[01:41] <micahg> please don't copy libreoffice to the release pocket, see bug #1062757
[01:41] <ubot2> Launchpad bug 1062757 in libreoffice "No menu bar under GNOME Shell & gnome-classic" [High,Confirmed] https://launchpad.net/bugs/1062757
[01:51] <infinity> micahg: Ugh, again?
[01:51] <micahg> yeah, it failed on arm* too, segfault
[01:53] <infinity> Then it ain't getting copied anyway.
[01:53] <micahg> right, I just noticed that afterwards though
[01:53] <infinity> But it sounds quite buggy in general for an rc->rc bump. :/
[01:54] <micahg> well, it says in the earlier changelog included 3.6.2.2, so I think the upstream versioning is wrong
[01:54] <micahg> infinity: did you see tyhicks's comment in the openssl bug?
[01:55] <infinity> Oh, no, haven't been back today.
[02:00] <infinity> micahg: Thanks for the nudge.
[02:00] <micahg> infinity: sure, when I saw the rejection e-mail I started panicking :)
[02:01] <infinity> micahg: I'm still pretty sure that that upload is wrong, but I'm willing to buy the "works like precise" argument while people investigate further.
[02:01] <infinity> It's at least conservatively consistent wrongness.
[02:02] <micahg> yeah
[15:01] <cjwatson> ^- test fixes only fixing FTBFS; webapps folks would probably appreciate a quick review
[15:02] <skaet> cjwatson,  yup,  made sense when I was looking at it,  have accepted.
[15:03] <skaet> cjwatson,   the grub2 one on the other hand....   who other than slangasek and yourself is best to review?
[15:13] <cjwatson> skaet: I asked slangasek to review it when he woke up; he did have a one-line contribution to it after some testing, but I don't think it's anything that should recuse him from review
[15:14] <cjwatson> aptdaemon> thanks
[15:14] <skaet> thanks cjwatson.   :)   ok,  saw him in there as an author, so wasn't sure about the level.   Thats fine then.
[15:16] <cjwatson> I tend to credit if in any doubt at all :)
[15:16] <skaet> :)
[15:17] <cjwatson> in this case he spotted a missing grub_errno cancellation
[15:18] <cjwatson> and of course the general design of what we're doing was following his high-level instructions
[15:18] <skaet> yes,  he's definitely the best choice for a second set of eyes then ;)
[15:19] <smartboyhw> :)
[15:22] <cjwatson> I think the remaining pieces following that upload are maintainer script code in grub-efi-amd64-signed to copy the loader image into the ESP, a shim-signed package once we get the MS signature, and a bit of fiddling in cdimage to use the signed loader
[15:23] <cjwatson> so the end is hopefully in sight, as it'd bloody well better be by this stage
[15:24] <cjwatson> oh, a corresponding grub-installer tweak too
[15:26] <cjwatson> ah, good, aptdaemon built
[15:34]  * skaet hoping the end is indeed in sight...  
[15:35]  * smartboyhw hope for it too
[15:35] <Laney> nah, the end is boring
[15:35] <smartboyhw> Laney, oh is it
[15:36] <skaet> Laney,  I like boring. boring is good.  ;)
[15:36] <Laney> :P
[18:46] <slangasek> grub2 LGTM
[18:47] <slangasek> (accepted)
[18:48] <Daviey> lets shake things up with a mass accidental sync from experimental.
[18:48] <Daviey> WEEEEEEEEEEEEEEEEEEEEEEEEee
[18:50] <slangasek> um
[18:54] <slangasek> cjwatson: so based on upstream's comments, I think we should be ok to drop the -partition_offset 16 option?
[18:57] <slangasek> cjwatson: or do you think we should pull 1.2.5?
[19:16] <phillw> cjwatson: do you have time for a PM?
[21:10] <cjwatson> slangasek: I'll read through the comments in detail tomorrow - haven't quite absorbed them yet
[21:11] <cjwatson> phillw: feel free to leave a message
[21:14] <infinity> slangasek / cjwatson: Based on my reading of the bug log, I suspect we'd be better off actually pulling the bugfix (either cherry-picking the commit he pointed to, or 1.2.5) than running with different options.
