/dev/chat - 2014-10-18

« first day (1470 days earlier) ← previous day next day → last day (3484 days later) »

12:03 AM

@Gilles - is this not what you're looking for?

lkml.org/lkml/2012/5/30/298

> Cong> This is a draft patch of implementing per-file drop caches.

Interesting. So can I do this from outside a process? I'm a
SysAdmin, so my POV is from noticing, finding and fixing performance
problems when the system is under pressure.

Cong> It introduces a new fcntl command F_DROP_CACHES to drop
Cong> file caches of a specific file. The reason is that currently
Cong> we only have a system-wide drop caches interface, it could
Cong> cause system-wide performance down if we drop all page caches

I cannot find out if this ever made it past this point though

@slm yes

none of the linux repos I can find have this code in their implementations of fcntl.c

@Gilles - You interested in applying that patch to fcntl.c?

12:18 AM

@slm no, I'm not recompiling a kernel for this

I'll add this as an A but keep looking

since it's of historical significance

12:45 AM

@Braiam That's sarcasm right?

@Seth Have you not met Braiam before?

He's all sarcasm 24/7

@Braiam http://unix.stackexchange.com/q/162803/4358
With all the command missing crap (the `command-not-found`, `id: not found`, `x-terminal-emulator: not found`, etc). I'd bet you money the user has gone and deleted something catastrophic

@slm Met him? I've never even heard of him! ;P

@slm I just reverted your edit. You just added information to the question that was already in the question

ah I missed it up there, thanks

why did he repost it, or did someone else edit it in?

1:00 AM

typical newbie behavior. add it as a comment, discover it looks like shit, adds it to the question instead, doesn't delete the comment

ah, OK,I processed the flag that it was too chatty, did you flag it then?

yes

doesn't show us who flagged it so I can only guess.

if you flag those as other then say something like I merged this crap in then I won't attempt to do the merging myself

I did :-)

The comment said too chatty

1:02 AM

keep reading :-)

going to double check

> Delete: too chatty (as in too long), and the content has been added to the answer.

s/answer/question/

The comments don't fully show in the crappy little window we get either

so I saw: too chatty ....

I'll go look on the history of the Q next time.

The next time it comes up I'll take a screenshot and open a meta about it too.

so we can get more visibility on the msgs

@Patrick I know he doesn't have cat, at very least

I see a system rebuild in someone's future

1:17 AM

mm... decisions, decisions... Chewbacca or Darth Vader?

@Braiam Darth Vader, of course.

@Gilles - you still around?

@FaheemMitha boo! I would for Yoda

3 hours later…

4:07 AM

@Gilles - figured out how to drop the file.

4:20 AM

@slm, just curious. you knew of such a thing or just searched now?

searched

I suspected it was possible from my previous work around fatrace research and my work in researching how the mm works in linux

@Ramesh ^^^

cool. I couldn't come up with any. I believe you are hungry to know more and more things :)

that and now we all know this, and we've created a giant marker on the internet for how to do it.

15 hours later…

7:31 PM

I feel like I'm working a bit hard here (see comments). Two questions: (1) am I wasting my time (2) am i missing anything? Debian users and/or people particularly familar with X, comment please.

unix.stackexchange.com/q/162807

If bugs.debian.org/668684 is really the problem the poster is seeing, that is puzzling. It should be fixed in his version, which appears to be the same as my version.

I ended up compiling libdrm and also Mesa 10.0.5. But I'm now facing another weirdness. I updated the outputs up there. — user3755746 5 hours ago

I don't have idea what was the version of Mesa he was trying first

@Braiam Yes. I've told him repeatedly to avoid local compiles. They're always problematic. In the hands of a beginner, doubly so.

@Braiam do you have any idea what is with this 0.6 thing?

7:46 PM

@FaheemMitha old bug, solved years ago, he was hitting it because he was using older sources

@Braiam According to info provided, his package version is same as the one currently in my system (7.6). Look at the dpkg -l info.

I'm watching the IT Crowd. Good stuff. Wish they would make more episodes, but doesn't look likely to happen.

@FaheemMitha the bug is in Mesa, not in the system libraries

@Braiam Oh? I missed that. What version of Mesa has the bug? He was seeing it when trying to compile Mesa.

@Braiam That answer says you deleted it. How come I'm still seeing it?

Your answer said "You have the libdrm_nouveau libraries, but the version is too old." So, that is not correct?

The question says Mesa 9.0.3 initially. Then he mentions Mesa 10 later.

