« first day (1483 days earlier)      last day (3454 days later) » 
00:00 - 18:0018:00 - 00:00

12:00 AM
@Seth sure
the answer is that he's out of luck, but that's not his fault
 
Cool, coming your way.
 
Midnight, finally! I can vote again!
 
0
Q: How can I compile and run a program designed for an older kernel version (2.6) in a newer kernel (3.16)?

Aaron FrankeI'm trying to run a driver for a network card, but it only supports kernels up to 2.6.32. I need to be able to use this piece of hardware to connect to the Internet. The device I'm using is "VIA Technologies VT6105/VT6106S [Rhine-III]". I know this is possible somehow, because I had the network...

 
@Seth going by the comments, smells XY
yup
 
Certainly possible.
@slm Did Linus ever reply to your message?
 
slm
12:27 AM
@Seth - nada
I'll probably broaden the appeal and post it on the linux kernel ml
unless someone else wants to, but probably better if a mod posts it there
 
12:41 AM
Hi @Seth.
@slm That's what I originally suggested.
 
slm
@FaheemMitha - thanks for the I told you so. I tried a direct route.
isn't this a dup?
1
Q: Autohot Key Equivalent?

Meer BorgIs there an equivalent product/method for AutoHot Key. For those that don't know, its a product that allows you to program your mouse movements and keyboard. This allows me to "macro" certain functions on programs instead of having to do it manually. In the old days they used to call them keyboar...

 
@slm I didn't find one
 
slm
0
Q: key replacement (like autohotkey?)

epizeuxisI have several broken keys on my laptop keyboard, which I replaced with key commands in Windows using autohotkey. Autokey, the closest linux alternative, has fallen out of development and won't run on my system. for example, I would like: ctrl+, == m

it's called autohotkey
he typed the name wrong
 
@slm Sorry. :-)
 
that's not the same question
they're using ahk for different things
 
slm
12:48 AM
yeah looking closer you're right, it's just about autohotkey
there was this Q before looking for alternative to autohotkey, maybe it was SU where I saw it when researching one of these Q's from before
 
 
3 hours later…
3:28 AM
Does anyone know how to save a page source from Chromium to a text file?
 
3:43 AM
@FaheemMitha Hi :)
 
@Seth Hey. What's happening?
 
@slm I can't say I expected him to, but :(
@FaheemMitha waves arms Flags. Flags everywhere.
But I'm taking the Linux course on edX. That's more interesting.
 
@Seth Flags? Huh?
@Seth He might have. He has time to write reams of drivel on G+
 
@FaheemMitha I've handled over 150 flags this week on the site I moderate and there are still more :/
 
@Seth Oh, right. SE moderation. Somehow I was thinking about flags in a different context. Must be tiring. How about your fellow mods?
 
3:48 AM
I'll just say we've had a TON of flags this month, so there are always more :)
 
Are these spam, users misbehaving or other?
How does it compare to U&L?
@Seth know anything about autotools?
 
I have no idea how it compares to U&L..
@FaheemMitha Sadly, no :/
Got a question about it though?
 
@Seth Ok.
 
@FaheemMitha Other mostly.
 
@Seth I'm messing with it.
 
3:51 AM
Unfortunately, probably due to Ubuntu's image as a newbie friendly Linux, we get a LOT of non answer answers ("me too" etc).
 
which was a good choice. I'm currently thinking the "Link against the library of common sources" one
@Seth Is that your major category?
 
IRC is sometimes nice, if people are actually there.
@FaheemMitha mm, not sure what you mean. Do you mean flag wise?
Or what site I spend time on?
 
@Seth yes. Re flags.
 
Ah. I'd guess so.. I don't have any real stats on that, but yes, probably.
 
So, bottom line, lots of "me toos"? Must get tedious. Maybe have a (re)education campaign. :-)
 
3:56 AM
Yes. Most of them come from new users so educating them is hard :/
Ooh, a vim question.
 
@Seth IRC is variable. Often there are totally silent sites with lots of people idling. It's like a ghost town. Or a zombie movie.
 
Yes, it is terribly annoying sometimes. And then there are the busy channels.. Phew!
So I was wondering.. How do you guys pronounce SUSE? The guy in the Linux course just pronounced it "soo-say", I always thought it was like "soose" (think moose). What do you say?
 
4:09 AM
@Seth I think I've used both versions, actually. Maybe ask some actual SUSE people.
I just asked that question above on 3 IRC channels, and got one "SO sux" and one "dunno" (from a DD).
 