[21:18] <phillw> cjwatson: the fate of PPC is pretty much, once again in you guys at -release, -kernel -xorg. We are running out of time for PPC. Could be carrying over a 'hatchet job' for 12.10 be done, with the issues being looked after (with hopefully a little urgency) to 13.04?
[21:18] <infinity> slangasek / cjwatson: It is, after all, a 1-char fix.
[21:18] <infinity> phillw: What do you mean a "hatchet job"?
[21:19] <phillw> infinity: it would be a boot option, which I think debian are currently using?
[21:19] <infinity> phillw: Note that PPC *does* work fine on some hardware, it's some specific X drivers that seem broken.  "It doesn't work on all hardware" probably isn't an excuse to release, but absoltely, getting those drivers fixed on that hardware would be shiny.
[21:20] <infinity> phillw: Or, rather, some combinations of KMS and X drivers.
[21:20] <infinity> phillw: Do you have exact boot options that people assume should work?  I'll test on my non-Apple hardware here and see if they generically apply in any meaningful way. :P
[21:21] <phillw> infinity: I'm not a kernel person :) I did ask & get a patch from red-hat for kernel. but I am way out of my knowledge area here.
[21:21] <infinity> phillw: Where's this patch?
[21:21] <phillw> infinity: I can email you the full list from adam of all the bugs that are reported.
[21:22] <infinity> Uhm, sure.  Does one of them have the patch referenced in it?
[21:22] <infinity> Cause I do so enjoy treasure hunts.
[21:23] <phillw> as A3 hit, 12.10 went from testing on VM to breaking both nvid and geof vid chip sets. as we have only few testers with the actual kit, this did rather muck up testing...
[21:23] <phillw> infinity: email address to be sent to?
[21:24] <infinity> adconrad@ubuntu
[21:24] <phillw> infinity: email address not recognised?
[21:25] <infinity> (.com)
[21:25] <phillw> lol
[21:26] <slangasek> infinity: what do you see in the bug log that indicates partition_offset 16 is needed?  The image I built without that option tests out fine under both UEFI and BIOS in a VM
[21:26] <phillw> you should have incomming. that is all that has been discussed since i got the nVid bug fix into the kernel, which does not help PPC.
[21:29] <cjwatson> phillw: I'm not taking responsibility for it.  Get a kernel/X person to say it should be done and I'll be happy to do that.
[21:30] <cjwatson> The fact that all the mails about this are 3000 lines long with references to dozens of different long bugs is probably not helping!  People really need to learn the virtue of being concise.
[21:30] <infinity> slangasek: My reading and understanding of the feature leads me to believe that it ends up being part of the whole "conservative in what you produce, liberal in what you accept" mantra.  I suspect most systems should boot perfectly fine without it, but "making it look like a conventional HDD MBR layout" is almost certainly going to fix booting on some broken/unforgiving firmware.
[21:31] <cjwatson> (Frankly half the time I lose track of what's even being asked for.)
[21:31] <infinity> slangasek: And, indeed, upstream points to that sometimes being the case.
[21:31] <phillw> cjwatson: I know you cannot, and I'm also really grateful for telling PPC to "Go away"... but, PPC is now stuck between a rock & a hard place.
[21:31] <phillw> *for you NOT telling*
[21:31] <infinity> slangasek: Of course, he also points to partition_offset *causing* one failure he knows about WRT Mac booting, but I do wonder if that might have been a manifestation of the bug he just fixed. :P
[21:32] <phillw> ~ 2 weeks left...
[21:32] <cjwatson> No, it's stuck between nobody cornering a kernel/X person and GETTING A STRAIGHT ANSWER.
[21:32] <infinity> slangasek: Either way, the bugfix is 1-character, seems not unreasonable that we should cherry-pick and test if it fixes your issue.
[21:32] <cjwatson> It should take five minutes.  Please stop making it my problem!
[21:33] <cjwatson> I'm sorry but I can't deal any more with reading through thousands of lines of stuff about this.  It shouldn't be this hard.
[21:33] <phillw> cjwatson: well, as I don't have a private number to sabdfl, can you suggest someone who has the authority to bash their heads together?
[21:33] <cjwatson> No.  Don't escalate to sabdfl.
[21:33] <cjwatson> That's silly.
[21:33] <cjwatson> I've pointed you to IRC channels - did you try?
[21:34] <infinity> phillw: Have you tried finding the references to the EXACT bugs (not lists of them), perhaps patches people have tested even, and then taking those clear, concise references to the Kernel or X people?
[21:34] <cjwatson> Did anyone try?
[21:34] <cjwatson> Or did people just send thousand-line e-mails with all the bugs they'd ever heard of vaguely to do with powerpc?
[21:34] <cjwatson> I'm sorry but I'm very frustrated by the nonproductive approaches.
[21:35] <infinity> phillw: Either way, doomsday prophecies about the death of PPC aren't helpful either.  A broken video driver is a broken video driver.
[21:36] <cjwatson> You have to be concise, and ask people who can do something about it.
[21:36] <phillw> infinity: cjwatson there is one thing I have learned.... Adam has taken time to herd up the bugs. for L-QA I am pretty strict. but you must also know that a 'fail to actually get LiveCD', 'fail to install' made it hard.
[21:36] <cjwatson> Because all those people are busy - which doesn't mean that you shouldn't ask them, but it means that you should be respectful of their time and not make them read through dozens of bug reports.
[21:37] <infinity> phillw: You claimed there was a patch that fixes things.  You haven't yet actually referenced that patch.  It's this sort of thing we're talking about.
[21:37] <cjwatson> Adam's mails are too long for me to be able to absorb.
[21:37] <phillw> we lost B1 and B2 beacuse of what are actually video chip issues? Please do cut us some slack here?... please?
[21:37] <infinity> phillw: Could be a lot less wailing and gnashing of teeth and a bit more "here's the patch we found".
[21:37] <cjwatson> I've been trying to give gentle advice on this for weeks now ...
[21:38] <cjwatson> But all I've got is extraordinarily lengthy mails and requests to just do hacky things anyway
[21:38] <cjwatson> When I've said that the reason I'm not doing that is that I have no idea whether it would break something else
[21:38] <cjwatson> And I am not qualified to decide
[21:39] <cjwatson> If somebody who is qualified to decide looks at it and says sure, then I'm quite happy to do it
[21:39] <cjwatson> If I'm actually told in clear and concise terms what to do
[21:39] <phillw> infinity: well, let's try the other issue... working fine at A3, totally broken just after... had that been something like AMD64, it would have been screamed at as a regression? Pleae, for god's sake think I'm being agreesive... But, it did work and then got totally broken. I know PPC is low on the list, but... we did have it working?
[21:40] <cjwatson> You asked some time after if we could regress the entire distribution to alpha-3 just for lubuntu powerpc.
[21:40] <cjwatson> We would not have done that for amd64 either, because we can't.
[21:41] <cjwatson> It's not possible.
[21:41] <cjwatson> I would expect that if this had happened for amd64 then somebody would have taken the time to work out how to describe the problem in concise terms
[21:42] <cjwatson> And work out what was architecture-specific, what was bootloader-specific, and what was video-card-specific
[21:42] <phillw> infinity: as it got broken at a basic level, my suggestion to cjwatson was look at what we had then, that worked. I do not know enough of kernel stuff to understand if the ppc kernel <> AMD64 etc. It was just a reference point.
[21:42] <cjwatson> And wouldn't have been conflating them all into amd64 being broken, because that's just unhelpful
[21:43] <cjwatson> phillw: it doesn't work that way, for any architecture
[21:43] <cjwatson> I redirected you away from that approach as a way to try to save you time from an unproductive approach
[21:44] <cjwatson> There are millions of lines of code involved.  You have to investigate considerably more before you get down to the point where it's worth "looking at what worked back then", or reverting, or whatever.
[21:45] <cjwatson> Targeted reverts are a tool in the armoury, but you don't reach for "revert the whole kernel across a month of development" as the first step.
[21:45] <cjwatson> Yes, it is useful to know when something last worked, but there's much more to a sensible debugging process than that.
[21:46] <phillw> cjwatson: which has been done, but the pointer forward is a fix from kernel / x-org team... Have we had any back from Adam's suggestions? The only one that got in was the redHat one, that infinity exlained better to me to ask to be included. I'm sure it has solved the i686 bugs etc, but it did not fix PCC.
[21:46] <phillw> It solved some bugs, but not all of them. For that alone it was incorporating into 12.10
[21:46] <cjwatson> AFAIK the nvidia fix you previously pointed to was already included.
