[00:02] <sergiusens> infinity: oh goodie, I had word that it wasn't going to be worked on
[00:02]  * sergiusens will just wait
[00:02] <infinity> Word from whom?
[00:03] <sergiusens> infinity: it's probably broken telephone (or whatever that game was called)
[00:20] <rsalveti> sergiusens: infinity: that's just because doko said he didn't have the time to fix such issues
[00:20] <rsalveti> but if you say you guys are on top of it, then even better
[00:21] <infinity> rsalveti: We're looking into it.  doko has a workaround if we can't get to a proper fix in a timely fashion.
[00:21] <infinity> rsalveti: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66368
[00:21] <rsalveti> great
[13:34] <arges> infinity: was maas ever granted an MRE for 1.7.5 in trusty?
[13:35] <arges> or is it still a no-go because of upgrading causing users to re-import images
[13:44] <sil2100> Hello! We would need an archive admin's assistance with signing-off a package that's ready for publishing
[13:45] <sil2100> We need an archive admin's +1 and review since it adds a new binary package
[13:45] <sil2100> https://ci-train.ubuntu.com/job/ubuntu-landing-017-2-publish/lastSuccessfulBuild/artifact/telepathy-ofono_packaging_changes.diff <- that's the packaging diff
[13:45] <sil2100> infinity, slangasek: ^ ?
[14:27] <infinity> arges: The import thing was fixed, or so they told me.  There's no blanket wave-it-through exception granted yet, but I intend to let that one in and see how it goes.
[14:28] <arges> infinity: shall i do the honors then
[14:28] <infinity> sil2100: Looks reasonable to me.
[14:28] <infinity> arges: I was going to look at it more closely after breakfast.
[14:28] <sil2100> infinity: \o/ thanks, publishing
[14:28] <arges> infinity: ack
[18:47] <ChrisTownsend> arges: Hi!  Are you on SRU duty today?
[18:49] <arges> ChrisTownsend: yes
[18:49] <arges> ChrisTownsend: lemme guess... unity?
[18:50] <ChrisTownsend> arges: Cool, yep, it reverts a commit that was causing a crash regression.
[18:50] <ChrisTownsend> arges: Do you need the debdiff?
[18:50] <arges> ChrisTownsend: makes it easier
[18:50] <arges> than doing it the manual way
[18:50] <ChrisTownsend> arges: Ok, give me 5 minutes and I'll attach it to the bug.
[18:58] <ChrisTownsend> arges: Ok, added the debdiff to https://bugs.launchpad.net/unity/+bug/1451613.
[18:59] <arges> ChrisTownsend: looks good accepted
[18:59] <ChrisTownsend> arges: Thanks!
[20:50] <boiko> hi, I have changed telepathy-ofono packaging to not build one of its subpackages in arm64, powerpc and ppc64el, and now update excuses says there are some old binaries left on those platforms
[20:50] <cjwatson> boiko: looking
[20:50] <boiko> cjwatson: it is the telepathy-ofono-ril-mc-plugin that is not building on arm64, powerpc and ppc64el anymore
[20:50] <cjwatson> boiko: rationale for the change?
[20:52] <boiko> cjwatson: the package needs libhybris-utils which is not available on those arches
[20:53] <cjwatson> boiko: aha, makes sense.  removed
[20:54] <boiko> cjwatson: as we initially landed this change in the vivid overlay ppa, we didn't realize the dependency was not available there
[20:54] <cjwatson> right
[21:13] <tedg> infinity, Were you able to have any luck with Mir on PPC?
[21:14] <infinity> tedg: I made a bit of progress, then got a bit distracted from my debugging, sorry.
[21:15] <tedg> infinity, K, I guess the question is more "close" or "delete all the binaries" :-)
[21:15]  * tedg assumes everyone builds from source
[21:16] <infinity> tedg: I think it's fixable, and fixing is the right direction to be going in.  Just need to dig up a round tuit or two to attack a weird testsuite crash with gdb.  Or ignore that test for the time being.
