joeghi> DavidWHodgins:  or to emulate 32bit?
21:04:23 <DavidWHodgins> We support 32 bit guests, not virtualbox on a 32 bit host
21:04:40 <joeghi> DavidWHodgins: I read in ML many people tests on real i586, like Athlon XP Barton.
21:04:57 <wilcal> Wow
21:05:15 <MageiaTJ> "Many" might be an exaggeration.
21:05:16 <DavidWHodgins> That must be painfull!
21:05:26 <joeghi> Felix IIRC.
21:06:10 <MageiaTJ> I only know of a few true i586 users in QA. I have a P4, but that's i686.
21:06:27 <DavidWHodgins> I looked last year and couldn't find any stores around here still selling i586 systems
21:06:34 <joeghi> DavidWHodgins: what I mean is that if you run virtualbox to emulate a 32bit guests it will have the SIMD instructions of the guest.
21:06:57 <MageiaTJ> You have to shop at church sales. ;-)
21:06:59 <joeghi> DavidWHodgins: I've a Pentium 133 in canteen
21:07:10 <wilcal> I set up a Vbox i586 Gnome client this morning
21:07:20 <wilcal> Works just fine
21:07:40 <joeghi> wilcal: which CPU is the Virtualbox host?
21:07:57 <wilcal> Intel i10 Dell laptop
21:08:10 <wilcal> 8 cores
21:08:11 <DavidWHodgins> joeghi: You're right. The build system can only warn about non-sse instructions being present, not reject the builds
21:08:33 <joeghi> wilcal: hwinfo --cpu will show you the simd instruction of your cpu.
21:08:47 <joeghi> wilcal: they'll be the same even under vbox/i586 guest.
21:09:02 <auroud_85[m]> <joeghi "wilcal: they'll be the same even"> You're right.
21:10:13 <DavidWHodgins> My oldest system has sse,sse2 in the cpu flags
21:10:50 <joeghi> I've a dual Coppermine PIII
21:10:55 <joeghi> in canteen too.
21:10:57 <DavidWHodgins> Finding non sse systems now is hard. :-)
21:11:32 <joeghi> qemu-system-i386  -cpu pentium would emulate a pentium (i586).
21:11:39 <joeghi> and generate traps.
21:11:52 <DavidWHodgins> We may have to drop support for non-sse cpu systems soon
21:12:05 <DavidWHodgins> That will be up to our kernel packagers
21:12:37 <Benmc> will have to put my athlon xp2400+ to bed
21:12:39 <joeghi> qemu-system-i386 -cpu help will show the list of cpu that can be emulated.
21:13:05 <joeghi> switching to i686+SSE2 would
21:13:15 <joeghi> have been fine 15 years ago, not today...
21:13:25 <joeghi> :-)
21:14:04 <joeghi> most problem of old systems is maximum quantity of ram.
21:14:36 <DavidWHodgins> Benmc:  does it show sse in  the cpu flags in lscpu?
21:14:59 <Benmc> yes
21:15:16 <DavidWHodgins> Then it's not affected. You can keep using it. :-)
21:15:23 <Benmc> but not sse2
21:15:52 <joeghi> DavidWHodgins: i686+SSE2 is basically i786.
21:15:58 <joeghi> aka pentium4.
21:16:06 <DavidWHodgins> My understanding is that this is about supporting a cpu that doesn't have even sse
21:16:30 <joeghi> just a step above p4 and you have 64bit x86_64
21:16:48 <joeghi> indeed there were even some p4 with the x86_64 set.
21:16:53 <auroud_85[m]> More than 20 years old CPU...
21:16:58 <MageiaTJ> There are 64-bit P4s, too. I had one.
21:17:29 <joeghi> auroud_85[m]: there were some CPU vendor that got licensed X86 and produce newer i586 with support for DDR etc.
21:17:36 <DavidWHodgins> We are talking about dropping support for really ancient hardware. The kind used on space missions. :-)
21:18:20 <auroud_85[m]> Yep. Such decision is for Council.
21:18:21 <joeghi> DavidWHodgins: you now the PowerPC 750 costs 200000$.
21:18:32 <MageiaTJ> Elon Musk must have a warehouse full.
21:18:32 <joeghi> DavidWHodgins: radiation hardened
21:18:41 <DavidWHodgins> Yes
21:19:06 <DavidWHodgins> Fewer things that can go wrong, since you know they will. :-)
21:19:17 <auroud_85[m]> But now, QA is warned, devs are sensibilized.
21:19:32 <auroud_85[m]> We should asks sysadmins also for BS.
21:19:41 <joeghi> DavidWHodgins: I wouldn't change anything in mga9 for i586. I'd let things as are
21:19:57 <joeghi> DavidWHodgins: unless dropping it and keeping just for 64bit compatibility.
21:20:06 <DavidWHodgins> Agreed. This is discussing the future
21:20:34 <DavidWHodgins> I'm against dropping 32 bit support completely, unless we have no choice
21:20:57 <wilcal> Does RedHat/Fedora support 32-bit?
21:21:23 <DavidWHodgins> Thanks to some 32 bit uefi controllers, some 32 bit support is now required for newer low end laptops
21:21:30 <joeghi> wilcal: FC switched to i686+SSE then dropped
21:21:57 <joeghi> DavidWHodgins: steam is 32bit...
21:22:42 <DavidWHodgins> I suspect the way forward will be to document somewhere packages that are not expected to work on non sse systems, but still include them in the repo
21:23:10 <joeghi> DavidWHodgins: actually the worst things for i586 are two: SSE+SSE2 slipping and lack of memory for linking.
21:23:15 <wilcal> Got to go here folkd see you all next week
21:23:26 <wilcal> Thanks to David for running the show
21:23:27 <joeghi> DavidWHodgins: and packager time
21:23:30 <DavidWHodgins> wilcal: Thanks
21:23:37 <wilcal> bye
21:23:41 <joeghi> bye
21:23:56 <auroud_85[m]> Bye wilcal
21:24:21 <DavidWHodgins> We just have to make sure non of the packages in base system minimal are in the list of non working packages
21:24:32 <ZekeMX> Hello everyone
21:24:40 <DavidWHodgins> Hi ZekeMX
21:24:50 <auroud_85[m]> <DavidWHodgins "We just have to make sure non of"> Agree
21:24:52 <MageiaTJ> Hi.
21:25:05 <DavidWHodgins> Just wrapping up this weeks qa meeting
21:25:11 <DavidWHodgins> Let's move on.
21:25:15 <DavidWHodgins> #topic * Anything else?
21:25:22 <auroud_85[m]> No.
21:25:44 <DavidWHodgins> Nothing else from me
21:26:05 <DavidWHodgins> Looks like countdown time then
21:26:09 <DavidWHodgins> t - 5
21:26:10 <Benmc> 3 x 7 to 8 magnitude earthquakes  north of NZ this morning
21:26:11 <ZekeMX> Does someone knows if the helpdesk system tbat mandriva had is still available?
21:26:14 <DavidWHodgins> 4
21:26:31 <DavidWHodgins> ZekeMX: No idea what they used
21:26:44 <DavidWHodgins> Anyone here know?
21:26:45 <ZekeMX> no.. ok- thanks
21:26:58 <DavidWHodgins> 3
21:27:01 <DavidWHodgins> 2
21:27:03 <DavidWHodgins> 1
21:27:08 <DavidWHodgins> Thanks for coming everyone
21:27:14 <DavidWHodgins> #endmeeting