=== yofel_ is now known as yofel | ||
=== pedro__ is now known as pedro_ | ||
=== jdstrand_ is now known as jdstrand | ||
=== plars_ is now known as plars | ||
=== Ursinha_ is now known as Ursinha | ||
penguin42 | anyone understand lib[open]blas/atlas? I've got a couple of bugs that are referring to crashes on newer AMD chips, it looks like they removed 3DNow instructions on some of them | 16:32 |
---|---|---|
hggdh | penguin42: I am trying to find out what to do with steam bugs | 17:11 |
penguin42 | thanks | 17:13 |
penguin42 | I expected some cracks from chemists; steam and tar.... | 17:14 |
ecolitan | hello, I'm trying to learn to triage, does someone have time to help with a specific bug i'm looking at first? | 18:40 |
penguin42 | sure - which bug | 18:40 |
ecolitan | 414422 | 18:40 |
ecolitan | because it looked easy :) | 18:40 |
penguin42 | now if you say bug 414422 | 18:41 |
ubot2 | Launchpad bug 414422 in sudo (Ubuntu) "sudo: unable to execute /bin/mv: Argument list too long" [Undecided,New] https://launchpad.net/bugs/414422 | 18:41 |
penguin42 | the bot does good things! | 18:41 |
ecolitan | bug 414422 | 18:41 |
ecolitan | bot doesnt answer for me | 18:41 |
penguin42 | huh, bad bot | 18:42 |
ecolitan | so according to the flowchart, first thing is to decide if it's really a bug | 18:42 |
penguin42 | yeh, and I'm not sure it is | 18:42 |
ecolitan | me neither | 18:42 |
ecolitan | i cant reproduce in bash or sh | 18:42 |
ecolitan | and suspect the cause was something with shell escapes | 18:43 |
penguin42 | well it is also 4 years old; I think there used to be fixed limits on argument length - I thought Linux was always quite generous/dynamic on them though | 18:43 |
penguin42 | ecolitan: Just because you can't repeat it doesn't mean that it doesn't happen for the originator | 18:43 |
ecolitan | ok | 18:44 |
ecolitan | so with a 4 year old bug thats probably very environment specific, is it worth asking follow up questions? | 18:45 |
penguin42 | ecolitan: Ah, try this sudo -s /bin/echo `seq -f "/a long tomb of data goes here %g" 1 10000` | 18:45 |
ecolitan | ahhhh | 18:46 |
penguin42 | doesn't happen without the sudo | 18:46 |
penguin42 | which I'll admit to being surprised by | 18:47 |
penguin42 | but is it a bug or just a limit, most Unix's have much stricter limits on arg limits | 18:49 |
ecolitan | 3572 was the limit of args | 18:49 |
ecolitan | hmmm | 18:50 |
penguin42 | 3573 here :-) | 18:50 |
ecolitan | if I make the string smaller, i can have more args | 18:50 |
penguin42 | yeh I think it's a total length thing | 18:51 |
penguin42 | I can get it to fail without the sudo if I turn the length up to be fair | 18:51 |
penguin42 | ecolitan: I'd say not-a-bug | 18:51 |
ecolitan | even if the behaviour is not in sudo documentation? | 18:51 |
penguin42 | this fails without the sudo - so the arg list does have limits /bin/echo `seq -f "/a long tomb of data goes here %g" 1 90000` | 18:51 |
penguin42 | it's just normal shell/process arg limits I think | 18:52 |
ecolitan | ok so according to the chart, neither a bug nor a support question, mark as invalid | 18:53 |
penguin42 | I think so; you could point to use of xargs or find to do the job without running into the length limit | 18:54 |
ecolitan | so no importance is set for invalidated bugs | 18:57 |
penguin42 | correct | 18:57 |
penguin42 | bbl - food calls | 19:00 |
ecolitan | thank you | 19:00 |
ecolitan | hopefully will participate thursday | 19:00 |
=== e11bits_ is now known as e11bits | ||
hggdh_ | penguin42: Steam bugs should be reported on Steam's BTS -- see https://wiki.ubuntu.com/Valve#Problems.3F | 20:15 |
penguin42 | hmm - I don't actually run steam myself and don't know much about it, other than those two bugs - I could create an account and report it on there but wouldn't be able to answer much in the way of questions | 20:17 |
penguin42 | hggdh_: So I'm thinking I delete the core files from those two bugs and then mark them public, dupe them and then report to valve? | 20:18 |
penguin42 | ah looks like it was fixed a few days ago | 20:21 |
penguin42 | right, done | 20:25 |
penguin42 | hggdh_: Thanks | 20:26 |
hggdh_ | penguin42: yea, I would say remove the coredump, open it up, and close invalid with a reference to Valve's BTS | 20:28 |
penguin42 | already done :-) | 20:28 |
hggdh_ | penguin42: I also had this response: the crashes were from their bundled libraries not ubuntu ones so launchpad doesn't make sense, for that bug you want to launch steam with STEAM_RUNTIME=0 steam then go to file->update and it should be working after the update | 20:29 |
hggdh_ | (the first part we already knew, the second is interesting (given that I never ran steam before) | 20:30 |
penguin42 | yeh makes sense; an interesting question I guess is whether something like tar should run with other libc as long as it was a sanely built one | 20:30 |
hggdh_ | penguin42: frankly, common (like "core") utilities should use the system. Otherwise we end up with another distro over the distro | 20:32 |
penguin42 | hggdh_: Yeh, I think it should be possible for something in the bug system to flag when another libc is in procmaps | 20:32 |
hggdh_ | we would have to hack apport for that. I wonder if it is worth the trouble | 20:34 |
penguin42 | yeh it's probably rare | 20:35 |
=== hggdh_ is now known as hggdh |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!