#archlinux-ports | Logs for 2017-10-27

[00:07:27] -!- guys has joined #archlinux-ports
[04:41:37] -!- tyzoid has quit [Ping timeout: 248 seconds]
[04:43:19] -!- tyzoid has joined #archlinux-ports
[05:20:13] -!- isacdaavid has joined #archlinux-ports
[05:32:20] -!- p71 has quit [Read error: Connection reset by peer]
[05:37:56] -!- p71 has joined #archlinux-ports
[06:43:43] -!- deep42thought has joined #archlinux-ports
[06:58:25] -!- deep42thought has quit [Quit: Leaving.]
[08:35:40] -!- deep42thought has joined #archlinux-ports
[08:58:30] guys is now known as guido_van_chuckl
[09:00:27] -!- isacdaavid has quit [Quit: isacdaavid]
[09:10:08] -!- shentey has joined #archlinux-ports
[09:11:29] -!- shentey has quit [Client Quit]
[09:11:31] -!- shentey_ has joined #archlinux-ports
[09:26:34] -!- shentey_ has quit [Read error: Connection reset by peer]
[09:27:54] -!- shentey_ has joined #archlinux-ports
[09:30:26] -!- shentey_ has quit [Client Quit]
[09:35:57] -!- guido_van_chuckl has quit [Ping timeout: 240 seconds]
[09:38:02] -!- guys has joined #archlinux-ports
[09:38:23] guys is now known as guido_van_chuckl
[10:59:19] bi is now known as bq
[11:11:22] -!- shentey has joined #archlinux-ports
[12:49:52] -!- shentey has quit [Quit: Konversation terminated!]
[14:27:40] -!- tyzoid_work has joined #archlinux-ports
[14:29:10] <tyzoid_work> hey deep42thought, would clisp work for building SBCL? It's also available as an arch package, so we could just add it as a build-dep
[14:30:16] <deep42thought> the problem might be of a different kind
[14:30:24] <tyzoid_work> oh?
[14:30:49] <deep42thought> I _guess_, that the compiler itself is not the real issue
[14:30:59] <deep42thought> because the binary provided from sbcl does not work, too
[14:32:04] <deep42thought> according to their site, clisp should work, though
[14:32:12] <deep42thought> so we could give that a shot
[14:32:21] <tyzoid_work> idk how much work it is, but if clisp works, then we have more info to give upstream.
[14:34:29] <deep42thought> I try to just replace the dependency and the compiler for building in the PKGBUILD - let's see, what happens :-)
[14:36:25] <deep42thought> x86_64 fails with the same error with clisp
[14:38:56] <deep42thought> ok, i686 now shows the same error as x86_64 - and not the sbcl debugger
[14:39:04] <deep42thought> I'm not sure if this is an improvement :-/
[14:39:18] <deep42thought> ok, I think it is - the error is emitted later :-)
[14:45:28] <tyzoid_work> deep42thought: Are we grabbing the right code?
[14:45:43] <tyzoid_work> i.e. are we grabbing from master/do they have a release branch?
[14:46:34] <deep42thought> this is the source release tarball
[14:48:35] <deep42thought> It might be, that some gcc-update broke the build https://gcc.gnu.org
[14:48:46] <deep42thought> but this message is using an older gcc than we have
[14:48:46] <deep42thought> ..
[14:55:43] <tyzoid_work> deep42thought: https://gcc.gnu.org
[14:55:56] <tyzoid_work> See the bug marked duplicate, it looks like the fix made it into 7.2
[14:56:31] <tyzoid_work> mainline is on 7.2, are we?
[14:56:40] <deep42thought> yep
[14:56:48] <tyzoid_work> hmm
[14:57:16] <tyzoid_work> So do you think it's a bug in the source, then?
[14:57:41] <tyzoid_work> also: is there a lisp compiler that isn't written in lisp we could test with?
[14:57:48] <deep42thought> it is some incopatibility of the source with some library
[14:58:55] <tyzoid_work> aka one of our .so's is too far up to date?
[14:59:17] <deep42thought> both ours and x86_64-archlinux's
[14:59:37] <tyzoid_work> \ah
[15:02:31] <tyzoid_work> deep42thought: btw, do you find the BM Sanity emails helpful?
[15:03:05] <deep42thought> not really
[15:03:18] <tyzoid_work> I can disable those, so they stop cluttering your inbox
[15:03:19] <deep42thought> but I need to make the web-output less sensitive
[15:03:47] <deep42thought> e.g. everything uses locks, but the sanity-test does not, because it once was run at the start of each script
[15:03:55] <deep42thought> so it can see some intermediate state
[15:03:58] <tyzoid_work> ah
[15:04:22] <tyzoid_work> mkdir -p /tmp/bm_sanity_check_lck || exit ?
[15:04:26] <deep42thought> my approach would be to only emit a warning if it fails $n consecutive times (e.g. n=3)
[15:04:49] <deep42thought> hmm
[15:04:57] <deep42thought> a separate sanity-check lock
[15:05:07] <tyzoid_work> empty directory creation/deletion are atomic
[15:05:20] <deep42thought> there is already flock
[15:05:27] <deep42thought> (in heavy use by the other scripts)
[15:05:29] <tyzoid_work> does that work in bash?
[15:05:43] <deep42thought> but the problem is, I don't want the sanity-check to lock all other processes
[15:05:46] <deep42thought> yes
[15:06:45] <tyzoid_work> deep42thought: Why not have the other processes put a sanity-check request into a FIFO
[15:07:00] <tyzoid_work> deep42thought: Then you can have a background process chug through them when it's free
[15:07:07] <deep42thought> the sanity check creates a file if it detects insanity
[15:07:13] <deep42thought> and the other scripts check for that file
[15:07:32] <tyzoid_work> so I don't really see any other way then blocking other builds while sanity check runs
[15:07:37] <deep42thought> but the problem is that the other scripts may create a (temporarily) insane build master
[15:07:43] <deep42thought> yep
[15:07:48] <deep42thought> unfortunately
[15:08:27] <deep42thought> or at least, the sanity check should ignore parts of the build-master if some other script is currently running
[15:08:28] <tyzoid_work> If you make a separate lock for sanity check, you could make it a shared lock amongst the build threads
[15:08:36] <tyzoid_work> then take it over exclusively to run the sanity check
[15:08:45] <tyzoid_work> that way, builds are still parallel, and the sanity check runs consistant
[15:09:20] <deep42thought> I might just skip sanity checks if other build processes are running
[15:09:28] <deep42thought> actually, why not :-)
[15:10:05] <tyzoid_work> deep42thought: Quick question, if I have a process waiting for an exclusive lock, and there are existing shared locks, the exclusive lock will wait/block
[15:10:16] <tyzoid_work> but can other shared locks be placed while the exclusive lock is waiting?
[15:10:33] <tyzoid_work> or will the existance of the exclusive request block new shared locks?
[15:11:09] <deep42thought> the shell's flock is not that complex, I think - it doesn't know the concept of "a shared lock"
[15:11:20] <deep42thought> but I might be missing something there
[15:11:58] <deep42thought> ah, no, I'm wrong: apparently it knows '--shared'
[15:12:40] <tyzoid_work> was just about to post that
[15:12:57] <tyzoid_work> so I guess we need to determine which of those two approaches flock takes
[15:13:18] <tyzoid_work> because if future shared locks get blocked, then we should be golden
[15:13:25] <tyzoid_work> we run shared, then request exclusive
[15:13:36] <tyzoid_work> future shared are blocked until exclusive gets the lock/releases the lock
[15:14:12] <deep42thought> it allows another shared lock
[15:15:03] <deep42thought> na, this is fine, too - the sanity-check is less crucial than recent/parallel updates of everything
[15:16:12] <tyzoid_work> deep42thought: We could introduce that functionality by adding an exclusive lock on a request object
[15:16:22] <tyzoid_work> but then we'd need to be careful about deadlocks
[15:16:36] <deep42thought> what do you mean by "request object"?
[15:18:47] <tyzoid_work> Just a file that flags that an exclusive lock request is in place
[15:19:23] <deep42thought> ah, ok
[15:19:24] <tyzoid_work> basically a lock to place a shared lock on the other object
[15:19:42] <deep42thought> the sanity-check really isn't that important
[15:19:50] <tyzoid_work> ok
[15:20:10] <tyzoid_work> btw, did you want me to remove that email alert?
[15:20:10] <deep42thought> it should just be run on a regular basis and if it complains, nothing else should be run anymore
[15:20:14] <deep42thought> no
[15:20:21] <tyzoid_work> ok, I'll leave it for you then
[15:20:44] <deep42thought> thanks
[15:40:30] <deep42thought> the commit message of the gcc-bug-fix explains the error: "Current glibc no longer gives the ucontext_t type the tag struct ucontext ..."
[15:40:43] <deep42thought> so it's sbcl's fault
[15:44:30] guido_van_chuckl is now known as guys
[15:56:08] <tyzoid_work> deep42thought: But that'd be an issue with the header file
[15:56:17] <tyzoid_work> any indication that they changed the implementation of that structure?
[15:56:24] <deep42thought> indeed
[15:57:00] <deep42thought> library/header .. where's the differenc, eh? ;-)
[15:57:01] <tyzoid_work> If the struct hasn't changed, we can add a patch to add the header until upstream changes their code
[15:57:12] <tyzoid_work> deep42thought: .so != .h :P
[15:57:15] <deep42thought> I'm currently compiling the git head
[15:57:23] <deep42thought> and so far it looks goot
[15:57:23] <tyzoid_work> ah, so that has the fix?
