#archlinux32 | Logs for 2018-07-08
Back
[00:26:25] -!- deep42thought has quit [Quit: Leaving.]
[00:31:48] -!- isacdaavid has joined #archlinux32
[01:21:58] -!- oaken-source has quit [Ping timeout: 264 seconds]
[01:46:54] -!- isacdaavid has quit [Read error: Connection reset by peer]
[02:21:47] -!- Cthulu201 has quit [Quit: Nowhere special. I always wanted to go there.]
[02:23:12] -!- Cthulu201 has joined #archlinux32
[07:15:04] -!- isacdaavid has joined #archlinux32
[08:23:46] -!- isacdaavid has quit [Quit: Leaving.]
[09:08:17] -!- deep42thought has joined #archlinux32
[09:08:18] <buildmaster> Hi deep42thought!
[09:11:17] <deep42thought> abaumann: do we want to comment on this: https://lists.archlinux.org ?
[09:11:17] <phrik> Title:[pacman-dev] interest in using meson over autotools? (at lists.archlinux.org)
[09:11:32] <deep42thought> it would be bad if pacman cannot be built, because of broken meson :-D
[09:32:04] -!- oaken-source has joined #archlinux32
[10:06:35] -!- oaken-source has quit [Ping timeout: 240 seconds]
[13:47:19] <buildmaster> dovecot is broken (says eurobuild3).
[15:39:52] -!- ahuillet has joined #archlinux32
[15:40:27] <ahuillet> hi
[15:40:33] <ahuillet> https://archlinux32.mirror.roelf.org
[15:40:33] <ahuillet> <<- this mirror in the mirror list appears out of date
[15:40:42] <ahuillet> am I mistaken? I think it should be taken off the list if it can't be updated
[15:40:57] <ahuillet> which leads me to my second question: do you guys need a mirror for archlinux32?
[15:47:21] <deep42thought> ahuillet: where would that mirror be?
[15:47:53] <ahuillet> in the mirrorlist?
[15:48:23] <deep42thought> no, geographically :-)
[15:48:27] <ahuillet> France :)
[15:48:52] <deep42thought> sounds good
[15:49:02] <deep42thought> thanks for the notice and the offer :-)
[15:49:39] <ahuillet> well the offer depends on how much storage and bandwidth that requires. do you know?
[15:50:24] <deep42thought> storage: 40GB (but might be more during big rebuilds - up to 80GB is realistic)
[15:50:39] <deep42thought> regarding bandwidth, I cannot really say something
[15:51:16] <deep42thought> well
[15:51:25] <deep42thought> regarding _traffic_ I cannot say something
[15:51:38] <deep42thought> regarding _bandwidth_: it should be >~ 100MBit/s
[15:54:12] <deep42thought> Is this within your expectations?
[15:54:38] <ahuillet> yes I believe that could work. I have to double check with someone else but that should be fine
[15:54:47] <ahuillet> we'll see what the average traffic ends up being.
[15:55:06] <ahuillet> do you have documentation somewhere of how to set this up?
[15:56:00] <deep42thought> hmm, I thought, we had ... let me search
[15:57:25] <deep42thought> looks, like we do not have
[15:58:34] <deep42thought> you can use the official guide from archlinux, but sync off mirror.archlinux32.org instead
[15:58:38] <deep42thought> (via rsync)
[16:08:35] <deep42thought> ahuillet: the avg. traffic on the master mirror is ~15MBit/s (but that hosts a lot more than the archlinux32 mirror) and on some other mirror it's one to three TB/month - but that also hosts some more stuff
[16:11:19] <ahuillet> that sounds workable
[16:56:32] -!- Polichronucci_ has joined #archlinux32
[16:57:57] -!- Polichronucci has quit [Ping timeout: 240 seconds]
[16:57:58] Polichronucci_ is now known as Polichronucci
[17:24:13] <ahuillet> deep42thought : rsync://mirror.archlinux32.org ?