Incidentally, the question isn't off-topic. It is a bit of a mess, because the poster is clearly in over his head. But he is at least cooperating as best he can, and he is being civil. Which is better than a lot of other people.

@slm ^^

8:06 PM

@FaheemMitha 10k

@Braiam Oh, I see. So over 10k means you can see deleted messages? Hmm.

@FaheemMitha 9.1.3

@Braiam Ah, I see. So he should be using Mesa 10 then.

@FaheemMitha the original problem is no more since OP is using 10.x instead of 9.x

@Braiam Except he is doing local installs, so it is conflicting with the system packages. I think I'll add a clarification.

8:10 PM

@FaheemMitha his original problem is the message about nouveau-libdrm soname version, that's solved, what are you saying is the problem now?

@Braiam Well, he needs to understand what his problem was. I didn't, so he probably doesn't either.

I just posted:

Having had a word with @Braiam in chat, a clarification. Apparently the 0.6 problem was not with libdrm-nouveau, but with Mesa, and was fixed in 9.1.3, according to @Braiam. I haven't checked that. Anyway, remove local installs, and backport Mesa 10 from testing/unstable. If you want help, post another question. — Faheem Mitha 1 min ago

Does that look Ok to you?

8:32 PM

dv-pls unix.stackexchange.com/a/162841 — using PS1 to test whether a terminal is present is completely wrong

> (…) mikeserv is correct (…) — Gilles

@Gilles - yeah already voted up his comments and DV'd the PS1 A'ers.

He's trying to explain to others that it's wrong, but I don't think they're following why.

I haven't read the Advanced Bash Scripting Guide, is it always that bad?

That page that goldilocks cited has several major errors

That section from the Advanced Bash-Scripting Guide has several errors. PS1 is not a reliable test to tell whether the shell is interactive. “If a script needs to test whether it is running in an interactive shell” is also confusing: it should be if some code needs to test — a script is usually not running in an interactive shell (but it can be, if it's sourced). Testing for i in $- is the correct way to test if the shell is interactive. Testing -t 0 or -t 2 is the correct way to tell if the script is running in a terminal, which is different from being interactive. — Gilles 4 mins ago

As for the -p bit, it contradicts the comment above that mentions a socket, for which the test would be -S. But I don't know what John Lange is trying to do there: if stdin isn't a terminal over SSH, it means that the script is not executed interactively, in which case it probably shouldn't do whatever it likes to do in a terminal, and in particular it shouldn't try to perform any user interaction. — Gilles 2 mins ago

8:55 PM

@Gilles the people on #bash recommend something else I think.

@Gilles Yeah, you have to always cross reference it. The examples are very hit or miss if they'll work. Stephane has provided them fixes a number of times but it's probably better to ditch it at this point.

is there a new FSH draft in the cooks?

9:14 PM

@Braiam in the cooks?

You mean, "in the works"?

9:27 PM

Apparently systemd is not default on jessie?

Would one notice this in a chroot?

@FaheemMitha phoronix.com/scan.php?page=news_item&px=MTgxNjM

@Braiam Yes, I read about that. GRs are always fun. I see Jackson isn't letting this drop.

I would actually prefer Debian to not depend on a particular init system, but wonder if there is enough manpower to deal with the extra layer of abstraction.

9:53 PM

iuculano.it/linux/apt-get-purge-chromium <- NOT GOOD FOR DEBIAN CHROMIUM USERS

@FaheemMitha I linked that some time ago

@Braiam Oh, you did? I didn't see it. Do you follow Debian Planet?

Oct 13 at 17:24, by Braiam

there we go http://www.iuculano.it/linux/apt-get-purge-chromium/

no

@Braiam ok

it appeared in my Gmail while reading a bugmail

9:58 PM

changelog.complete.org/archives/… <- JOHN GOERZEN ON SYSTEMD

@Braiam oh

Some of this might be teething problems. Hard to say. Anyone using Jessie care to comment? I should also mention John is very level-headed. And a DD. Not necessarily synonymous terms...

10:20 PM

Anyone familiar with Debian packaging, quilt in specific?

@Seth I am, somewhat. Probably more than most. What's up?

quilt is usually used to generate Debian patches. as in debian/patches.

I need to patch a program I'm packaging because it uses ffmpeg and so far the safest way to meet the dependencies seems to be patching to avconv. I made the patch, but now when I attempt to build builddeb fails at quilt with:

> bzr: ERROR: An error (1) occurred running quilt: None

The working tree was created by an older version of quilt. Please run 'quilt upgrade'.

When I run quilt upgrade I get:

> The quilt meta-data in .pc/ are already in the version 2 format; nothing to do

@Seth You need to stick a @ in front, or I might not be notified.