[21:18] <infinity> Well, there's also this failure, which just reeks of bad math somewhere:
[21:18] <infinity> Expected: is equal to 0x100349df340
[21:18] <infinity>   Actual: 0x349df340 (of type gbm_device*)
[21:19] <cjwatson> that reeks of G_POINTER_TO_INT abuse or whatever it is
[21:19] <tedg> Hmm, kgunn do you know who infinity could ping about that? ^
[21:20] <kgunn> tedg: infinity might ping racarr or kdub in #ubuntu-mir
[21:24] <boiko> cjwatson: what does the regression result mean? should I do something on my side?
[21:25] <cjwatson> boiko: I don't know, ask CI folks
[21:26] <infinity> Not shockingly, I'm amazingly suspicious of any function called "reinterpret_cast"...
[21:26] <infinity> Dear C++, WTF.
[21:27] <cjwatson> not really a function
[21:27] <infinity> cjwatson: A compiler directive, apparently, but yes.
[21:28] <slangasek> I hear C++11 adds a philip_glass_interpretation < > directive
[22:02] <infinity> tedg: Oh, FFS.  I should have tried ppc before ppc64el.
[22:02] <infinity> tedg: http://bazaar.launchpad.net/~mir-team/mir/development-branch/view/head:/src/server/scene/gl_pixel_buffer.cpp#L59
[22:02] <infinity> tedg: That's a bit more porting work than I was willing to sign up for.
[22:02] <infinity> tedg: So, yeah, let's look into removals. :(
[22:03] <tedg> Haha, funny. At least it's documented :-)
[22:03] <tedg> infinity, Can we just disable the tests?
[22:04] <infinity> tedg: I suppose we could do that instead.  But we run into the potential for breaking rdeps if something links against a Mirish thing and then actually uses it.
[22:04] <infinity> tedg: But, in most cases, I imagine the rdeps are fairly lame, except when running in a unity8/mir world.
[22:04] <tedg> infinity, At least in the UAL case you'd have to be using a pretty desktop specific feature…
[22:07] <infinity> tedg: The only scary rdep is gtk3.  Which is kinda super scary.  But it currently doesn't try to link against libmir on ppc, and we can just keep it that way.
[22:07] <infinity> tedg: All the other rdeps seem pretty mir-specific to start with, so I doubt we can make things worse by letting them build but suck a little.
[22:08] <cjwatson> I remember looking at that and thinking it wasn't *super* difficult ...
[22:08] <cjwatson> IIRC it's just that file
[22:08] <infinity> cjwatson: Well, there are two other bugs to hunt too, but yes, I think that file (that function, even) might be the only thing that needs *porting*, rather than bug fixing.
[22:08] <tedg> infinity, WFM, at least in the near term I don't imagine we'll have a lot of Mir users on PPC.
[22:09] <infinity> cjwatson: But I'm not in the mood to wrap my brain around endian transforms of whatever that is (pixel colourspaces?)
[22:09] <cjwatson> It strikes me as the sort of porting that needs pencil and paper, indeed
[22:10] <infinity> tedg: For big-endian PPC, it might become relevant in the community as the world moves more towards Mir, Wayland, or bust.
[22:10] <tedg> Back a long time ago (before gcc 3) that was the only case I found that tail recursion was faster than a for loop :-)
[22:10] <infinity> tedg: For ppc64el, IBM has made noises about desktoppish things "in the future".
[22:10] <infinity> tedg: But neither is today.
[22:10] <tedg> That would be kinda cool. I have a warm spot in my heart for PPC.
[22:11] <tedg> Though I imagine I'll have an arm64 desktop first :-/
[22:11] <infinity> tedg: Well, unless some OpenPower partners go small/low-power in the near future, any POWER8 "desktop" kit would be high end workstations right now, not exactly home user stuff.
[22:12] <tedg> DON'T JUDGE MY HOME!
[22:12] <tedg> ;-)
[22:12] <infinity> tedg: But there's been noise about shrinking P8 and P9 for more specialised uses, so never say never.
[22:12] <infinity> That said, with the whole CAPI thing, a P8 workstation would be a force to be reckoned with.
[22:12] <infinity> SO MUCH GPU BANDWIDTH.
[22:13]  * infinity shakes.
[22:16] <infinity> tedg: So, I'll hang on to this WIP branch, but lemme test if it vomits out interesting results if I just skip the testsuite.
[22:17] <tedg> infinity, Cool, thanks!