[17:24:42] <deep42thought> rsync://mirror.archlinux32.org
[17:26:54] <ahuillet> thx
[17:35:36] -!- isacdaavid has joined #archlinux32
[17:36:28] -!- abaumann has joined #archlinux32
[17:36:29] <buildmaster> Hi abaumann!
[17:36:57] <abaumann> deep42thought: pacman and meson. Well he can definitely change it if he shows me how to bootstrap archlinux then.. :->
[17:40:05] <abaumann> should I answer.
[17:40:06] <abaumann> ?
[17:40:32] <abaumann> oh? I'm not on the pacman-dev list. :-)
[17:52:17] <deep42thought> abaumann: do so!
[17:54:18] <abaumann> mmh. how do you reply to an archived mail? setting In-Reply-To (if I can find it)..
[17:55:05] <deep42thought> I forwarded on of the mails to you, hope that helps
[17:55:16] <abaumann> ah. that's good. thank :-)
[17:56:17] <abaumann> *thanks
[18:12:26] <abaumann> ok. done.
[18:12:53] <abaumann> btw: I'm working on the i486 branch of 'builder' and 'packages' and I'm backporting the patches needed to build from 'bootstrap32'. So far no problems. :-)
[18:14:36] <deep42thought> I saw, that you simply made a new i486 branch on packages - is this maintainable?
[18:15:04] <deep42thought> because this is not what I had in mind, but it might make thinks simpler
[18:16:29] <abaumann> so, what's in your mind then?
[18:16:46] <abaumann> It was just the first thing I came up with and it was working.
[18:16:52] <deep42thought> I thought of a modification repository similar to what we already have
[18:17:04] <deep42thought> e.g.: x86_64 + packages32 -> i686
[18:17:13] <abaumann> ah. and then patch the builder accordingly..
[18:17:13] <deep42thought> and i686 + packages32-i486 -> i486
[18:17:21] <abaumann> ok. yeah. why not.
[18:17:29] <abaumann> the thing I did was with voidLinux in my mind. :-)
[18:17:39] <deep42thought> I don't know that
[18:17:59] <abaumann> one central PKBUILD, if CARCH for all architecture dependend things.
[18:18:06] <abaumann> ok, not called PKGBUILD there. :-)
[18:18:26] <deep42thought> yeah, totally valid
[18:18:44] <abaumann> basically: I take the things in 'if CARCH = i486' from the branch into a new repo, adapt patching in builder and voila.
[18:18:48] <abaumann> no problem. :-)
[18:18:50] <deep42thought> we could even drop the i486 branch at some point (when it's "stable") and just have $CARCH switches like upstream used to have
[18:19:04] <deep42thought> I prefer your way
[18:19:17] <abaumann> ah. for sure. we would drop the branch.
[18:19:20] <deep42thought> only downside I see: the builder will reschedule _all_ architectures if you only change one of them
[18:19:35] <deep42thought> but this should not happen too often, I guess
[18:19:41] <abaumann> yeah. that's a problem.
[18:19:46] <deep42thought> plus: I need to rethink blacklisting
[18:20:16] <abaumann> * abaumann forget his pasta with salsa and has to do some cooking things.. FAST.. AFK.. BRB
[18:20:25] <deep42thought> cu
[18:22:01] <abaumann> back
[18:22:08] <deep42thought> that was fast O.o
[18:22:14] <abaumann> me fast chef :-)
[18:22:19] <deep42thought> what did you cook: minute rice?
[18:22:30] <abaumann> no. spaghetti con sugo
[18:22:41] <abaumann> al pomodoro..
[18:22:42] <abaumann> :-)
[18:23:00] <abaumann> blacklisting? yeah. the un-blacklisting mainly is an issue currently.
[18:23:52] <deep42thought> that, too
[18:24:07] <deep42thought> but I think, you mean something else by "unblacklisting" than me :-)
[18:24:47] <abaumann> whitelisting? ;-)
[18:27:29] <deep42thought> no, my issue is, that if b is not compiled, because a is blacklisted and b depends on a: If then a is unblacklisted, then b will not be scheduled, currently
[18:28:28] <abaumann> oh. that's also not good.
[18:32:20] <elibrokeit> abaumann: how do you currently bootstrap autotools?