[21:46] <cjwatson> But obviously that's not relevant to radeon.
[21:47] <phillw> and it also does not fix the nVid issue in PPC
[21:47] <cjwatson> That's been buried in massively long mails.
[21:47] <cjwatson> They were so long I hadn't even noticed.
[21:48] <phillw> it is either a kernel fix (required) or a boot parameter.
[21:48] <cjwatson> Then get a kernel guy to sign off.
[21:48] <cjwatson> That's all I want.
[21:48] <phillw> cjwatson: I'm really, really sorry to be pecking your head. You are clearly a guy who wants PPC to happen.
[21:49] <phillw> cjwatson: is not infinity a kernel guy?
[21:50] <cjwatson> If infinity decides that the boot parameter's the right thing to do, then I trust his judgement; but he's a foundations developer on my team, not full-time kernel.
[21:50] <cjwatson> So I would respect his right to abstain.
[21:51] <phillw> cjwatson: so, if no one will make a decision, is PPC dead?
[21:51] <cjwatson> No, because that's been an exaggeration right from the start.
[21:52] <cjwatson> Lubuntu powerpc might have trouble with this release.
[21:52] <phillw> cjwatson: it can not pass the iso tests, so it will be dead.
[21:52] <cjwatson> But there's more to powerpc than Lubuntu (e.g. see Ben Collins' recent posts on planet), and there's more than this release.
[21:53] <cjwatson> Dead implies not getting back up.
[21:53] <phillw> cjwatson: so there is a ubuntu / kubuntu / xubuntu version?
[21:53] <cjwatson> Ben's been doing server work.
[21:55] <cjwatson> The architecture is clearly not just about video card bugs ...
[21:55] <phillw> cjwatson: I assure you, I'm much rather that L-PPC goes out with out with a release note than nothing at all. I'm also going to be tests for PPC-Server so we keep kate happy, that is not reliant on desktop video chips :)
[21:56] <cjwatson> Not always generalising bugs encountered on powerpc to all of powerpc would really help.
[21:56] <cjwatson> As I think infinity said above.
[21:58] <phillw> Jonathan, one of the Lubuntu devs also will test PPC server on 'real kit'. All I want, honest.... please... just an insatallable system for 12.10 PPC? The stuff dropped late, took some finding out.. but...... can we get something working for 12.10?
[21:59] <phillw> cjwatson: was Adam always as PITA as me? :D7
[22:02] <cjwatson> Assuming you mean "o jordan" rather than infinity, I've not dealt with him much before.
[22:02] <cjwatson> Begging really isn't helpful.
[22:04] <Daviey> If anyone wants to review freeipmi, please feel free to reject if you feel the -Wunused-result is too intrusive... I actually don't think ti's worth it.... but MIR suggetsed it.
[22:04] <Daviey> (my patch, so i won't sob)
[22:07] <cjwatson> Daviey: I don't think it's too intrusive, but it's buggy.  You've declared _err_exit's second parameter as int, but are passing strerror() to it.
[22:08] <Daviey> cjwatson: Ah.
[22:08] <Daviey> I suck.
[22:09] <cjwatson> And you seem to be using it from different translation units without a header declaration, which is confusing
[22:11] <cjwatson> It kind of looks like you meant this to be varargs or something, but that's probably OTT in context.  You could just pass errno (being careful to call it something else in the formal parameter) and append : %s directly in _err_exit, then exit 1 or something
[22:11] <cjwatson> The rest of it looks fine, but check over the compiler warnings since this looks as though it should have triggered some ...
[22:13] <Daviey> Yes, looking now, i can see warning: passing argument 2 of '_err_exit' makes integer from pointer without a cast
[22:13] <cjwatson> Daviey: so rejecting for now, but happy to review a corrected version tomorrow if nobody's done it before I fall back out of bed
[22:14] <Daviey> I won't touch it before i awake. Thanks for looking.
[22:29] <Daviey> A maas package is just about to hit the queue which is quite intrusive, please do not accept it.  I want to review in the morning.
[22:29] <Laney> you should reject it then :-)
[22:30] <infinity> Daviey: Yeah, just reject it and review and accept it from REJECTED in the morning.
[22:30] <infinity> Daviey: Much saner than trusting everyone to read backscroll. ;)
[22:31] <Daviey> Good thinking