[03:39] <Son_Goku> RAOF, hey, it seems like your Git repository is incomplete
[03:39] <Son_Goku> it's missing all the newer tags from Launchpad
[03:40] <RAOF> As in 0.28? Yeah, probably.
[03:40] <Son_Goku> yeah
[03:40] <Son_Goku> RAOF, also, now that it's on GitHub officially-ish, I could submit my spec to be merged into the repository
[03:41] <Son_Goku> you might want to talk to mvo about how he manages to update the spec and debian stuff for the snapd repo and implement the same thing for mir
[03:41] <Son_Goku> and wire up CI building for Mir on Fedora
[03:41] <RAOF> Pull requests welcome!
[03:42] <RAOF> (CLA permitting, I think?)
[03:43] <RAOF> Yeah, building on Fedora in CI should be a simple as adding a Fedora lxd image.
[03:44]  * RAOF loves that Riot has a “you seem to be shaking the phone in frustration, would you like to file a bug report?” dialogue.
[03:45] <RAOF> Now (?) that we've disabled merge proposals on LP we'll need to do a final sync from there.
[03:46] <Son_Goku> I signed the CLA last year for snapd stuff
[03:46] <Son_Goku> I believe that still applies for Mir
[03:46]  * Son_Goku grumbles slightly about the continued existence of the CLA...
[03:49] <Son_Goku> hmm, looks like there's something wrong with building the package with unit tests enabled
[03:52] <Son_Goku> extracting debug info from /builddir/build/BUILDROOT/mir-0.28.1-1.fc26.x86_64/usr/bin/mir_unit_tests_mesa-kms
[03:52] <Son_Goku>  /usr/lib/rpm/debugedit: canonicalization unexpectedly shrank by one character
[03:52] <Son_Goku> RAOF ^
[03:52] <Son_Goku> something weird is going on with how mir_unit_tests_* binaries are being compiled
[03:53] <RAOF> ??!
[03:54] <Son_Goku> this usually happens when something has something like a double slash is encoded in some path
[03:55] <Son_Goku> because canonicalization of file paths would reduce it by a character
[03:55] <Son_Goku> so like somewhere in CMake, these things are getting compiled with paths that have double slashes in it
[03:55] <RAOF> mir_unit_tests_mesa-kms shouldn't *have* any debug info; it's a symlink!
[03:56] <RAOF> But that's entirely possible
[03:56] <Son_Goku> no it isn't
[03:56] <Son_Goku> it's legit binary
[03:56] <Son_Goku> -rwxr-xr-x. 1 root root 118M Nov  1 23:52 mir_unit_tests_mesa-kms
[03:56] <Son_Goku> it's also HUGE
[03:57] <duflu> Son_Goku, https://bugs.launchpad.net/mir/+bug/1473330  :)
[03:57] <ubot5> Ubuntu bug 1473330 in Mir "mir_*_tests binaries are suspiciously large. Up to 248MB before stripping." [Medium,Triaged]
[03:59] <Son_Goku> RAOF: https://paste.fedoraproject.org/paste/jY4AHO9htZkJdk2nEapVlw
[03:59] <Son_Goku> this is what the built tree looks like
[04:00] <Son_Goku> debug symbols not finished splitting, though
[04:02] <RAOF> Hm. Have we changed the way we wrap this executables?
[04:02] <Son_Goku> well, this is the first time I was able to build the unit tests
[04:02] <Son_Goku> they've been broken before
[04:03] <RAOF> Hm, no.
[04:03]  * Son_Goku is sending a build of mir 0.28.1 without unit tests to build in copr
[04:04] <Son_Goku> once we can get the unit tests, then we'll be in a state where putting this in Fedora proper could be a thing
[04:04] <RAOF> In my build tree, mir_* is a symlink to mir_*.bin.
[04:04] <RAOF> Oh, sorry, wrong.
[04:04] <RAOF> mir_* is a symlink to wrapper
[04:04] <Son_Goku> that seems like debian foo
[04:04] <Son_Goku> because it's not here
[04:04] <RAOF> (Which sets some environment variables and then calls argv[0].bin)
[04:05] <RAOF> That's not debian foo; I'm running a raw cmake build.