[18:33:10] <elibrokeit> what new challenges does python bring in, except for the larger base?
[18:33:46] <elibrokeit> also mailman provides reply links to set message-id and persist threads for people who didn't get the email :p
[18:38:50] <abaumann> elibrokeit: I was in a cooking hurry, now I am in the list.. ;-)
[18:39:16] <abaumann> you are right, python is a little bit more depending on other packages in order to bootstrap.
[18:41:30] <abaumann> actually, the dependencies of python are not so bad, the modules are (especially bootstrapping the basic python modules) and the build system of python..
[18:45:31] <elibrokeit> does nothing else needed to bootstrap the system, rely on this? I mean, you need to build python at some point...
[18:46:47] <abaumann> IIRC it was my only motivation to build python...
[18:47:00] <abaumann> .. on the other hand, also gdb needs python in it's final build.
[18:47:02] <elibrokeit> meson only needs python, ninja, and, for building, setuptools
[18:47:15] <abaumann> I like the *only* :-)
[18:48:43] <abaumann> In order to get a functional build slave I need all of core (or better what core should be nowadays) and meson is definitely part of it.
[18:48:55] <elibrokeit> well, you suggested "especially bootstrapping the basic python modules"
[18:48:56] <abaumann> otherwise building X-stuff for instance is not possible.
[18:49:16] <elibrokeit> systemd needs meson anyway
[18:49:35] <abaumann> yeah.. so.. I lost :-)
[18:49:35] <deep42thought> btw: I recently removed our meson patches - so we build identical to archlinux now
[18:49:48] <abaumann> ah. that's good.
[18:49:57] <elibrokeit> which may not be needed for booting a chroot, but libsystemd is needed anyway...
[18:49:57] <deep42thought> we have a working dmd build now :-)
[18:50:26] <abaumann> elibrokeit: yeah, systemd is the killer argument. If we need it there, pacman is no longer a real issue.
[18:50:34] <abaumann> *jippie*
[18:50:42] <abaumann> dmd works, and without us doing anything :-)
[18:50:45] <elibrokeit> do you follow https://bootstrappable.org BTW
[18:50:46] <phrik> Title:Bootstrappable builds (at bootstrappable.org)
[18:51:43] <abaumann> no. interesting page.
[18:52:22] <elibrokeit> also meson is both a spec and an implementation, they've specifically left open the possibility that people should be able to reimplement meson using other programming languages
[18:53:00] <abaumann> ah. I thought this is common sense when coding?
[18:53:32] <elibrokeit> Maybe bootstrappable.org would be interested in sponsoring a minimal meson implementation for bootstrapping purposes?
[18:53:47] <elibrokeit> Tell them that it's needed to bootstrap systemd-based systems!
[18:54:10] <deep42thought> c-meson :-D
[18:54:39] <abaumann> I thought it was called D-meson? ;-)
[18:54:53] <deep42thought> well, there are many mesons
[18:55:05] <abaumann> physicists.. *tsts* ;-)
[18:55:17] <deep42thought> particle zoo, section mesons
[18:56:04] <ahuillet> Server = https://archlinux32.agoctrl.org
[18:56:07] <ahuillet> give it a shot?
[18:56:26] <ahuillet> cron rule syncs it every hour
[18:57:09] <elibrokeit> abaumann: you going to ask them?
[18:57:29] <elibrokeit> or interested in writing your own build system to aid in bootstrapping?
[18:57:55] <deep42thought> ahuillet: looks good
[18:58:07] <deep42thought> you skip the isos and the irc-logs on purpose, I guess
[18:58:51] <abaumann> https://github.com
[18:58:53] <phrik> Title:Move to native implementation · Issue #1990 · mesonbuild/meson · GitHub (at github.com)
[18:59:13] <ahuillet> indeed. no point mirroring the irclogs
[18:59:42] <abaumann> elibrokeit: my problem is, I'm not very fond of meson.
[18:59:56] <abaumann> It's just another try to replace make, there where to many of them in the past.