[15:57:25] <tyzoid_work> nice
[15:57:25] <deep42thought> good
[15:57:36] <deep42thought> dunno, I'm just a monkey hitting keys
[15:57:39] <tyzoid_work> lol
[15:57:50] <tyzoid_work> I am groot
[15:57:56] <tyzoid_work> buildmaster is groot
[15:58:12] -!- yokel has quit [Remote host closed the connection]
[16:00:37] -!- yokel has joined #archlinux-ports
[16:12:50] <deep42thought> ok, x86_64, git HEAD built fine with clisp, I'll now try i686
[16:14:11] <fsckd> y'all are awesome. thanks for working so hard on arch32. :D
[16:14:28] -!- deep42thought has parted #archlinux-ports
[16:14:33] -!- deep42thought has joined #archlinux-ports
[16:18:40] -!- deep42thought has quit [Quit: Leaving.]
[16:22:54] <tyzoid_work> fsckd: Thanks :)
[16:23:11] <tyzoid_work> fsckd: deep42thought deserves most of that credit, though
[16:25:44] <fsckd> :D
[17:37:58] <brtln> guys, let's face it
[17:38:36] <brtln> all credits for this should go to me
[17:38:37] * brtln hides
[17:40:32] -!- deep42thought has joined #archlinux-ports
[17:41:41] <deep42thought> brtln: you need to hide harder :-)
[17:43:04] <guys> brtln: you do know what goes hand-in-hand with the credits?
[17:44:21] <brtln> Not sure if I want to :D
[17:44:38] <deep42thought> this implies, you never got credit :-/
[17:44:56] <deep42thought> brtln: I appreciate your work!
[17:46:36] <guys> brtln: BUGSSSSSSSSSSSSSSS
[17:48:06] <deep42thought> fsckd: are you using archlinux32
[17:48:20] <deep42thought> if so, it might be nice, if you'd test some packages :-)
[17:49:20] <fsckd> sorry, i wanted to but my 32bit computer has boot problems. :(
[17:50:00] <deep42thought> you can install archlinux32 on 64 bits, too ;-P
[17:51:24] <fsckd> if my father hasn't thrown away his old half broken laptop, i may run it on that :)
[17:51:39] <deep42thought> :-)
[17:52:07] <fsckd> emachines laptop
[17:52:11] * fsckd waits for everyone to stop vomiting
[17:52:26] <fsckd> i think 256 MB RAM?
[17:54:22] <deep42thought> hmm, the i686 build for sbcl crashed :-/
[17:55:03] <deep42thought> the one using clisp
[17:55:06] <deep42thought> and the git HEAD
[19:08:06] guys is now known as privilege
[19:14:53] <tyzoid_work> deep42thought: that sucks
[19:15:15] <tyzoid_work> Same error? Or different error?
[19:15:25] <deep42thought> not really sure
[19:15:28] <deep42thought> maybe I'll have some time and muse this evening to look deeper into that issue
[19:15:46] <tyzoid_work> deep42thought: Quick thought, can you patch in the definition of that structure and see if it compiles?
[19:15:55] <tyzoid_work> I doubt that glibc has actually changed that structure yet
[19:16:00] <tyzoid_work> so it should "work"
[19:16:11] <tyzoid_work> and it'd be temporary patch until upstream fixes it for good
[19:16:22] <tyzoid_work> this would be on their source tarball
[19:16:29] <deep42thought> well, HEAD compiles for x86_64 with clisp - so they fixed the issue
[19:16:43] <deep42thought> but it fails for i686
[19:16:49] <deep42thought> probably at a different point
[19:17:12] <tyzoid_work> right, but something's up with i686. We (think we) know the issue for the current i686 source, right?
[19:17:17] <deep42thought> anyway, an error in HEAD is probably more important for upstream
[19:17:44] <deep42thought> it's the same as for x86_64 for current stable
[19:17:57] <deep42thought> but this is fixed in HEAD (at least for x86_64)
[19:18:11] <deep42thought> but as I said: maybe I have some time and investigate further this evening
[21:25:30] -!- titus_livius has joined #archlinux-ports
[22:08:26] -!- deep42thought has quit [Quit: Leaving.]
[22:09:02] -!- tyzoid_work has quit [Quit: Page closed]
[23:15:45] -!- p71 has quit [Read error: Connection reset by peer]
[23:20:59] -!- p71 has joined #archlinux-ports
[23:33:46] -!- shentey has joined #archlinux-ports
[23:35:26] -!- shentey has quit [Client Quit]
[23:36:46] -!- shentey has joined #archlinux-ports
[23:37:51] -!- shentey has quit [Client Quit]
[23:39:41] -!- shentey has joined #archlinux-ports
[23:43:29] -!- shentey has quit [Client Quit]