« first day (1426 days earlier)      last day (3511 days later) » 
00:00 - 15:0015:00 - 00:00

3:02 PM
I would personally much prefer @terdon's suggestion to the tag wiki.
 
@mikeserv I'm working on that now. I hope we end up having both, that way everyone is happy.
 
I sure as hell don't. The wiki is not centralized - it is not a single entity.
anyone can stick anything in the wiki and it could languish for gods' kow how long.
 
@mikeserv, I drafted whatever we discussed as a writeup here.
1
Q: per process private file system mount points

RameshI was checking unshare command and according to it's man page, unshare - run program with some namespaces unshared from parent I also see there is a type of namespace listed as, mount namespace mounting and unmounting filesystems will not affect rest of the system. What ex...

 
@Ramesh - I know. The votes there are mine.
 
@mikeserv thanks. Is there something that is missing or something wrong in it?
 
3:07 PM
well... it is lacking in detail.
All of that stuff about the kernel and vfs... That is primary to the discussion on namespaces - it is not ancillary.
 
Thanks. Can you please let me know what other details needs to be added?
 
Like you tried to equate it yesterday to the external hdd. Most people will do the same. Those that don't understand the difference have no cause to care.
 
@Braiam I found it.
@Gilles That would involve a fair amount of repetitiion, possibly. Depending on the subject. But no harm doing that as long as somone else doesn't try to edit it out. Has this been done / is this being done somewhere?
@casey The point that was puzzling me is that git doesn't try to clone the submodules locally. Maybe there is a reason for that - it just isn't obvious to me.
The people on #git ignored me.
@Patrick any idea?
 
@mikeserv, vfs is from kernel am I right?
 
How much rep does one need to edit tag entries?
 
3:15 PM
@FaheemMitha, 5000 I guess.
 
@Ramesh oh
 
@Ramesh - yes.
 
@Gilles hmm, not everyone can do this, then.
 
@FaheemMitha - I would be more worried about what doesn't get edited out. For as long as I've been coming to this site How can we organize some tag cleaning? has been a HOT meta question.
 
@mikeserv thanks. Added some more details to the answer.
 
3:17 PM
Ok, I guess others can make the edits, but it has to be approved. I guess that is reasonable.
@mikeserv Do you agree with putting canonical questions in tag wikis?
 
I am firmly opposed.t seems to me it would be a pretty arbitrary place.
 
Oh, you already said no, sorry
 
And it defeats the purpose of the tag anyway.
 
@mikeserv it actually seems reasonable to me. but i'd like to see where it had been done.
@mikeserv Why?
 
Do that and the tag ceases to be a tag and becomes instead a single link.
 
3:19 PM
@derobert what do you think?
@mikeserv Huh?
No, I'm pretty sure gilles means inside the text for that tag.
 
I mean, if the tag description points to a canonical answer - then what good are all of the others that description should describe?
 
sudo echo "helo there"
 
Password:
 
********
 
@mikeserv I don't follow.
 
3:22 PM
Sorry, try again.
 
not in /etc/sudoers file. The incident will be reported :)
 
sudo shutdown 0
 
The system will shutdown now.
 
@mikeserv No, the idea is to have certain Q&As in the tag's wiki. Like they do here for example:
 
@user965347 @mikeserv @Ramesh what the hell?
 
3:23 PM
@FaheemMitha, system is going for halt now.
 
@terdon Presumably just a link.
Are you guys playing some really obscure game?
 
@terdon. ok. that is better. In a FAQ format, maybe.
 
@mikeserv That's what I was saying. As was @Gilles. Not sure what you thought we were saying...
 
command not found: Are
 
@FaheemMitha I'm not git expert, but clone probably only works on the repo that is named, and submodules are technically their own git repos. That is probably why the submodule commands were created in the first place. who knows.. (not me)
 
3:25 PM
I still don't like though that it will be so decentralized.
 
@casey Yes, maybe. I don't know anything about git, except that is is too unixy for me. (Can't believe I'm saying that.)
@mikeserv Stop it.
 
this is another SO tag wiki that itself is basically a canonical answer: stackoverflow.com/tags/fortran-iso-c-binding/info
 
But, honestly, I feel a lot less strongly about that.
 