That's the problem with programming language, natural language has grown to discriminate sound, while coders are silent speakers :)
 
@prosodycontext Wow, that's almost poetic. :-)
 
<3 I guess I've been running through that string of judgement for long enough to have a feel for the echo. ;) @FaheemMitha
 
slm
@Seth We've had 113 combined for the wk 8-)
@Seth - what's your guys turn around time on flags?
@Seth I've always heard it pronounced Sooos Seh
 
@slm ~2ish hours, I believe.
@slm Interesting.. ok then :)
 
4:19 AM
@prosodycontext visiting?
 
@FaheemMitha I don't know anyone who uses SUSE xD
Anyway, time for bed over here.
 
@Seth Take care.
What is the current active mod strength on AU? I think these sites could use more mods.
 
@FaheemMitha Hmm? You mean why here? Trying to reach out I guess. Finding good chat rooms is hard.
 
@prosodycontext Define good.
 
:).... yyuupp..
 
slm
4:20 AM
That's not bad, we've been ~20-40 mins.
night Seth!
 
E.g. U&L has probably 2 active, with a bunch of other high rep users helping out.
 
slm
We have 3 active mods and probably 10 active high rep users
With another tier of 10 new users that have been active
Michael, Terdon and myself are here most of the day
 
@slm Don't see Michael in chat much. At least, recently.
 
slm
the load seems to be very much equally divided which each of us handling roughly a 1/3 of the flags and such
he's just as active as terdon and myself, but tends to be quieter about it
we chat with him as needed in various mod chatrooms
 
@slm I see. That's good.
 
slm
4:26 AM
yeah, I like it since being new it's nice to have others to draw off of, there's a mod room for all the mods across the sites where we can ask for guidance and help too.
there's a very extensive support network in place that is completely invisible to the visitors
when you combine that w/ the community managers, the other mods, the high rep users, and everything else it's actually a very well staffed operation
 
@slm Good to know.
 
4:46 AM
Its nice having my security blanket back in place (my 4x2TB raid5 back into a normal non-degraded state). I've never appreciated my backups and raid disk redundancy more than this week...
 
5:08 AM
@casey That's a lot of space. How much is your total usable space?
 
/dev/mapper/vgr5-data 5858043904 4360729996 1497313908 75% /data
 
In human readable form? :-)
-h
 
/dev/mapper/vgr5-data 5.5T 4.1T 1.4T 75% /data
 
Wow, that's a lot of space.
@casey know anything about autotools?
 
@FaheemMitha I used to.
but my knowledge on setting it up for a project is a little over 10 years old
 
5:11 AM
@casey Not any longer?
 
what do you need to know?
 
@casey Nothing specific. I'm just messing with it.
I have a project that uses scons. I'm experimenting with building it with autotools.
 
@FaheemMitha thats nothing. this is a lot of space:
/dev/glade_scratch    4.7P  3.8P  857T  82% /glade/scratch
/dev/glade_u          786T   13T  774T   2% /glade/u
/dev/glade_p          9.3P  3.9P  5.4P  42% /glade/p
 
@casey That's some supercomputer? And space, like everything else, is relative.
 
thats the shared storage on yellowstone
 
5:15 AM
@casey Which is a supercomputer?
@casey "that's not a knife, this is a knife".
 
I also have this for local backups:
/dev/mapper/backupvg-backup 7.2T 5.2T 1.6T 77% /mnt/backup
 
@casey You must spend a lot of money on HDDs.
 
unfortunately. I spent ~$500 on disks just this week to replace my two failed disks and to have two more in cold storage to replace the other two when they die (I'm assuming it'll happen before long)
its a necessity of working remotely though. I need the resources to work with my datasets locally and backup is a must
 
5:32 AM
@casey Sure.
You used to use autotools and no longer? Option a: you use another build system. Option b: you just write plain non-portable Makefiles or similar.
 
right now plain makefiles meet my needs
I used it a project a long time ago just to check it out
if I had a complicated C/C++ project I was working on, i'd probably go for it again
 
@casey Ok. So portability is not an issue for you?
 
the code is portable
 
@casey I meant portability of the build
 
the build is, its not not a complicated build
the only portability issue needing to manually put the path of the hdf5 library in the makefile
but not even autotools is great to solve that issue
 
5:49 AM
@casey oh
 
there is an m4 macro to test for and find hdf5 and its necessary libraries to link with (using h5cc to get the library list), but it cannot deal with anything but the simplest install
which won't work for a lot of cases in real use
and cmake can't deal with finding hdf5 unless it has been installed with a specific cmake helper file
so instead I just use:
# HDF5 Support -------------------------------------------
# Uncomment the following lines to enable HDF5 support
# Make sure that the path in HDFINC points to your HDF5 install
HDFFLAGS=-DHDF5
HDFINC=-I/usr/include
HDFLIB=-lhdf5_fortran -lhdf5hl_fortran -lhdf5
 
