| RAOF | Hmmmmmmmmmmmm: | 01:20 |
|---|---|---|
| RAOF | ``` | 01:20 |
| RAOF | /usr/include/c++/12/bits/char_traits.h:431:56: error: ‘void* __builtin_memcpy(void*, const void*, long unsigned int)’ accessing 9223372036854775810 or more bytes at offsets -4611686018427387902 and [-4611686018427387903, 4611686018427387904] may overlap up to 9223372036854775813 bytes at offset -3 [-Werror=restrict] | 01:20 |
| RAOF | ``` | 01:20 |
| RAOF | I feel that accessing `9223372036854775810` or more bytes *might* be a mistake 🤷♀️ | 01:20 |
| RAOF | Saviq: I've enabled ppc64el in ppa:mir-team/dev on the basis that we reasonably frequently see FTBFS on that platform and it would be nice to get a heads up before uploading to the archive. | 07:16 |
| Saviq | RAOF (he/they): ofc, we should have the same coverage there | 07:42 |
| RAOF | Damnit, how do you really turn off LTO now? | 07:53 |
| alan_g[m] | <RAOF> "Damnit, how do you really turn..." <- It doesn't "just work"? | 08:02 |
| Saviq | WDYM? | 08:05 |
| Saviq | `-DMIR_LINK_TIME_OPTIMIZATION=OFF` | 08:05 |
| Saviq | Maybe something overrode it? Look for `COMMON_CONFIGURE_OPTIONS` in the log, as it prints what's passed to CMake? | 08:06 |
| RAOF | (Sorry, bleeding nose emergency) | 08:30 |
| RAOF | Ubuntu enables LTO by default now. `DEB_BUILD_MAINT_OPTIONS=optimize=-lto` is meant to turn that off, but didn't on this build. | 08:31 |
| RAOF | Damnit, miriway crashed and I lost the half-filed gcc bug. | 08:32 |
| RAOF | Oh, well. EOW it is! | 08:32 |
| Saviq | Right, and we do it if `DEB_BUILD_OPTIONS=noopt`; but I suppose we may need to set it if we encounter `MIR_LINK_TIME_OPTIMIZATION`? Or rework things in a way that one follows the other? | 09:33 |
| Saviq | I mean, this should also look for `optimize=-lto`, not just for `noopt`? | 09:34 |
| Saviq | https://github.com/MirServer/mir/blob/aa0a01e1b9d46cebf47650243dbf2245c5e9025f/debian/rules#L34-L38 | 09:34 |
| alan_g[m] | <Saviq> "I mean, this should also look..." <- https://github.com/MirServer/mir/blob/aa0a01e1b9d46cebf47650243dbf2245c5e9025f/debian/rules#L34-L38 | 11:14 |
| alan_g[m] | I think RAOF says we set that, but it didn't work. But I don't know which build he's referring to | 11:14 |
| Saviq | I think this: | 11:15 |
| Saviq | https://launchpad.net/ubuntu/+source/mir/2.9.0-0ubuntu1 | 11:15 |
| Saviq | Unless something completely different | 11:16 |
| alan_g[m] | Hmm looking at the ppc64el build I see the strange error he quoted. But there's nothing disabling LTO | 11:21 |
| alan_g[m] | Or trying to | 11:21 |
| Saviq | Yup, that. | 11:22 |
| alan_g[m] | So, I think the idea is to check for ppc64el and disable LTO? | 11:22 |
| Saviq | That's what I was thinking, I'll try in a test PPA. | 11:24 |
| Saviq | alan_g if you turn CapsLock on, then Alt+Tab to another app, do you get small caps? | 11:25 |
| alan_g[m] | Was about to switch to Miriway to try that. Having a reproducer would be very helpful | 11:25 |
| alan_g[m] | I so seldom hit CapsLock... | 11:26 |
| alan_g[m] | \o/ Capslock only unlocks on the original app | 11:28 |
| alan_g[m] | Or... not quite consistent. But definitely enough of a reproducer | 11:31 |
| Saviq | Yup | 11:35 |
| Saviq | Ah. I think you can only unlock in the original app, but need to kick it twice. | 11:49 |
| RAOF | > <@alan_g:matrix.org> https://github.com/MirServer/mir/blob/aa0a01e1b9d46cebf47650243dbf2245c5e9025f/debian/rules#L34-L38 | 11:51 |
| RAOF | > I think RAOF says we set that, but it didn't work. But I don't know which build he's referring to | 11:51 |
| RAOF | I'm referring to my local build, where I set `MIR_LINK_TIME_OPTIMIZATION=OFF`, but it still gets built with LTO. | 11:51 |
| RAOF | (in a cross-building `sbuild`0 | 11:52 |
| RAOF | s/0/)/ | 11:52 |
| RAOF | Yeah, please see if you can disable LTO in a PPA build. | 11:53 |
| RAOF | Ideally, I guess, we'd just not build any of `tests/` with LTO? There's no particularly good reason to, and if we didn't we could turn off fat LTO and speed the build up a little. | 11:54 |
| RAOF | All LTO of `tests/` seems to accomplish is to help gmock expose bugs in the LTO toolchain 🤪 | 11:55 |
| RAOF | And now, good night! | 11:56 |
| Saviq | Well, that should be easy, just set `-fno-lto` in `tests/CMakeLists.txt`? Probably no need for `-fno-fat-lto-objects` then? | 11:59 |
| Saviq | Good night! | 11:59 |
| Saviq | I mean... https://github.com/MirServer/mir/blob/aa0a01e1b9d46cebf47650243dbf2245c5e9025f/tests/CMakeLists.txt#L116-L117 | 12:03 |
| Saviq | <alan_g[m]> "I so seldom hit CapsLock..." <- I have it mapped to Compose in GNOME, which results in me hitting it every now and then… and my caps go wild :) | 12:19 |
| Saviq | Which reminds me... | 12:19 |
| -GitHub[m]:#mir-server- **[MirServer/mir]** Saviq opened [issue #2630](https://github.com/MirServer/mir/issues/2630): Support mapping a Compose key... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/532fb44cc237dd2eb806cad4e81cb00aa78cf4c0) | 12:21 | |
| Saviq | RAOF (he/they) this got LTO disabled, but the error is the same, if that's what you were fighting: | 13:12 |
| Saviq | https://launchpad.net/~saviq/+archive/ubuntu/mir-test/+build/24348505 | 13:12 |
| -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths assigned AlanGriffiths to [issue #2592](https://github.com/MirServer/mir/issues/2592): CapsLock state gets stuck | 14:07 | |
| Saviq | Good weekend all o/ | 16:34 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!