I think mercurial certainly has a much better ui. No competition, really.
 
I was thinking about hovering my mouse over a tag and being pointed to an answer.
 
3:27 PM
@casey @terdon is the idea to have links in the tag wiki, or paste the entirety of the question in there? I'm ok with the former, not so much with the latter.
 
I thought that was a terrible idea. Thankfully, it was also a stupid one.
 
@FaheemMitha in that case, submodules are just references to other git repos that serve as "shortcuts" so you don't have to clone/fetch all of them on your own. You could accomplish the same effect by 'git clone xyz; cd xyz; git clone 123'
 
Actually the latter seems like a terrible idea.
 
@FaheemMitha The former.
 
@casey Hmm, but if they are present locally, it would make more sense to copy them locally, I'd have thought.
@terdon Good
@casey the mercurial equivalent of git submodules is subrepos, and the first thing everyone tells you about subrepos is not to use them. i never have, so i have no idea how they work.
 
3:29 PM
@FaheemMitha git doesn't make that distinction. every git repo that is cloned to your machine is local, but it is still a git rep, and clone probably just doesn't traverse repo boundaries
 
@casey Apparently not.
Could be a waste of bandwidth, though.
 
@FaheemMitha no such thing as sub-repos afaik in git. Every submodule is a full fledged repo you could clone/fetch/push/pull on its own
 
@casey i see.
maybe not exactly the same concept.
 
its mostly a dependency thing
 
@casey what is?
 
3:32 PM
the need for submodules, at least that I've seen them used for
 
@casey oh, right.
 
e.g. if I'm writing something that depends on some library for a build dep, I can just pop that libraries git repo as a submodule to my own
then I can always stay in sync
and when people clone my repo, they can just do the submodule thing to bring in all of the sources I don't keep in my own repo
and they stay in sync without me duplicating the code
its nice for that
 
@casey Can I specify a version of that repo?
 
sure, if that repo is tagged or branched or you want to reference a specific commit
 
@casey Yes, that's the context in which the project i'm cloning is using it.
@terdon so you think the tag idea makes sense?
 
3:34 PM
I like it because if I'm doing it this way I'm much more likely to contribute back upstream with pull requests for things I fix locally.
 
it would certainly be better than fishing around for that meta question.
 
versus if I copied it into my repo and had no upstream dependency
 
@casey right. less overhead.
so you are mostly a git user, then?
 
@FaheemMitha everything I do is in git locally and some of it pushed to github
 
@casey what did you use before that?
 
3:36 PM
I keep all of my code, my vimrc, my emacs.d, my presentations and publications tex sources, etc
 
Ditto for mercurial.
 
@FaheemMitha cvs for code and nothing for the non-code
 
@casey ok
 
git didn't make a lot of sense to me coming from CVS, but once I wrapped my head around its paradigm I found I liked it a lot
 
You don't use the tex site? You don't even have an account.
 
3:37 PM
Hmmm. I just got a "what do you think?" ping delivered to my phone for a conversation that's apparently long over... And, cvs, svn, then git.
Wait. Actually, rcs first.
 
@casey Oh. I don't find it a good ui though
 
I vaguely recall MPW having something... if so, I bet I used that as well.
 
@FaheemMitha git has whatever UI you want to put on top of it really
 
@derobert Sorry about that. Not really over as such, we're sort of meandering towards concensus. so what do you think about the tag idea?
 
command line is fine for me, and gitweb is bundled with it
but you could scale all the way up to gitlab's UI or host on github
if you want a sophisticated UI
 
3:39 PM
@FaheemMitha I quickly looked and couldn't figure out what the conversation was about...
 
@casey command line ui, to be precise. not really a big gui fan when it comes to vc.
 
99% of my git usage is commit and push
 
@derobert brief summary. @Gilles advocates putting links to canonical questions inside the tag wiki. after some initial resistance, i think those of us here are more or less on board. was just wondering what you thought
@casey You don't use any incremental thingy?
 
@FaheemMitha incremental thingy?
 
like when you are in the process of working on a cset, and you want everything saved as you go along?
 
3:41 PM
I branch for some stuff, but mostly work in HEAD
 
@FaheemMitha Seems fine to me. As long as we're talking about the wiki, and not the excerpt.
 
once I progress to a point that I "release", I'll probably branch more
and do development in branches that merge back to HEAD for releases
 
@derobert I'm fuzzy on the distinction.
 
@FaheemMitha I don't have time to read that in detail, but doesnt git already do that to some degree. The current source is the sum of every commit since the initial one
 
@FaheemMitha excerpt has no format and is severely limited (500 char max), while the wiki is akin to any post, allows markdown and 30k characters
 
3:43 PM
and every commit is in the history
 
@casey That's true for both of them...
Well, maybe there is a 5 minute rule, like other posts, but other than that.
 
@derobert that was in ref to the git disco
 
@casey ah
 
@casey not quite the same thing. i can find you a digest version if you want.
The point is making the commits the way you want while not destroying any history. Nifty trick. Still very beta. I've personally found 3 bugs in the evolve extension in the short time I've been using it. then again, I seem to spend an incredible amount of time bug finding and reporting. Pain sometimes.
@Braiam ok
 
I'm trying to envision making a commit in git that doesnt preserve all of the history
 
3:49 PM
@casey well, rebase will do that, sort of. Not really, as the history is still around, if you know the ID (or check the reflog), but... it's no longer on the branch.
 
@casey You're missing the point. You make temporary csets, then alter them to make a final one.
But the old history sticks around.
We're all very excited about it.
@derobert evolve is similar to reflog, but better. of course.
 
@FaheemMitha perhaps I am. Isn't a cloned repo that is no longer at HEAD akin to a temporary changeset, and then committing it makes it final?
I've never used mercurial though, so my terminology there is non-existant
 
@FaheemMitha Mercurial folks always claim their stuff is better. Its a side-effect of the heavy metal poisoning, I suspect.
3
 
@FaheemMitha yes, it's being done even now on Unix & Linux and many other sites
@FaheemMitha google.com/…
 
@Gilles ok. examples? So i can see how it is done.
@derobert :-)
Better overall, dunno. Better UI, yes, definitely.
 
4:01 PM
> All registered users can submit edits for tag wikis.
 
@casey how is a repo comparable to a cset? don't follow.
@derobert Sheesh, two stars? What is with you people.
@casey imagine you are writing stuff, and it is sort of ok, and you want to save it, but you don't want to make a final commit? or you make three commits in a row, but then decide you would like to rearrange their contents?
so, with evolve, you can make commits as you go along, and change things later. i've been using mq for that for a while, but it wasn't really designed for that. evolve is.
 
@FaheemMitha that's what git rebase has done for a long time... Well, that and a lot of other things.
 
fold the commits into each other, rearrange or delete the contents.
 
yep
 
@derobert I don't use git, so I have no idea.
do the original contents stick around in history?
 
4:05 PM
@FaheemMitha google.com/…
That's the section title I use for lists of questions in a tag wiki. The title is just a personal habit, other people use different ones.
if I disappear, blame this crappy wifi
 
@Gilles Ok, thanks.
 
@FaheemMitha they're still there (in your repository) until you decide to remove them. You can find them with the reflog.
 
Anyway, evolve is the bomb. As they say in North America. Not meant in a violent sense, I should hasten to add. Much of NA slang sounds violent. "hit on" for example.
@derobert right.
 
@Gilles "Sept. 4--- Well-know Stack Exchange moderator vanishes, police investigating reports of an attack by a WiFi router. Sources which have asked $SITE not to disclose their identity say Anonymous may be involved."
4
(that'd of course be followed by some quote that the "journalist" made up, and then an attempt to explain what Anonymous is. That, of course, being the funniest part of the article.)
@FaheemMitha I'm not sure anyone still says something is "the bomb", except ironically.
 
@derobert Glad to hear it.
Now if only they could get rid of "hit on". And "hot", while they are about it.
 
4:22 PM
Those have been around for a while, so I doubt they're going anywhere.
 
@derobert Pity
 
@FaheemMitha thanks for clearing that up
 
@mikeserv You're welcome.
 
@derobert i like it when Anonymous is a terrorist organization. discounting the terror just the organization bit is always hilarious on its own.
@FaheemMitha look - a reply!
 