I can't find anything online at all, which is disconcerting.

I could take a look.

10:31 PM

@FaheemMitha If you'd prefer :)

If you want to email me, it is faheem at faheem dot info

A lot of people don't like being pinged all the time so I normally avoid it.

@Seth Well, fair enough. I wouldn't want to be pinged constantly either. But if you are trying to get someones attention...

@Seth I've never seen that error message. A patch is a patch. Weird.

@FaheemMitha Sure I understand :)

Why not just recreate the patch(es) from scratch?

Oh, you made the patch from scratch? What is your version?

10:34 PM

That's doable, but I was hoping for a more long term solution, since the program is still under active development as time goes on manually patching would get harder and more complicated.

Oh, when you said manual I thought you meant edit all the files and keep track of it manually.

@Seth I didn't mean that. You presumably made the patch using quilt in the first place?

Yes.

What VCS are you using with it, git or hg?

quilt --version says I'm using 0.61.

@FaheemMitha This is Launchpad so I'm using bzr. Does it matter?

@Seth No.

I meant, what Debian version.

I'm quilt 0.60 here.

Debian stable.

If you want to point me to your packaging I can take a look right now.

10:36 PM

I'm not actually using Debian itself (I thought you would know that, my bad). I'm using Ubuntu 14.04 right now, which is based on Debian unstable (as it was in March/April I think)

@Seth Ok.

@FaheemMitha What specifically would you want to look at? I don't have it at any publicly available location right now, since I can't get it to build.

Regardless, can test on stable. Or on jessie (have a chroot here)

I verified the quilt format version is at 2 manually by checking .pc/.version.

@Seth Your current version of the packaging.

Maybe an expert could diagnose your problem from the description alone, but I can't.

10:39 PM

Sure.

If this is a fresh attempt at packaging, I'm puzzled by the older version of quilt thing.

What if I tar up my source directory and upload it to a cloud client where you could download it?

That seems to be the easiest option.

Or I could start over and give you a copy of all the commands I ran and their output.

@Seth Sounds reasonable. You don't have the debian source files for this, i assume?

.dsc etc.

I guess this would be better tested on jessie. What is the sw name?

no, because the build fails.

All those are generated during the build.

@Seth Right

This is a from scratch packaging effort?

10:42 PM

this is the software: github.com/colinkeenan/silentcast/releases/tag/v2.0

This is from scratch, yes. I've packaged the older versions before, but I always start brand new for the next version.

@Seth Are you building against current head? If so, just send me the debian dir? You could tar it up and send as an email attachment, even. It can't be very large.

Oh, I see. there is a release. So i unpack that, and stick the debian dir in?

Yes, pretty much.

I'm getting the debian directory now.

(you might have to apply the patch, I'm not terribly certain how automatic quilt is)

@Seth quilt should be pretty automatic. you are using 3.0 (quilt), right?

Yes.

ok. so, email sent? i downloaded the tarball

or you could stick the tarball somewhere and give me the link. i don't care.

10:52 PM

Just packaging.tar.gz. Ignore the rest.

8.9 mb?

what do you have in there?

I prefer something that is wgettable, personally

I don't really have a method of generating a wgettable link.

Is it really 8.9MB?

Hm.

Oh, that's the .bzr file apparently.

Dunno if you need/want that or not.

@Seth builds without fuss here. What command are you using to build it?

Also, it doesn't seem to actually do anything. What language is this?

Python, maybe

I use debuild -uc -us. afk

11:08 PM

yes, I was just debugging with someone in irc, the normal debuild command works, but bzr debbuild does not.

Even more odd :S

@FaheemMitha A combination of bash+python.

So I guess that's as far as it goes. Something bzr does messes it up.

thanks for your help @FaheemMitha

@Seth sure

I think I'll purge my build environment completely and try starting over.

@Seth Sounds like a plan.

I'm doing it in a VM so it should be too bad.

The error itself is a huge pain tho, because no one else seems to have even encountered it.

@Seth Hmm. is this a bzr-buildpackage deal? Assuming bzr-buildpackage exists.

Grocery stores in Norway use Ubuntu, apparently

11:25 PM

@FaheemMitha I'm not sure what you mean.

@Seth your bzr debbuild thing. is it part of a larger package?

oh, I think it's part of the bzr package.

@Seth ok

« first day (1470 days earlier) ← previous day next day → last day (3484 days later) »

Transcript for

Oct '1418

/dev/chat

General discussion for unix.stackexchange.com. If you have a q...

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile