#archlinux32 | Logs for 2019-06-28
Back
[00:15:31] -!- eschwartz has quit [Ping timeout: 258 seconds]
[01:09:13] <elibrokeit> deep42thought, abaumann: so someone managed to build llpp I see... I noticed a while ago it did not build and I realized my package is broken on parallel builds.
[01:11:15] <elibrokeit> Btw what's the makedepends on libglvnd that you add?
[01:15:22] <buildmaster> pentium4/sagemath-doc is broken (says buildknecht).
[01:54:45] -!- eschwartz has joined #archlinux32
[02:02:14] -!- eschwartz has quit [Read error: Connection reset by peer]
[02:03:57] -!- eschwartz has joined #archlinux32
[02:09:48] -!- yans has joined #archlinux32
[02:11:37] -!- samantaz has quit [Read error: Connection reset by peer]
[02:24:38] -!- eschwartz has quit [Read error: Connection reset by peer]
[02:26:09] -!- eschwartz has joined #archlinux32
[02:34:48] -!- MrBIOS has joined #archlinux32
[02:48:18] -!- eschwartz has quit [Read error: Connection reset by peer]
[02:49:22] -!- eschwartz has joined #archlinux32
[03:13:23] -!- eschwartz has quit [Read error: Connection reset by peer]
[03:14:34] -!- eschwartz has joined #archlinux32
[03:32:49] -!- ofara_ has joined #archlinux32
[04:04:43] -!- eschwartz has quit [Read error: Connection reset by peer]
[04:05:46] -!- eschwartz has joined #archlinux32
[04:27:13] -!- thePiGrepper has quit [Ping timeout: 246 seconds]
[04:30:18] -!- ofara has quit [Ping timeout: 272 seconds]
[04:50:29] -!- eschwartz has quit [Read error: Connection reset by peer]
[04:51:59] -!- eschwartz has joined #archlinux32
[05:03:47] -!- eschwartz has quit [Ping timeout: 245 seconds]
[05:44:38] -!- MrBIOS has quit [Quit: MrBIOS]
[05:46:56] -!- MrBIOS has joined #archlinux32
[05:58:25] -!- yans has quit [Ping timeout: 248 seconds]
[08:38:00] -!- deep42thought has joined #archlinux32
[08:38:00] <buildmaster> Hi deep42thought!
[08:38:00] <buildmaster> !rq deep42thought
[08:38:01] <phrik> buildmaster: <deep42thought> "und wieder ist ein Tag vollbracht und wieder ist nur Murks gemacht"
[08:40:44] <deep42thought> elibrokeit: dunno what the additional makedepends is for (my guess: there is a depends on libgl which needs to be satisfied by libglvnd at build time)
[08:40:58] -!- abaumann has joined #archlinux32
[08:40:58] <buildmaster> Hi abaumann!
[08:40:58] <buildmaster> !rq abaumann
[08:40:59] <phrik> buildmaster: <abaumann> I hope the placebo effect works also for coffeine :-)
[08:41:03] <deep42thought> Hi abaumann!
[08:41:09] <abaumann> hi deep42thought
[08:41:10] <deep42thought> How was the bike trip?
[08:41:19] <abaumann> I'm back, together with my lazy build slaves..
[08:41:26] <deep42thought> !grab abaumann
[08:41:26] <phrik> deep42thought: Bingpot!
[08:42:02] <abaumann> On the db-update -p: I didn't use -p anymore, I just pushed individual packages
[08:42:18] <deep42thought> hmm, ok
[08:42:21] <abaumann> OTOH the thing could write a log file, so we see who did what when. :->
[08:42:36] <deep42thought> oh, indeed
[08:43:14] <abaumann> tirsek: corrupt GNU_PROPERTY_TYPE (5) size: 0, yeah noticed that too, it's the dynamic loader croaking about non-existing GNU special stuff in ELF binaries.
[08:44:51] <deep42thought> hmm, I wonder where, how and what we should log
[08:45:04] <abaumann> mail_log, ssh_log, cli_log?
[08:45:13] <deep42thought> ah, in the database
[08:45:17] <abaumann> the problem is the who.
[08:45:20] <deep42thought> and only the manual commands
[08:45:27] <abaumann> If we both operate as master on the buildmaster.
[08:45:28] <deep42thought> lib/load-configuration >:-)
[08:45:35] -!- ofara_ has quit [Ping timeout: 252 seconds]
[08:45:49] <deep42thought> that script is sourced by literally all other scripts right at the beginning
[08:46:10] <deep42thought> it doesn't matter _who_ did something - it's only interesting _if_ it was done
[08:46:17] <abaumann> good place, though it should then be renamed to load-configuration-and-log-commands-to-database ;-)
[08:46:33] <abaumann> yeah. true.
[08:47:12] <abaumann> you rescheduled some packages? perlish stuff?
[08:47:20] <deep42thought> yes
[08:47:34] <abaumann> ah, explains the blue bump in the buildmaster graph :-)
[08:47:36] <deep42thought> those were still signed with tyzoid's now-expired key
[08:47:41] -!- ofara_ has joined #archlinux32
[08:47:42] <deep42thought> and I could not build lilypond-devel
[08:48:00] <abaumann> a -devel packages (abaumann sniggers)
[08:48:20] <deep42thought> it's more like lilypond-git
[08:48:21] <deep42thought> ...
[08:49:34] <abaumann> ah. that's an AUR package
[08:49:41] <deep42thought> yes
[08:49:56] <abaumann> so, the normal version is not "bleeding" enouhg. ;-
[08:49:57] <abaumann> )
[08:50:38] <deep42thought> stupid story: I compiled the newest (devel) version for crux, not noticing, that it was "devel", then I added "\version $version" strings in all my sources, then it refused to compile my notes on arch, so I built lilypond-devel, there, too
[08:52:55] -!- eschwartz[m] has quit [Write error: Connection reset by peer]
[08:52:55] -!- z3ntu has quit [Write error: Connection reset by peer]
[08:56:02] <abaumann> ERROR: Please install required programs: /usr/bin/fontforge >= 20110222 (installed: )
[08:58:01] <deep42thought> proposed log columns: id, date, command, parameters, shell
[08:58:11] <abaumann> shell?
[08:58:18] <deep42thought> if it was run from a shell or not
[08:58:26] <deep42thought> (e.g. systemd service)
[08:58:37] <deep42thought> or should we log the manual commands only?
[08:59:17] <abaumann> ah.
[08:59:30] <abaumann> yeah, no, good idea, didn't thing about that.
[08:59:42] <abaumann> *think
[09:00:08] <abaumann> fontforge: Unable to init server: Could not connect: Connection refused
[09:00:08] <abaumann> (fontforge:6): Gdk-WARNING **: 08:59:57.070: cannot open display:
[09:00:11] <abaumann> really now.
[09:00:46] <abaumann> so, I think the prober fails because it has no X environment
[09:02:15] <abaumann> umount /tmp, mount /tmp
[09:02:15] <abaumann> mount: /tmp: can't find in /etc/fstab.
[09:02:19] <abaumann> I love systemd
[09:02:36] <abaumann> not that /tmp is something you _should_ do actually in a running system.
[09:02:45] <abaumann> ..but OTOH, there _are_ typos. :-)
[09:02:58] -!- z3ntu has joined #archlinux32
[09:06:26] <buildmaster> i686/c2hs are broken (says eurobuild6-2).
[09:08:29] <deep42thought> log should work now ...
[09:08:34] <abaumann> (y)
[09:09:17] <deep42thought> hmm, this table is filling rapidly :-/
[09:09:32] <abaumann> you log _everything_ now?
[09:09:35] <deep42thought> already 23 commands logged
[09:09:36] <abaumann> pings for instance
[09:09:37] <deep42thought> yes
[09:09:40] <deep42thought> :-/
[09:10:20] <deep42thought> let's run it for some time and see how it evolves ...
[09:10:26] <deep42thought> I have to go afk for a while
[09:10:31] <abaumann> ok. cu later.
[09:10:35] <deep42thought> cu
[09:14:47] <buildmaster> i686/haskell-hackage-security is broken (says buildknecht2).
[09:30:15] <buildmaster> i686/haskell-iproute is broken (says buildknecht).
[09:37:39] -!- eschwartz[m] has joined #archlinux32
[09:39:03] <deep42thought> 803 logged commands so far
[09:45:50] <deep42thought> I'll skip logging "slave-build-connect" and "ping-from-slave": The former is useless, because everything is set up in environment variables, the latter is useless, too, because all details are already logged in ssh_log
[09:56:25] <abaumann> sound good
[09:56:39] <deep42thought> somehow it only safes the first parameter, though
[09:56:46] <deep42thought> the query looks sane
[09:57:56] <deep42thought> ah, stupid me
[09:58:17] <deep42thought> I still had a "group by" in my query looking for the newly logged command ...
[09:58:30] <deep42thought> all fine :-)
[10:01:38] <buildmaster> i686/haskell-socks are broken (says eurobuild3).
[10:01:46] <abaumann> /home/master/builder/bin/return-assignment: line 234: mismatched interface file versions (wanted : command not found
[10:01:58] <deep42thought> errr
[10:02:40] <deep42thought> try again
[10:02:51] <abaumann> I have to update the slaves?
[10:02:54] <deep42thought> no
[10:03:06] <abaumann> ah, I see /home/master is buildmaster side
[10:03:22] <deep42thought> I forgot to add "\" before continued lines in a find expression
[10:03:55] <deep42thought> I wonder why shellcheck did not complain ...
[10:06:50] -!- AndrevS has joined #archlinux32
[10:11:14] <deep42thought> https://archlinux32.org
[10:11:16] <phrik> Title: command-log (at archlinux32.org)
[10:16:11] <deep42thought> damn, I squashed a fly which was inside my monitor - I should have thought about how I get it out before doing so ...
[10:16:43] <abaumann> ..now you have to wait till the spider inside the monitor eats the fly ;-)
[10:17:19] <deep42thought> fun fact: the fly is /behind/ the lcd (but unfortunately in front of the backlight)
[10:17:56] <buildmaster> i686/haskell-torrent is broken (says buildknecht2) - I rescheduled: haskell-bencode.
[10:18:19] <abaumann> mmh. I wonder how it got there..
[10:18:55] <abaumann> ii-answer is in the command log?
[10:19:11] <deep42thought> yes
[10:19:11] <abaumann> well, no harm done, but..
[10:19:40] * deep42thought adds another command to the growing blacklist
[10:21:02] <buildmaster> pentium4/haskell-th-reify-many is broken (says eurobuild6-5) - I rescheduled: haskell-th-expand-syns.
[10:21:36] <deep42thought> ah, my broken-line-thing held back all the broken haskell packages :-/
[10:23:24] <abaumann> careful with the next mariadb upgrades..
[10:23:40] <buildmaster> i686/haskell-foundation is broken (says eurobuild6-3) - I rescheduled: haskell-basement.
[10:25:06] <deep42thought> yeah
[10:25:15] <deep42thought> I hesitate to do it ...
[10:25:28] <deep42thought> maybe it's wisest to set up the new database on buildmaster-new
[10:25:31] <abaumann> .. on the buildmsater-new first.
[10:25:33] <abaumann> exactly :-)
[10:25:34] <deep42thought> and simply play in a backup
[10:26:04] <abaumann> just updated my main webserver with customers on it, seems fine.
[10:33:05] <abaumann> mmh. My Raspberry PI 3 works in every socket, with every USB cable, but not if connected in the server room area to a power line..
[10:33:27] <abaumann> ..maybe too many machines are connected :-)
[10:33:42] <deep42thought> use thicker wires
[10:33:58] <deep42thought> I used to power all my machines from a single socket
[10:34:29] <deep42thought> ... because we had only one fuse for all the sockets in our flat anyways
[10:35:31] <abaumann> this would explain some of my power-outages when I enable turbo freeze in the freezer and in parallel start the toaster :-)
[10:35:41] <abaumann> everything is on one fuse
[10:36:00] <abaumann> oh well. I put a new power line from the other fuse there, see, what happens
[10:36:10] <buildmaster> i686/haskell-fast-logger is broken (says eurobuild6-4) - I rescheduled: haskell-easy-file.
[10:36:45] <abaumann> mmh. I think the new buildmaster can make good use of 2 threads (one for mysql and one for scripts, xz, bsdtar, git, etc.)
[10:36:59] <abaumann> That's what I can read from the stats of the current buildmaster
[10:37:24] <abaumann> one cpu >90%, one 0-80% doing something
[10:37:46] <deep42thought> yeah, I use 'xzcat database.xz | mysql -u root -p' to inject the backup
[10:38:10] <deep42thought> or are you talking about load on the current buildmaster and extrapolating to the new?
[10:38:23] <abaumann> current one. I'm referring to your email.
[10:38:31] <deep42thought> ah, ok
[10:38:47] <deep42thought> I sent cpu stats, too?
[10:38:48] <abaumann> I have no worries, we can not make good use of the rest of the powerof..
[10:38:53] <abaumann> no.
[10:38:56] <abaumann> just timing.
[10:39:08] <abaumann> but I life-checked in to on the old buildmaster
[10:39:13] <abaumann> *it
[10:39:42] <abaumann> I have funny nice ideas again :-)
[10:40:00] <deep42thought> !grab abaumann
[10:40:00] <phrik> deep42thought: Tada!
[10:40:21] <abaumann> build slaves: how many packets did they build per time? How long are they in which phase: "waiting for assignment", "building", "uploading"
[10:40:37] <deep42thought> sounds like a job for munin
[10:40:49] <abaumann> exactly :-)
[10:46:27] <buildmaster> i686/haskell-configurator-ng is broken (says buildknecht2) - I rescheduled: haskell-data-ordlist.
[10:52:44] <buildmaster> i686/haskell-src-meta is broken (says eurobuild6-2).
[10:54:43] <buildmaster> i686/haskell-config-ini is broken (says eurobuild6-3).
[10:54:44] -!- abaumann has quit [Ping timeout: 272 seconds]
[10:56:42] -!- abaumann has joined #archlinux32
[10:56:42] <buildmaster> Hi abaumann!
[10:56:42] <buildmaster> !rq abaumann
[10:56:43] <phrik> buildmaster: <abaumann> a babe is a checksum error
[11:02:41] <buildmaster> i686/haskell-singletons are broken (says eurobuild6-5) - I rescheduled: haskell-th-desugar.
[11:03:53] <deep42thought> actually, I would blacklist all the commands that are started from within slave-build-connect -- they are already logged in ssh_log
[11:11:13] <buildmaster> i686/haskell-bytestring-tree-builder is broken (says buildknecht2) - I rescheduled: haskell-base-prelude.
[11:12:11] <buildmaster> i686/haskell-hslua-module-text is broken (says eurobuild6-1).
[11:12:36] <buildmaster> i686/haskell-aeson is broken (says eurobuild6-2).
[11:13:00] <buildmaster> i686/haskell-hslua-module-system is broken (says eurobuild3).
[11:16:15] <buildmaster> i686/haskell-data-serializer is broken (says eurobuild6-5) - I rescheduled: haskell-parsers.