@mikeserv Ah, the reply thingy.
 
4:36 PM
@FaheemMitha so... where's my cookie?
 
@derobert It is exactly the same as Al Queida. Or however that is spelt.
@mikeserv Cookie? Cite needed.
 
@FaheemMitha They're boring these days. It's all about IS now!
 
@derobert IS?
 
@FaheemMitha - I don't think this is helping.
 
4:51 PM
@FaheemMitha sorry, what's the question?
 
The Islamic State (IS) (Arabic: الدولة الإسلامية‎ ad-Dawlah l-ʾIslāmiyyah), formerly the Islamic State of Iraq and the Levant (داعش) (ISIL /ˈaɪsəl/) or the Islamic State of Iraq and al-Sham (ISIS /ˈaɪsɪs/), is a Sunni jihadist group in the Middle East. In its self-proclaimed status as a caliphate, it claims religious authority over all Muslims across the world and aspires to bring much of the Muslim-inhabited regions of the world under its political control, beginning with Iraq, Syria and other territory in the Levant region, which includes Jordan, Israel, Palestine, Lebanon, Cyprus and part of...
 
5:29 PM
@terdon - is this an example of a canonical answer?
oh. I see the comment...
 
@Gnouc - is that... a compliment?
 
Does anyone know of any wrapper scripts I can run on scientfic linux that can actively check my nagios board and display results
 
@mikeserv Finally! I can award that bounty. I think I just pushed you over 10k.
n/m, that's network wide...
 
Yeah, I do have the potty break answer...
 
5:34 PM
@mikeserv: Yes.
 
30
A: How should bathroom breaks be handled during written exams to avoid cheating?

mikeservForbidding use of restroom facilities is extreme and more than a little ridiculous. If a student is resourceful enough to defeat your exam's purpose (which should be to measure a student's capability in a particular subject) without your certain discovery given only the use of a few minutes and a...

@Gnouc - much appreciated then. But, I mean, it's not that tricky. It's the same thing you'd do in awk with its FS right?
 
@mikeserv: No, I use perl, after a confusion with sed. Still get downvote :)
 
@derobert - thanks very much, though.
@Gnouc - not mine - I haven't voted there except on the question. But I mean, if you were to use awk, that's basically how you would do it, right? I don't know awk very well (which is to say - at all), but from what I see here, it usually looks similar.
shell $IFS is the same - there's nothing to fear. There are just the two extra considerations - 1 - IFS whitespace, and 2 - it only splits expansions.
 
@mikeserv: Yes, of course. I never consider shell as a solution for this, so I see yours tricky :D
 
@Gnouc - oh, and 3 - and filename expansions which you usually have to guard against with set -f but which is not a factor in this case.
@derobert - I usually don't care for the points, but I am really interested to see how the site changes at 10k when all of the deleted shit starts popping up. I know I've deleted some really ugly stuff, and so I have a pretty morbid curiosity about it.
 
5:55 PM
@mikeserv There isn't actually that much of it that you run across.
 
just break me down, man. stomp on a dream.
@Braiam - when you get this, guy came back with results.
 
6:11 PM
Hmmm. It's lunch time. I leave you with this question on compiling Linux ... wow. Where to start on that. Well, part of it is off-topic (e.g., coding standards). Other than that.... Maybe after lunch. If no one beats me to it. (And if its still open.)
 
@derobert - please answer that question. I actually had the link in the copy buffer to paste it here too before wifi-guy distracted me.
 
 
2 hours later…
8:14 PM
At last I get a chance to quote this link.
Yaay, I got a chance to provide this link. stackoverflow.com/a/1732454/1742825Ramesh 2 mins ago
 
 
1 hour later…
9:28 PM
@mikeserv What isn't? Sorry, I feel asleep.
 
@FaheemMitha - I am also sorry you feel that way.
 
@Patrick The question was about git submodules.
8 hours ago, by Faheem Mitha
09:55 < faheem_> moritz: consider i've cloned a repos, including submodules. call it foo. then if i do -> git clone foo foo-clone, shouldn't foo-clone get everything locally?
8 hours ago, by Faheem Mitha
9:58 < faheem_> i see, it rewrites the url for the main repo, but not for the submodules. i wonder why
@mikeserv huh?
 