[19:00:42] <elibrokeit> abaumann: it's not replacing make, it's replacing autotools :p
[19:00:57] <elibrokeit> replacing make itself, with ninja, is an implementation detail
[19:01:09] <abaumann> yeah. for most software I think a plain Makefile does the job. Just a personal opinion.
[19:01:22] <abaumann> most things autotools do are obsolete nowadays..
[19:01:50] <abaumann> and I have seen too many projects doing a test for let's say (uint64_t) and then happilly using long longs in the code.
[19:03:18] <elibrokeit> ehm....
[19:03:20] <elibrokeit> 2018-01-18 06:56:02 PM amcrae make sans autotools is a really bad idea...
[19:03:20] <elibrokeit> 2018-01-18 06:56:20 PM amcrae for any project compiling more than one file and having actual configure options
[19:04:16] <elibrokeit> autotools does a terrible job of providing options, but the solution is not to remove the actual features it *does* offer
[19:04:39] <elibrokeit> the solution is to create some alternative which *does* provide configuration options
[19:04:46] <elibrokeit> like, meson
[19:05:05] <elibrokeit> also note that pacman needs to work on macOS and Windows
[19:05:15] <abaumann> cmake then?
[19:05:37] <abaumann> ok, it does quite some things even more terrible than autotools :-)
[19:05:45] <elibrokeit> CMake is like meson except nearly as ugly as autotools...
[19:05:55] <elibrokeit> so why not just use meson :p
[19:07:22] <abaumann> can meson cross-compile?
[19:07:35] <elibrokeit> FWIW replacing Make with ninja is not really an interesting topic, because ninja is basically "what would Make be like if we removed support for pattern rules and just supported explicit target: file ==> command generator"
[19:08:46] <abaumann> meson needs python at runtime, autotools need m4, perl at develop-maintain-time.
[19:08:46] <elibrokeit> https://mesonbuild.com
[19:08:46] <phrik> Title:Cross compilation (at mesonbuild.com)
[19:09:04] <elibrokeit> Well, meson's generated ninja files are in theory standalone, they might hook into meson to regenerate if your build.meson gets modified but otherwise they just run commands blindly using meson
[19:09:08] <ahuillet> deep42thought : what's the best frequency of rsync for the mirror? (I use the reposync.sh script)
[19:09:13] <elibrokeit> err, using ninja
[19:09:43] <deep42thought> ahuillet: 1h is fine
[19:09:59] <deep42thought> if you increase it, please check lastsync before actually invoking rsync
[19:10:11] <abaumann> so. no issue then: meson and pyton on the host to cross-compile a pacman, then rebuild pacman when you have python and meson inside the target architecture.
[19:11:19] <abaumann> you could write pacman in python (alsa called yum). ;-)
[19:11:55] <elibrokeit> abaumann: sure :)
[19:33:58] -!- Polichronucci has quit [Ping timeout: 264 seconds]
[19:44:40] -!- Polichronucci has joined #archlinux32
[19:52:03] -!- abaumann has quit [Quit: leaving]
[21:31:54] -!- eduardoeae_ has quit [Quit: eduardoeae_]
[21:32:25] -!- Polichronucci has quit [Ping timeout: 244 seconds]
[21:38:51] -!- Polichronucci has joined #archlinux32
[21:53:36] -!- Polichronucci has quit [Ping timeout: 244 seconds]
[22:13:16] -!- Polichronucci has joined #archlinux32
[22:15:20] -!- Polichronucci_ has joined #archlinux32
[22:17:57] -!- Polichronucci has quit [Ping timeout: 265 seconds]
[22:17:57] Polichronucci_ is now known as Polichronucci
[23:02:24] -!- Polichronucci_ has joined #archlinux32
[23:04:23] -!- Polichronucci has quit [Ping timeout: 268 seconds]
[23:04:24] Polichronucci_ is now known as Polichronucci
[23:45:42] -!- Polichronucci has quit [Ping timeout: 268 seconds]
[23:46:10] -!- Polichronucci has joined #archlinux32
[23:51:12] -!- deep42thought has quit [Quit: Leaving.]