[11:18:38] <buildmaster> i686/haskell-http-api-data is broken (says buildknecht) - I rescheduled: haskell-cabal-doctest.
[11:18:43] <buildmaster> i686/haskell-base-prelude is broken (says eurobuild6-3).
[11:19:04] <buildmaster> pentium4/haskell-base-prelude is broken (says eurobuild6-1).
[11:19:15] <abaumann> uiui.
[11:19:50] <buildmaster> pentium4/haskell-either is broken (says eurobuild6-2).
[11:20:53] -!- abaumann has quit [Remote host closed the connection]
[11:23:37] <deep42thought> my local replication slave made the mariadb update without problems
[11:42:51] -!- abaumann has joined #archlinux32
[11:42:51] <buildmaster> Hi abaumann!
[11:42:51] <buildmaster> !rq abaumann
[11:42:53] <phrik> buildmaster: <abaumann> bugs are always smarter than you are :-)
[11:43:16] <abaumann> so, should I update the buildmaster?
[11:43:48] <buildmaster> pentium4/haskell-mono-traversable is broken (says eurobuild6-1) - I rescheduled: haskell-foldl.
[11:54:30] <buildmaster> i686/haskell-wai-cors are broken (says eurobuild3) - I rescheduled: haskell-base-unicode-symbols.
[11:56:21] <abaumann> let's do it.. I feel motivated to break things. ;-)
[11:57:08] -!- finsternis has quit [Excess Flood]
[11:57:52] -!- finsternis has joined #archlinux32
[12:00:11] <abaumann> hum. does restarting mysql on the buildmaster kill the replication slave?
[12:00:31] <buildmaster> pentium4/haskell-x509-system is broken (says eurobuild6-5) - I rescheduled: haskell-x509-store.
[12:07:01] <deep42thought> it might kill it
[12:07:07] <deep42thought> but it's necessary nevertheless
[12:07:23] <abaumann> I'm just browsing through the mysql config on the buildmaster
[12:07:33] <abaumann> the update went ok. also the sanity-check afterwards
[12:07:39] <deep42thought> \o/
[12:07:51] <abaumann> did you configure much in the mariadb config?
[12:08:13] <deep42thought> not much but some stuff
[12:08:23] <deep42thought> you added some config, too
[12:10:03] <abaumann> yep.
[12:10:06] <abaumann> let's see now..
[12:10:33] <abaumann> hope, I didn't kill replication
[12:10:36] <deep42thought> btw: i added the database "mysql" to the dump
[12:10:45] <deep42thought> in the hope, that we back up the users, etc., too
[12:10:54] <abaumann> good idea.
[12:11:14] <abaumann> that's why I am hesitating to suggest we should migrate to the new buildmaster..
[12:11:20] <abaumann> ..everything runs so nicely now :-)
[12:11:23] <deep42thought> lol
[12:11:34] <abaumann> so, one idea is to break things, so we have a reason to move :->
[12:11:50] <deep42thought> "it's too quiet - let's rip our build system apart and set it up from scratch on a new box"
[12:12:11] <abaumann> exactly :-)
[12:12:21] <deep42thought> or we could just annoy your hoster again with 200% cpu load ;-)
[12:12:25] <abaumann> bad moto, if you want to make a career in administration..
[12:13:38] <abaumann> is the replication slave pushing from the buildmaster to the other instance, or is the other instance pulling data from the buildmaster?
[12:14:04] <deep42thought> pulling
[12:14:13] <abaumann> aha.
[12:19:41] <buildmaster> pentium4/haskell-wl-pprint-terminfo is broken (says eurobuild6-3) - I rescheduled: haskell-wl-pprint-extras.
[13:24:42] <buildmaster> i686/haskell-profunctors are broken (says eurobuild6-5) - I rescheduled: haskell-bifunctors.
[13:33:52] <buildmaster> i686/haskell-reducers are broken (says eurobuild6-5) - I rescheduled: haskell-semigroupoids.
[13:37:31] <buildmaster> i686/haskell-checkers are broken (says eurobuild6-2).
[13:51:02] -!- ofara has joined #archlinux32
[13:52:20] -!- slacka123 has joined #archlinux32
[14:08:19] -!- ofara_ has quit [Quit: ofara_]
[14:08:41] -!- ofara_ has joined #archlinux32
[14:27:22] -!- quequotion has joined #archlinux32
[14:28:52] -!- djmoch has quit [Ping timeout: 246 seconds]
[14:32:41] -!- djmoch has joined #archlinux32
[14:42:02] <abaumann> ninja/samurai: uses cc, which is linked to /usr/bin/cc
[14:42:09] <abaumann> /usr/bin/gcc I mean
[14:42:16] <abaumann> so, how to change that to let's say tcc?
[14:43:02] <abaumann> I can change the cc symlink (nice on a build slave), write my own ninja rule (but then, why at all have ninja embedd a default rule for compiling C)
[14:44:17] <abaumann> *abaumann is completly confused about those new tools supposedily being much better than older tools..
[14:46:16] <abaumann> So, if ninja doesn't like reading environment variables (because they supposedly exist on Unix only - last time I checked, there where tons of enviroment variables on Windows), why can't you pass them as options to ninja.
[14:46:45] <abaumann> I should open a #why-all-modern-build-systems-suck (rant of the day)
[14:48:50] <abaumann> so the solution is to use another tool on top, generating the proper ninja tools you want
[14:53:39] <deep42thought> you seem to be alone with your opinion - #why-all-modern-build-systems-suck is empty
[14:54:39] -!- quequotion_ has joined #archlinux32
[14:56:20] -!- quequotion has quit [Ping timeout: 244 seconds]
[14:56:21] quequotion_ is now known as quequotion
[14:58:04] <elibrokeit> deep42thought: how does shellcheck know what is a command and what is an argument? Trick question: it doesn't, so it cannot detect your broken find
[14:58:39] <deep42thought> yeah, that's true, but it finds so many non-issues, that I thought, it might look into $PATH, too
[14:59:00] <elibrokeit> abaumann: ninja's motto is also "don't use rule expansions because they are slow, let something like meson generate literally every rule"
[14:59:08] <elibrokeit> So, meson respects $CC
[14:59:35] <elibrokeit> deep42thought: whose PATH? The target system could be literally anyrhing
[14:59:42] <elibrokeit> *anything
[15:00:24] <abaumann> PATH is handled by the shell, so _that_ one should be respected.
[15:00:51] <abaumann> sorry elibrokeit, I'm patching samurai now for SAMUFLAGS..
[15:00:57] <deep42thought> :-D
[15:01:11] <abaumann> more working, less ranting :-)
[15:01:32] <abaumann> deep42thought: just tested the new buildmaster with a parallel build of linux.
[15:01:37] <abaumann> looking quite ok.
[15:01:42] <elibrokeit> abaumann: awesome :)
[15:02:04] <abaumann> learning again about the perils of C and strtok
[15:13:03] -!- quequotion_ has joined #archlinux32
[15:14:13] -!- quequotion has quit [Ping timeout: 245 seconds]
[15:14:15] quequotion_ is now known as quequotion
[15:23:33] <buildmaster> i686/haskell-crypto-pubkey-types are broken (says eurobuild6-5) - I rescheduled: haskell-asn1-encoding.
[15:27:33] <buildmaster> i686/haskell-securemem is broken (says eurobuild6-1) - I rescheduled: haskell-byteable.
[15:30:28] <deep42thought> I wonder why all this haskell rescheduling is necessary in the first place
[15:31:46] <abaumann> *abaumann shrugs
[15:32:15] <buildmaster> i686/haskell-cryptohash is broken (says eurobuild6-4) - I rescheduled: haskell-cryptonite.
[15:35:19] -!- MannerMan has quit [Quit: MannerMan]
[15:36:14] -!- MannerMan has joined #archlinux32
[15:39:57] <abaumann> deep42thought: the bike trip was nice, but the weather was a disaster..
[15:40:09] <deep42thought> oh, how so?
[15:40:12] <deep42thought> too hot?
[15:40:16] <abaumann> ..pedalling conformably was only possible before noon.
[15:40:20] <abaumann> yeah.
[15:40:29] <deep42thought> ... or at night
[15:40:58] <abaumann> Lake Constance has some very nice places.
[15:42:02] <abaumann> *abaumann still reads tons of chat logs to see if he has to answer more questions..
[15:43:02] <abaumann> aha, https://bugs.archlinux.org
[15:43:03] <phrik> Title: FS#63015 : [lib32-glibc] ld warning: /usr/lib32/ld-linux.so.2: corrupt GNU_PROPERTY_TYPE (5) size: 0 (at bugs.archlinux.org)
[15:43:13] <abaumann> So, it's not Archlinux32 specific
[15:43:15] <buildmaster> i686/haskell-crypto-cipher-types are broken (says eurobuild6-5).
[15:44:18] <deep42thought> phew :-)
[15:44:37] <elibrokeit> abaumann: strtok is the reason I'm scared to do it myself :D that and the fact that I don't actually *know* C
[15:44:45] -!- MannerMan has quit [Quit: MannerMan]
[15:44:50] <abaumann> hehe
[15:45:07] <abaumann> GNU_PROPERTY_TYPE (5) size: 0
[15:45:13] <abaumann> is some zero-sized marker of sorts
[15:45:23] <abaumann> the binutils code is.. well.. not that documented :-)
[15:45:27] * elibrokeit is much more comfortable in bash and python
[15:47:58] <buildmaster> i686/haskell-http-client is broken (says eurobuild6-2) - I rescheduled: haskell-cookie.
[15:49:46] <buildmaster> i686/haskell-crypto-random is broken (says eurobuild6-3).
[15:50:22] <deep42thought> what exactly does this error mean?:
[15:50:22] <deep42thought> Crypto/Hash/Types.hs:21:1: error:
[15:50:22] <deep42thought> Bad interface file: /usr/lib/ghc-8.6.4/site-local/byteable-0.1.1/Data/Byteable.dyn_hi
[15:50:22] <deep42thought> mismatched interface file versions (wanted "8065", got "8064")
[15:50:22] <deep42thought> |
[15:50:22] <deep42thought> 21 | import Data.Byteable
[15:50:22] <deep42thought> | ^^^^^^^^^^^^^^^^^^^^
[15:50:36] <deep42thought> _which_ package needs recompilation?
[15:51:23] <abaumann> haskell-byteable?
[15:51:41] <deep42thought> that's what I currently reschedule in that case
[15:51:55] <abaumann> mmh. maybe that one fails for different reasons
[15:51:58] <deep42thought> but the sheer amount of needed reschedules makes me wonder whether that is correct
[15:56:37] <abaumann> what I was wondering: if the Haskell train derails one, the only safe options is to go back with all modules to the previous version and then build everything again
[15:56:44] <abaumann> sort of bootstrapping from t
[15:56:51] <abaumann> the last known good version
[15:57:02] <deep42thought> haskell is un-bootstrappable
[15:57:05] <abaumann> but, it's not easy to back (especiialy in time)
[15:57:11] <abaumann> *go back
[15:58:17] <deep42thought> one needs to be able to compile all haskell-* modules without any haskell-* module present at the beginning
[15:58:44] <deep42thought> as far as I understand the problem
[15:58:48] <abaumann> do we have haskell on i486?
[15:58:54] <buildmaster> i686/haskell-cookie is broken (says eurobuild6-2) - I rescheduled: haskell-data-default-class.
[15:59:02] <abaumann> if yes, it was auto-bootstrapped :-)
[15:59:10] <deep42thought> ls: cannot access '/mnt/archlinux32/pool/ghc-*-i486.pkg.tar.xz': No such file or directory
[15:59:19] <abaumann> explains something
[15:59:40] <abaumann> *abaumann puts another language/environemnt onto the list of to-boostrapping-things
[15:59:46] <deep42thought> :-D
[15:59:53] <deep42thought> maybe, ghc depends on rust?
[16:00:13] <deep42thought> ==> Installing missing dependencies...
[16:00:14] <deep42thought> error: target not found: ghc-static
[16:00:14] <deep42thought> error: target not found: haskell-hscolour
[16:00:29] <deep42thought> errrm, ghc depends on haskell-hscolour????
[16:00:34] <buildmaster> i686/haskell-x509-validation is broken (says nlopc43).
[16:00:53] <abaumann> yeah, very important to get those color outputs on the screen while bootstrapping a compiler :->
[16:01:24] <deep42thought> that - and it is absolutely necessary to use it in such a low-level language as haskell
[16:02:06] <abaumann> :-)
[16:03:28] <abaumann> yeah. it may depend on a static version of itself, that's quite normal.
[16:03:57] <deep42thought> we should add the following peace of logic to all (re)scheduling stuff: if any haskell-* package enters the build-list, then all runtime dependencies should be rescheduled as well
[16:03:58] <abaumann> https://www.joachim-breitner.de
[16:03:59] <phrik> Title: Thoughts on bootstrapping GHC – Blog – Joachim Breitner's Homepage (at www.joachim-breitner.de)
[16:06:19] <abaumann> It's also a fact that the more features a language has, the harder bootstrapping gets: because you need more and more subsets of the same language to build upon another
[16:10:53] <abaumann> ghc should be standalone (minus featuritis packages)..
[16:11:24] <abaumann> I fear to propose a "Haskell package version reset"
[16:11:42] <abaumann> Can you force an older version of ghc in build-support?
[16:12:00] <abaumann> naeh. everything too complicated..
[16:13:25] <deep42thought> what do you think about https://mirror.archlinux32.org ?
[16:13:26] <phrik> Title: #archlinux32 | Logs for 2019-06-28 (at mirror.archlinux32.org)
[16:15:59] <elibrokeit> abaumann: oof, ghc is sadness. I bet deep42thought will remember how I helped him fix some bootstrapping issues, because the stupid things won't build due to cycles, and the solution is to rebuild without check()... But doing that will change the soname.
[16:16:19] <deep42thought> yeah, that's madness
[16:16:24] <deep42thought> and yes: I remember
[16:16:27] <abaumann> lol
[16:16:49] * deep42thought proposes dropping *all* haskell check()s
[16:17:12] <abaumann> and then the filenames will be different than the 64-bit ones..
[16:17:19] <abaumann> ..and somebody will depend on that..
[16:17:23] <deep42thought> they are different anyways
[16:17:31] <deep42thought> there is a hash in the file name!
[16:17:46] <deep42thought> I strongly doubt, that this will be architecture agnostic
[16:18:17] <abaumann> give new kids old tools and have them have new brilliant ideas..
[16:19:04] <abaumann> How is Haskell rebult on 64-bit?
[16:19:25] <deep42thought> oh, look at the time! I have to go ...
[16:19:33] <abaumann> ok.
[16:19:37] -!- deep42thought has quit [Quit: Leaving.]
[16:19:38] <abaumann> cu
[16:22:21] -!- abaumann has quit [Quit: leaving]
[16:28:21] -!- quequotion has quit [Remote host closed the connection]
[16:35:56] <elibrokeit> abaumann: felixonmars has a custom script that rebuilds the entire Haskell world in topological order based on whether any of its depends, makedepends, or checkdepends have been updated by even a pkgrel.