@FaheemMitha - yeah. I get that a lot.
 
@mikeserv I'll bet.
@Patrick @casey replied to me. But it still seems odd to me.
 
@FaheemMitha - like here. Ramesh just politely let it go.
 
9:35 PM
@mikeserv you should do professional standup
@mikeserv ramesh probably thought he had fallen asleep and was having a really strange dream.
You don't question things in dreams, you see.
 
It's not that funny. But that it passed without mention was very funny to me.
 
@mikeserv You should get a audience of millions of American TV watchers. They're easily amused. Particularly children.
Is the reply thingy now gone with the wind?
 
@FaheemMitha - I take exception to that. I am not a child.
 
@mikeserv Are you sure?
 
fm - we're the only ones talking here.
who else could I be talking to?
 
9:41 PM
/me thinks I should hook eliza up to this chat.
 
who is eliza?
 
@mikeserv glad you asked.
ELIZA is a computer program and an early example of primitive natural language processing. ELIZA operated by processing users' responses to scripts, the most famous of which was DOCTOR, a simulation of a Rogerian psychotherapist. Using almost no information about human thought or emotion, DOCTOR sometimes provided a startlingly human-like interaction. ELIZA was written at MIT by Joseph Weizenbaum between 1964 and 1966. When the "patient" exceeded the very small knowledge base, DOCTOR might provide a generic response, for example, responding to "My head hurts" with "Why do you say your head hurts...
I specifically had in mind the emacs version.
 
Which is the picture I was just looking at.
 
@derobert Thanks. I'd never heard of them before.
 
Evening folks
I've installed transmission-daemon on my server, and it works great
But every now and then the service mysteriously and spontaneously stops.
I'm running CentOS 6
 
9:57 PM
Is it spontaneous? I used to have a WRT54G that would belly-up occasionally due to there being too many simultaneous active bittorrent connections.
 
Is it possible to get a notice when a service sometimes stops?
 
The thing is, most of those torrent programs don't clean up after themselves soon enough and the thread count just piles on.
like expensive sheets
my solution was to change a DD-WRT setting for connection time-out.
I don't know how to do same in Linux. @Braiam probably does.
You should ask a question.
 
Seconded
Along the lines of how can I diagnose this, perhaps?
 
@FaheemMitha that behavior makes perfect sense to me. your submodules URLs are defined in a plain old file called .gitmodules. That file is part of the repo. URLs aren't going to change unless you update the file
 
@Patrick Ok. Just seems like a waste of bandwidth, since presumably the submodules you want are available there locally.
 
10:08 PM
ah, so you're asking why the clone didn't copy them? Remember git is decentralized. Cloning from a filesystem path is the exact same as cloning from a remote URL
 
@mikeserv Doesn't appear to be a "too many connections" issue.
It happens even when I don't have any torrents active.
 
right - you don't have to be up/downing to maintain a connection though.
 
Connection to what?
 
you can watch it with ss
 
I removed all torrents
No announcers, no trackers, no peers.
 
10:10 PM
oh, well, then it's probably not.
which makes this chatroom - at least as far as i am concerned - an even less likely source for an answer than before
 
The funny thing is, it's not the first time I tried to run a torrent client on it, and that one crashed randomly as well
No error message no nothing
 
it was transmission both times
?
 
@Patrick Ok. I wonder if there is some way to get it to do a local copy. Any idea?
 
i always liked rtorrent
 
@mikeserv First one was rtorrent
 
10:13 PM
wow. i'm batting 0.
 
@FaheemMitha dont think so
 
@Patrick Ok, thanks.
 
well, i dunno man. but you really should ask a proper question about it.
 
Will do
 
and - you can always redirect stderr at process launch to some log and then check its tail after a crash.
 
10:20 PM
@mikeserv mmm?
 
I think that is a nogo, anyway, @Braiam. It was about shortening tcp timeouts, but probably not relevant actually.
But the other thing - the one I asked you to look at yesterday? That guy came back.
With the results from your script.
 