@casey Fair enough. But that sort of thing doesn't scale very well. And taken to extremes puts a large burden on a potential user.
 
what makes hdf5 difficult to deal with in particular, and common to many programs with Fortran 90 bindings is that to use pre-compiled modules requires the Fortran compiler can find the requisite .mod file, and this file is compiler dependent. The response is that many installs will have hdf5 compiled against gfortran, ifort and pgf90 and you have to use the one corresponding to the compiler you are using to compile your code
@FaheemMitha its more common than you'd think. The users of my code are writing programs that make use of my code, so if they can't handle making a change in a makefile they have bigger problems
so I'm not worried about it.
and in any case, the HDF support is by default commented out and if you have gfortran installed a simple "cd src; make" will work out of the box
If i did move to autotools the method I'd use to tackle the issue would to require a `--with-hdf5=/path/to/hdf/install" flag passed to configure
and then I'd have to write a macro to test some things about it (e.g. is it compatible with $FC and what libraries do I need to add to the link to make it work e.g. -lsz)
so it would still require user intervention and knowledge to use
 
6:43 AM
@casey sure, autotools may not make sense in your case. but it is still widely used, and I think for good reason.
 
 
8 hours later…
2:18 PM
@terdon Hi, got your comment! I don't remember what flag you're talking about though...
 
@JohnWHSmith You flagged this as "Not an answer":
(you can't see it but it should take you to the question)
It looked like an answer to me so I dismissed the flag, then saw that it was repeating something already given stating "Yay! It worked for me". Anyway, I just felt bad about dismissing a valid flag and some people get worked up about things like that so I left you that message :)
 
@terdon that Q needs cleanup, I see some regurgitated answers
 
@Braiam Yes, on it now.
 
@terdon Oh yes, it was more like a duplicate answer though. But well, as long as it's been removed ;)
 
@Braiam is that you downvoting?
It was really weird, I watched each A get downvoted as I was reading it!
 
2:24 PM
@terdon there are answers that lists the amount of partitions not physical disks
 
@Braiam I know, I wasn't arguing against the downvote, just saying it was strange to watch it happen. I was reading just behind you and in fact deleting the As just after your downvote.
:)
 
just how many solutions using fdisk we need?
 
@Braiam all of them apparently...
 
The second one does a little bit of formatting through grep but the first still provides the correct solution.
 
I see community protected that Q. Would be nice if that had any real effect
 
2:29 PM
@casey It probably did, just imagine all the crap that was never posted.
 
@terdon 10 rep is an incredibly low barrier to entry. True it'll stop anon posts or brand new signups, but thats about it
and I'm not saying it shouldnt be protected, I would have done it myself if I had the rep to do so
 
I know, I know. I'm just assuming that even more crap would have been there without the protection.
 
I just wish it was more effective -- perhaps 50 rep would be a better threshold. Still low, but at least you've been trusted with commenting by that point.
I'm sure if I bothered to search meta.SE, I'd find this suggestion had been made before though
 
Hi guys. Terdon, how many flags does U&L get a day, approximately. Is there some easy way to get a total count?
 
@FaheemMitha ~16
 
2:36 PM
@Braiam How do you know?
 
10 hours ago, by slm
@Seth We've had 113 combined for the wk 8-)
113/7=way off my number
 
and those are likely flags requiring ♦ attention, not all flags (though this is probably what you were really after).
 
@casey Yes, I wanted all flags, though what is a flag that does not require attention?
 
close flags, dup flags don't go into the mod queue
 
Neither do VLQ flags I think.
 
2:46 PM
and some of the others, the just funnel questions into the review queues
yea, nor VLQ
 
@terdon they do
 
In iptables, :INPUT ACCEPT [527:57388] means allow open connections from ports 527 to 57388 right?
 
@Braiam You sure? I thought they went straight to the review. Hmm, no, I've handled some actually, I guess you're right.
 
comment flags and other(♦) flags go to the mods
 
> VLQ and NAA flags immediately enter the mod flag queue in all cases
but that probably will change
17
A: Final (binding) Votes to Close and Delete should trigger a revision to be saved

Shog9This strikes me as a good idea for moderator accountability - it's all too easy for malicious moderators such as yourself to hand-wave about the grace period right now, when in truth you know that the original answer was 20 paragraphs of beautifully-written prose and example code the like of whic...

why I keep linking to Shog??
 
2:56 PM
@Braiam but they also go into the review queues if the post is unreviewed and disappear from the mod queue if the community closes/deletes them.
 