[04:05] <RAOF> I don't build packages during development ;)
[04:06] <RAOF> All those are being bulit from CMakeLists.txt with mir_add_wrapped_executable(), which is defined in cmake/MirCommon.cmake
[04:07] <RAOF> Oooooh1
[04:07] <RAOF> !
[04:07] <Son_Goku> https://paste.fedoraproject.org/paste/yg7YM9HFJSPR9a93cGIpmQ
[04:07] <RAOF> No, that's right.
[04:07] <RAOF> You've run make install.
[04:07] <Son_Goku> this is my cmake command
[04:07] <RAOF> As is right and proper.
[04:07] <RAOF> And we don't use the wrapper on installed binaries, because we can assume the libraries are also installed.
[04:07] <RAOF> Be about your business ;)
[04:07] <Son_Goku> okay then
[04:07] <Son_Goku> :)
[04:07] <Son_Goku> you had me panicking for a bit there
[04:10] <Son_Goku> https://copr.fedorainfracloud.org/coprs/ngompa/Mir/build/656918/
[04:10] <Son_Goku> no unit tests yet
[04:10] <Son_Goku> since, well, debugedit says there's something wrong with them
[04:11] <RAOF> Hrm.
[04:12] <RAOF> “strings bin/mir_unit_tests_mesa-kms.bin| rg //” does not show up any // paths for me.
[04:13] <Son_Goku> hmm
[04:14] <Son_Goku> the person who would know more about this is mjw (Mark Wielaard) in #rpm.org
[04:14] <Son_Goku> he is the one who hacks on debugedit and knows more than anyone else I know about debug symbol stuff
[04:14] <Son_Goku> but he's not online yet
[04:14] <RAOF> There are a bunch of PATH/../../foo.h in there.
[04:14] <Son_Goku> that could do it too
[04:15] <Son_Goku> but I'll probably be passing out soon, since it is past midnight here
[04:15] <RAOF> But I'd expect that to result in canonicalisation dropping more than one character :)
[04:15] <RAOF> Heh.
[04:15] <RAOF> Good news, everyone!
[04:15] <Son_Goku> it's easy enough for someone to reproduce though
[04:16] <RAOF> I've been accidentally testing that GNOME Shell passes my wayland conformance suite.
[04:16] <RAOF> And it does!
[04:16] <Son_Goku> RAOF: ?
[04:16] <Son_Goku> :)
[04:17] <RAOF> (Not that there's much of a conformance suite yet, mostly the infrastructure for one)
[04:19] <Son_Goku> anyway, anyone can reproduce the problem I have by downloading https://copr-be.cloud.fedoraproject.org/results/ngompa/Mir/fedora-26-x86_64/00656918-mir/mir-0.28.1-0.fc26.2.src.rpm and using `rpmbuild -ra --with check mir-0.28.1-0.fc26.2.src.rpm`
[04:20] <Son_Goku> I suppose on a debian system with rpm installed, if you have the debian builddeps installed, you could do it with `rpmbuild -ra --nodeps --with check mir-0.28.1-0.fc26.2.src.rpm`
[04:20]  * RAOF fires up positive-seasnail
[04:26] <RAOF> Son_Goku: What's the way to install all the build-deps of an rpm?
[04:29] <Son_Goku> $ rpm -ivh https://copr-be.cloud.fedoraproject.org/results/ngompa/Mir/fedora-26-x86_64/00656918-mir/mir-0.28.1-0.fc26.2.src.rpm
[04:29] <Son_Goku> $ sudo dnf builddep -D "_with_check 1" --spec ~/rpmbuild/SPECS/mir.spec
[04:31] <Son_Goku> (ordinarily, we'd use the SRPM directly, but we want more builddeps than the default)
[04:31] <Son_Goku> and once you have the builddeps installed
[04:31] <Son_Goku> you can just do the following:
[04:31] <Son_Goku> $ rpmbuild -ba --with check ~/rpmbuild/SPECS/mir.spec
[04:39] <RAOF> Ta
[04:47] <RAOF> Ah.