@mikeserv it could be that the kernel exhausted the connections tables (I don't know exactly how it's called, just that there's something like that)
 
@Braiam Happens when no torrents on the list
So I don't think it's the connections thing.
Also, it's a VPS, it's supposed to serve a lot of people at once.
 
yeah - that was what was killing my router. so shortening the timeouts managed it.
 
@MadaraUchiha are you allowed to run torrents in that VPS?
 
10:24 PM
@Braiam Yes
 
11:08 PM
@mikeserv absurdly long answer to that question posted... unix.stackexchange.com/a/153842/977
... of course, interestingly, it seems it was put on hold while I was writing that :-(
Nice of the site to not bother telling me.
Odd that I could still post the answer.
 
@derobert - The same thing happened to me earlier with this (almost answer) comment...
GNU: sed -sn '2{F;p}' *|paste -sd'\t\n'mikeserv 4 hours ago
I was so mad at Ramesh!
@derobert - halfway through. But this is really good.
 
@derobert you can answer a question up to 4 hours of being closed
 
@mikeserv sorry man! I tend to go with general public.
 
11:23 PM
No, I know. It was fine.
 
Done! Nominated to reopen.
 
It probably should have been closed. But the go with public thing may not be so fine...
 
But I voted to close since the OP did not care to provide an input and expected output.
I mean the question may be clear. But providing an example will do good for more broader audience.
 
@Braiam weird. when I was doing mine I got a little message bar at the top that said (this question is now closed - no new answers will be accepted).
 
@mikeserv Oh, there may have been a bar at the top, but, well, the top was nowhere near...
 
11:25 PM
dammit! I should have just hit the button.
It's exactly 4 hours ago now though.
what's the point of f()? A noop?
A cheap one?
 
if you have enough text, you can still put the answer
 
@mikeserv The void f(); up top tells the C compiler there will be a f() defined somewhere... The one inside the if calls that f()
Of course, there isn't an f() defined, so the linker will error out on -O0.
 
I know - but there's nothing there. Why does it call it?
 
That code doesn't do anything useful---it's just an example.
 
oh.
of course.
 
11:28 PM
Actually, I bet if you compile that with -O2 gcc will emit a program that does literally nothing
 
does killall work in case of a forkbomb?
 
@Ramesh nope
 
As in, I bet gcc optimizes the entire program to int main() { return 0; }
@mikeserv so I guess then that program does have a use, its an implementation of true :-/
 
But this answer suggests it can be done with killall.
11
A: How do I kill a forkbomb process?

Colin Pickardreboot the computer if you can't, you could try: killall -STOP -u user1 killall -KILL -u user1 If by "permanent solution", you mean preventing this happening again, well you can't really, but you can simply reduce the ulimit for the problem users and ignore them.

 
@Ramesh It depends on the forkbomb. If it manages to kill them faster than they respawn, it'll work.
So, it may work on accidental fork bombs. Purpose-built ones are probably too fast.
 
11:31 PM
@derobert ok thanks.
 
attempting with -STOP first may help, too. That won't free their process slots for re-use.
 
2
A: Has the Israeli Intelligence, or Army (Mossad, IDF) ever been implicated in human trafficking?

BobsonGoogle says: No. This answer brought to you by the same level of effort put into the question.

 
@derobert hooray! it's fun to be right.
 
Anyway, good night.
 
@Ramesh I've accidentally outpaced killall on many occasions I'm weirdly proud to say.
 
11:38 PM
@mikeserv From what I understood so far, the signal can be blocked by kernel code.
So do you mean, you outplaced them using kernel code may be?
 
what signal?
I mean I outpaced it.
 
sigkill and sigstop
 
As in - it couldn't kill my accidental forkbombs fast enough - they spawned too quickly.
@Braiam - that is hilarious
 
@mikeserv, I am not sure I follow your comments.
 
Well - you directly quoted me. Immediately following that quote there is a big bold word that reads: References. But there is no reference to the quote.
 
11:47 PM
Ah ok. Sorry about that. Just a minute.
Do you mean I should paste the url to our chat over there in the references section?
 
I dunno - it's your thing. It just seemed to be an incomplete References section to me.
 
@mikeserv, sure. I am figuring out how to get the chat transcript.
 
00:00 - 15:0015:00 - 00:00

« first day (1426 days earlier)      last day (3511 days later) »