[16:37:17] <elibrokeit> It requires manual work to resolve build cycles, by modifying the PKGBUILD, running with --nocheck, building lots of things, and then doing a second rebuild pass after reverting the changes, bumping the pkgrel again, and running with check enabled.
[16:43:30] <buildmaster> i686/cbindgen is broken (says eurobuild6-2).
[16:54:26] <djmoch> elibrokeit: i've always wondered about that. is haskell that fragile that everything needs to get rebuilt if the pkgrel changes?
[16:55:15] <djmoch> or is the script just really eager?
[16:56:19] <buildmaster> pentium4/npm is broken (says eurobuild6-2).
[16:59:27] -!- finsternis has quit [Excess Flood]
[17:00:58] -!- finsternis has joined #archlinux32
[17:24:35] <buildmaster> i686/dart is broken (says eurobuild6-3).
[17:28:39] <buildmaster> pentium4/dart is broken (says nlopc43).
[17:31:25] <elibrokeit> djmoch: Haskell has no stable abi
[17:31:43] <elibrokeit> Every time that script recompiled a Haskell package, the siloname changes
[17:31:46] <elibrokeit> *soname
[17:32:20] <elibrokeit> Haskell shared libraries are always libfoo-HASH.so
[17:33:20] -!- thePiGrepper has joined #archlinux32
[17:33:33] <elibrokeit> The hash depends on the version of all dependencies with infinite recursion, the command line used to build it, and I don't even know what else
[17:48:22] <djmoch> that's ... not great
[17:50:09] -!- deep42thought has joined #archlinux32
[17:50:09] <buildmaster> Hi deep42thought!
[17:50:09] <buildmaster> !rq deep42thought
[17:50:11] <phrik> buildmaster: <deep42thought> I like ignoring systemd, but: why?
[17:50:34] <deep42thought> elibrokeit: as I understand it, you cannot possibly build all haskell-* packages with check() enabled in the end
[17:51:37] -!- AndrevS has quit [Remote host closed the connection]
[17:51:58] <deep42thought> or at least not in a way, that the check uses the package what will end up in the repository in the end
[17:52:06] <elibrokeit> deep42thought: you enable them one by one, relying on the fact that most/all build cycles are ultimately not needed at runtime
[17:52:50] <deep42thought> I understand that part
[17:53:12] <deep42thought> but consider packages a and b, where a run-depends on b an b check-depends on a
[17:53:51] <elibrokeit> Compile b without check, compile a. A is linked to b, but b is not linked to a.
[17:53:53] <deep42thought> so you build b w/o check, then a, then b with check (against old a), then a again - but then you don't know anything about cooperation of b and a anymore
[17:54:08] <elibrokeit> Then compile b with check, but you don't need to recompile a
[17:54:16] <deep42thought> why?
[17:54:24] <deep42thought> if I recompile b, I need to recompile a
[17:54:42] <elibrokeit> Because it works as is, it's only a checkdepends
[17:55:11] <deep42thought> but recompiling b (wheter with or without check does not matter) makes recompiling a necessary
[17:55:16] <elibrokeit> If you do recompile it anyway, just to check that the tests work, then all is fine and well for a, although b does need rebuilding
[17:55:38] <elibrokeit> Checkdepends are not rundepends...
[17:55:49] <elibrokeit> It's not a true build cycle
[17:56:16] <elibrokeit> If it were a true build cycle then the soname changes would be impossible to reconcile ;)
[17:57:36] <deep42thought> ah, ok, so the checked b never enters the repos?
[18:01:54] <elibrokeit> sure it does. It just does not invalidate a
[18:01:58] <elibrokeit> a is not linked to b
[18:02:16] <elibrokeit> b is linked to a, but a's testsuite is the only thing that links to b
[18:02:43] <elibrokeit> the fact that the testsuite is used as part of the hash input to determine the final soname of a, does not mean that a itself links to b
[18:02:59] <elibrokeit> it just means that a is non-deterministically named depending on whether b was present in the build environment...
[18:03:48] <deep42thought> you swapped the meaning of a and b
[18:05:17] -!- deep42thought has parted #archlinux32
[18:05:21] -!- deep42thought has joined #archlinux32
[18:05:22] <buildmaster> Hi deep42thought!
[18:05:22] <buildmaster> !rq deep42thought
[18:05:23] <phrik> buildmaster: <deep42thought> otoh I have the feeling, archlinux32 dies anyway if I get overrun by a bus
[18:06:56] -!- eschwartz has joined #archlinux32
[18:07:03] <deep42thought> "it just means that a is non-deterministically named depending on whether b was present in the build environment..." - this means, that compiling a needs recompilation of b which "invalidates the outcome of the check of a
[18:07:19] <deep42thought> sry, gotta go
[18:07:21] -!- deep42thought has quit [Quit: Leaving.]
[18:12:57] -!- thePiGrepper has quit [Ping timeout: 245 seconds]
[18:19:46] <elibrokeit> deep42thought: no, it does not need a recompilation... it is simply that a recompilation will change the resulting generated code, and therefore require relinking... because if your generated code changes, then you need to recompile the haskell runtime tree.