[04:47] <RAOF> I think it's more likely to be a trailing /
[04:58] <RAOF> Son_Goku: Fixed in Fedora 27, apparently :)
[05:00] <RAOF> https://bugzilla.redhat.com/show_bug.cgi?id=304121
[05:00] <ubot5> bugzilla.redhat.com bug 304121 in rpm "debugedit prints: canonicalization unexpectedly shrank by one character" [Medium,Closed: rawhide]
[06:56] <RAOF> Yes! And now that it's hooked up, Mir also passes this particular test!
[07:01] <RAOF> So I guess I get this applied to a git tree, break up some megafiles, and propose.
[13:32] <Son_Goku> [09:03:58 AM]  <mjw>	pmatilai, O, right, yes, it might also be the some source file is encoded as some/dir//source.c with a double slash which some code might or might not remove and then the lengths don't add up.
[13:32] <Son_Goku> RAOF ^
[19:30] <Pharaoh_Atem> RAOF: so I ran ctest and got the following error:
[19:30] <Pharaoh_Atem> 17: [ FAILED ] 1 test, listed below:
[19:30] <Pharaoh_Atem> 17: [ FAILED ] MesaGraphicsPlatform.connection_ipc_package
[19:31] <Pharaoh_Atem> 17: [ RUN ] MesaGraphicsPlatform.connection_ipc_package
[19:31] <Pharaoh_Atem> 17: [2017-11-02 14:54:39.357042] mesa-kms: Using DRM device /dev/dri/card1
[19:31] <Pharaoh_Atem> 17: [2017-11-02 14:54:39.357132] mesa-kms: Using DRM device /dev/dri/card0
[19:31] <Pharaoh_Atem> 17: unknown file: Failure
[19:31] <Pharaoh_Atem> 17:
[19:31] <Pharaoh_Atem> 17: Unexpected mock function call - returning default value.
[19:31] <Pharaoh_Atem> 17: Function call: drmAuthMagic(8, 32659)
[19:31] <Pharaoh_Atem> 17: Returns: 0
[19:31] <Pharaoh_Atem> 17: Google Mock tried the following 1 expectation, but it didn't match:
[19:31] <Pharaoh_Atem> 17:
[19:31] <Pharaoh_Atem> 17: /builddir/build/BUILD/mir-0.28.1/tests/unit-tests/platforms/mesa/kms/test_platform.cpp:108: EXPECT_CALL(mock_drm, drmAuthMagic(mtd::IsFdOfDevice("/dev/dri/card0"),_))...
[19:31] <Pharaoh_Atem> 17: Expected arg #0: Is an fd of DRM device /dev/dri/card0 (one of: { 10 })
[19:31] <Pharaoh_Atem> 17: Actual: 8
[19:31] <Pharaoh_Atem> 17: Expected: to be called once
[19:31] <Pharaoh_Atem> 17: Actual: never called - unsatisfied and active
[19:31] <Pharaoh_Atem> 17: /builddir/build/BUILD/mir-0.28.1/tests/unit-tests/platforms/mesa/kms/test_platform.cpp:108: Failure
[19:32] <Pharaoh_Atem> 17: Actual function call count doesn't match EXPECT_CALL(mock_drm, drmAuthMagic(mtd::IsFdOfDevice("/dev/dri/card0"),_))...
[19:32] <Pharaoh_Atem> 17: Expected: to be called once
[19:32] <Pharaoh_Atem> 17: Actual: never called - unsatisfied and active
[19:32] <Pharaoh_Atem> 17: [ FAILED ] MesaGraphicsPlatform.connection_ipc_package (7 ms)
[19:35] <RAOF> Pharaoh_Atem: Harrumph. Please file a bug, that's the test having an implicit dependency on the order udev enumerates devices in, I think.
[19:45] <Pharaoh_Atem> okay
[19:45] <Pharaoh_Atem> where do I file bugs now?
[19:45] <Pharaoh_Atem> is it still LP?
[19:46] <Pharaoh_Atem> ah, it moved to GitHub too
[19:46] <Pharaoh_Atem> that makes things easier
[19:48] <Pharaoh_Atem> RAOF: https://github.com/MirServer/mir/issues/8
[19:48] <RAOF> Ta