=== cpaelzer_ is now known as cpaelzer [11:00] @pilot in === ChanServ changed the topic of #ubuntu-devel to: Archive: Mantic Open | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Bionic-Lunar | Patch Pilots: waveform [11:59] morning [11:59] starting my +1 maintenance shift week [11:59] hit me [12:13] ahasenack: ignore all r-bioc-*, they're in transition in Debian [12:13] phew [12:13] $ ./find-proposed-cluster -s mantic [12:13] r-bioc-biocgenerics 160 [12:13] (...) [12:14] i retried all build failures in mantic-proposed [12:17] looking at haskell then [13:38] https://launchpad.net/ubuntu/+source/haskell-gi-gio/2.0.30-1build3 might be what will unblock it, armhf build ongoing [13:38] 5h building, and counting, ouch [13:43] hm, previous build took just 1h18min [13:43] wonder if it's stuck [13:48] LocutusOfBorg: I wonder if https://launchpad.net/ubuntu/+source/haskell-gi-gio/2.0.30-1build3/+build/26482467 should be cancelled, and retried [13:48] it feels stuck [13:51] * ahasenack hovers over the cancel button [14:04] restarted the build [14:06] * ginggs looks at libmedia-convert-perl vs ffmpeg, which might unblock bctoolbox [14:17] this will take a bit, so next. rust-phf [14:28] ahasenack, thanks, I'm monitoring them, haskell and ocaml should go in 2-3 days [14:28] (ocaml today I guess) [14:28] ffmpeg/vim are the worst painful ones [14:28] they block *lots* [14:28] that armhf builder is slow [14:28] I'm doing a build in another arm machine, started both at the same time, and mine is at [240 of 300] [14:29] while the lp one is at [15 of 300] [14:29] but at least it's not stuck [14:29] for some reason armhf for ghc regressed a lot [14:29] and we never found why [14:29] you mean in terms of speed? [14:29] yes, ghc became SO SLOW in building [14:29] :/ [14:29] even ghc in building itself [14:29] and I remember vorlon debugging it some bits [14:30] one thing I noticed is that it's not using parallel jobs [14:30] I had to switch to unregisterised build for armhf, that worked better, and then use that ghc to build the registerised one [14:30] even though I passed --jobs=80 [14:30] so 79 cores are idling now in this machine [14:43] ahasenack, I don't remember, let me chec [14:43] *check [14:44] hm, anyone familiar with rust-pallete here? It has a versioned build-dep: https://salsa.debian.org/debian/rust-palette/-/blob/debian/latest/debian/control#L15 [14:44] on librust-phf-0.10, and we now have 0.11 [14:44] sounds like a simple bump, and it builds fine with the new one [14:44] * ahasenack wonders how that migrated in debian [14:45] well, should be simple to prove now that -palette is an ftbfs in debian [14:46] Fix "support for DEB_BUILD_OPTIONS=parallel=n broken" dh_haskell_build: Remove support for parallel=n. (Closes: #536589) [14:46] it's even uninstallable in debian now, and salsa doesn't take merges for it [14:46] wow, 2009 [14:47] LocutusOfBorg: maybe it's fixed now? :) [14:47] libmedia-convert-perl passed on ppc64el and s390x, holding thumbs for arm64 [14:48] (seems the new ffmpeg pushed into OOM, I suspect that because it passed on armhf and amd64) [14:50] ahasenack, https://github.com/ghc-proposals/ghc-proposals/pull/540 [14:50] -ubottu:#ubuntu-devel- Pull 540 in ghc-proposals/ghc-proposals "jsem: parallelism semaphores" [Merged] [14:50] maybe cabal will support them *soon* [14:51] https://github.com/haskell/cabal/issues/976 [14:51] -ubottu:#ubuntu-devel- Issue 976 in haskell/cabal "Parallelise cabal build over modules" [Open] [14:51] https://well-typed.com/blog/2023/08/reducing-haskell-parallel-build-times/ [14:51] something from this month, after 14 years it got implemented :D [14:51] so, somebody has to put a note on calendar to enable again parallel building? [14:53] that pr was merged last year, [14:54] yeah [14:54] 976 still open [15:01] @pilot out === ChanServ changed the topic of #ubuntu-devel to: Archive: Mantic Open | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Bionic-Lunar | Patch Pilots: N/A [15:27] wow: dpkg-deb (subprocess): compressing tar member: zstd error: Allocation error : not enough memory [15:28] $ free -h [15:28] total used free shared buff/cache available [15:28] Mem: 250Gi 98Mi 249Gi 232Ki 728Mi 250Gi [15:28] armhf userspace, I wonder if that was it (32bit: does armhf have something similar to PAE?) [15:28] but the kernel is arm64 (it's the trick of running armhf lxd on arm64 host) [15:29] LocutusOfBorg: that was my "local" build of haskell-gi-gio in armhf [15:30] * ahasenack -> lunch [15:30] meh, the kernel arm64 and chroot armhf is what was probably the culprit [16:24] ok, rust-palette 0.7.2+dfsg-3 will help/fix the rust-phf migration [16:24] it's building in debian [17:06] ooh, looks like bctoolbox and ffmpeg and friends are migrating \o/ [17:07] with that, I will EOD! [17:10] o/ [19:02] !dmb-ping [19:02] bdmurray, kanashiro, rbasak, seb128, sil2100, teward, utkarsh2102: DMB ping [19:03] rbasak: as stated last meeting by me, this one and the next one i have predefined work obligations. Can't be there today. [19:03] *is neck deep in training today8 [21:22] ddstreet: mapreri: I want to reject #2030675 from backports. No developer backing the request, and FTBFS as is. "It's possible the source package is too new, and would need modification" - not our job, and with no developer backing the request, I think this is a prime case for a rejection. Thoughts before I pull the "reject" button out? [21:22] Finished 31 seconds ago (took 7 hours, 18 minutes, 48.9 seconds) [21:22] finally [21:24] ahasenack: rust being chaotic? :P [21:25] haskell [21:25] build takes forever [21:31] ah, ye, that's one of the nastys to build