[18:20:13] -!- eschwartz has quit [Read error: Connection reset by peer]
[18:20:26] <elibrokeit> You do not need to recompile the makedepends or checkdepends tree though!
[18:21:08] -!- eschwartz has joined #archlinux32
[18:21:22] -!- thePiGrepper has joined #archlinux32
[18:26:15] -!- MrBIOS has quit [Quit: MrBIOS]
[18:33:22] -!- isacdaavid has joined #archlinux32
[18:44:34] <elibrokeit> deep42thought: llpp has a makedepends on ocaml-lablgl, which requires libgl.
[18:45:18] <elibrokeit> But it's a runtime dependency...
[18:46:17] <elibrokeit> and llpp depends on glu directly. glu depends on libgl.
[18:47:12] -!- thePiGrepper has quit [Ping timeout: 272 seconds]
[18:47:19] <elibrokeit> Why do tons of things on the archlinux32 website say "not found in pkg-api"
[19:02:53] -!- eschwartz has quit [Read error: Connection reset by peer]
[19:04:06] -!- MrBIOS has joined #archlinux32
[19:04:19] -!- eschwartz has joined #archlinux32
[19:08:37] -!- MrBIOS_ has joined #archlinux32
[19:08:44] -!- MrBIOS has quit [Ping timeout: 272 seconds]
[19:08:44] MrBIOS_ is now known as MrBIOS
[19:24:46] -!- oaken-source has joined #archlinux32
[19:24:46] -!- oaken-source has quit [Client Quit]
[19:25:11] -!- oaken-source has joined #archlinux32
[19:43:03] -!- thePiGrepper has joined #archlinux32
[19:49:05] -!- thePiGrepper has quit [Read error: Connection reset by peer]
[19:49:43] -!- eschwartz has quit [Read error: Connection reset by peer]
[19:50:31] -!- eschwartz has joined #archlinux32
[19:57:55] -!- yans has joined #archlinux32
[20:09:32] -!- thePiGrepper has joined #archlinux32
[20:10:45] -!- samantaz has joined #archlinux32
[20:12:25] -!- eschwartz has quit [Read error: Connection reset by peer]
[20:13:42] -!- eschwartz has joined #archlinux32
[20:37:46] -!- eschwartz has quit [Read error: Connection reset by peer]
[20:38:54] -!- eschwartz has joined #archlinux32
[21:02:09] -!- eschwartz has quit [Read error: Connection reset by peer]
[21:03:06] -!- eschwartz has joined #archlinux32
[21:28:26] -!- eschwartz has quit [Read error: Connection reset by peer]
[21:29:18] -!- eschwartz has joined #archlinux32
[21:40:50] -!- thePiGrepper has quit [Ping timeout: 258 seconds]
[21:43:40] <buildmaster> i686/haskell-foldl is broken (says eurobuild6-5) - I rescheduled: haskell-mwc-random.
[21:51:38] -!- thePiGrepper has joined #archlinux32
[21:52:16] -!- eschwartz has quit [Read error: Connection reset by peer]
[21:53:30] -!- eschwartz has joined #archlinux32
[21:55:28] -!- eschwartz has quit [Read error: Connection reset by peer]
[21:55:52] -!- samantaz has quit [Ping timeout: 244 seconds]
[21:57:48] -!- samantaz has joined #archlinux32
[21:57:53] -!- eschwartz has joined #archlinux32
[22:02:54] -!- thePiGrepper has quit [Ping timeout: 272 seconds]
[22:18:12] -!- eschwartz has quit [Read error: Connection reset by peer]
[22:18:36] <buildmaster> i686/haskell-cryptonite-conduit is broken (says eurobuild6-1) - I rescheduled: haskell-conduit-extra.
[22:19:05] <buildmaster> i686/hedgewars are broken (says nlopc43).
[22:19:05] -!- eschwartz has joined #archlinux32
[22:20:15] <buildmaster> i686/haskell-cryptohash-conduit is broken (says eurobuild6-2).
[22:20:23] -!- thePiGrepper has joined #archlinux32
[22:28:45] -!- deep42thought has joined #archlinux32
[22:28:45] <buildmaster> Hi deep42thought!
[22:28:45] <buildmaster> !rq deep42thought
[22:28:46] <phrik> buildmaster: <deep42thought> rcf you just detuned and are now receiving a different program - please execute /join #debian
[22:29:06] <deep42thought> elibrokeit: "not found in pkgapi" is the price we have to pay for updating the pkgapi asynchronously
[22:32:20] <deep42thought> or is it a persisting problem?
[22:33:52] <elibrokeit> the error is gone
[22:44:41] -!- isacdaavid has quit [Quit: Leaving.]
[23:03:04] -!- eschwartz has quit [Read error: Connection reset by peer]
[23:04:18] -!- eschwartz has joined #archlinux32
[23:22:21] -!- deep42thought has quit [Quit: Leaving.]
[23:54:19] -!- eschwartz has quit [Read error: Connection reset by peer]
[23:55:32] -!- eschwartz has joined #archlinux32