@casey we were talking about what goes to the mod queue and what not, VLQ and NAA's does goes there, even if it's handled by the community ;)
 
@Braiam yes, but it is important distinction to make since other flags that go to the mod queue cannot be handled by the community
 
I'm not sure how that's relevant to the main question "what goes to the mod♦ queue?"
 
@Braiam well, if were going to get pedantic about it, I was answering:
19 mins ago, by Faheem Mitha
@casey Yes, I wanted all flags, though what is a flag that does not require attention?
and because the VLQ and NAA flags can be handled by the community, they do not require mod attention
 
naa?
 
3:03 PM
not an answer
 
not an answer
 
slm
3:16 PM
@casey protection should be at least 300-500 IMO. 10's a joke
> This strikes me as a good idea for moderator accountability - it's all too easy for malicious moderators such as yourself to hand-wave about the grace period right now, when in truth you know that the original answer was 20 paragraphs of beautifully-written prose and example code the like of which has never been seen before.
love the sarcasm
 
3:32 PM
"We’ll be using Ubuntu 10.04" which "I got the 2.6.32 kernel". I would use another newer guide. — Braiam 8 mins ago
why people keeps following +3 yr old guides and hopes it works exactly the same?
 
Because Google has been well-ranking them for longer, and people don't like going beyond the second results page.
 
Braiam that seems like a perfectly valid question. The OP wants to modify the kernel so we can assume they're not a newb and that guide is what they found. Neither of your comments is particularly helpful. Can you suggest a newer guide?
 
@terdon nope
 
slm
@JohnWHSmith I'd be happy if ppl would actually do a search and read the 1st page of results 8-)
 
@terdon I actually googled and got the first result csee.umbc.edu/courses/undergraduate/CMSC421/fall02/burt/…
that comment was helpful
 
3:42 PM
@slm too much expectation :)
 
slm
@Ramesh - yeah foolish me, use a search engine to look for solutions to my problems. I'm such a dumb @$$
 
@Braiam The one you just deleted is helpful, assuming that is a newer guide that would help the OP. Remember that 1) your google-fu might be stronger than theirs and 2) google's results depend on you google profile so you don't know if they would find it.
 
@slm we all are :)
 
The one I deleted just suggests using a newer guide without actually offering one. Since it came after the quite rude "Why do you need to know", I deleted it to stop things from escalating.
 
3:46 PM
@terdon "Why do you need to know" is not rude, it prevents XY situations
asking WHY is essential
 
@Braiam, I would probably phrase it as It would be better if you let people know why do you need it.
 
@Ramesh err... isn't obvious? to clarify the question?
> Use comments to ask for more information or suggest improvements.
 
@Braiam yeah, but some users might feel little offended if you plainly ask "Why do you need to know?". Again, I do not see it as rude but simply saying some users might perceive it as offensive.
 
@Braiam Well, yes, but this is a pretty specific case. "How can I add a new system call" is precise enough to make it unlikely that it's an XY problem. Unfortunately, you worded it very abruptly "Why you need to know" sounds like someone should demonstrate a right to the information. Next time, please use something like "Why do you need this?" or "What are you trying to do" etc. "Why you need to know" really sounds aggressive.
@Braiam why did you delete your other comment? The one giving a link to a guide? That one does seem useful.
(and I undeleted the one I deleted, you're right, I shouldn't have. Sorry.)
 
@terdon because it's a guide to the 2.2 kernel
alhail SO
 
3:53 PM
Ah.
So, finding a newer one is harder than it seems huh?
 
yeah, just added kernel 3.3 to the search
 
Nice, +1 on the answer.
 
hah, just came across a Q on SO with someone trying to do this in Fortran:
> open(unit=unit,file=\"pwd -P/$(OUTDIR)/\"//trim(filename))
 
shivers
 
trying to put shell script into what ends up being a system call
no wonder it doesn't work....
 
4:07 PM
I actually find it quite difficult to find good references for kernels after 2.6... It isn't too difficult when you're looking for something specific, but for general documentation, that's tricky...
 
@slm any ideas on this? CentOS stuff.
2
Q: CentOS 7 boots too fast and network is not ready when executing cron scripts

DHSI have just upgraded from CentOS 6.5 to 7.0 and I am not too happy as the new systemd is probably giving me problems. It seems it is just simply booting too fast, starting up processes asynchronously and screwing up service dependencies. For example I have a few scripts setup in crond which are ...

What is the equivalent of /etc/default/rcS for CentOS? That should be there right?
 
slm
@JohnWHSmith Yeah it can be tricky to find things, but I've amassed a lot of points by doing nothing more than taking either fragments of the OP's Q and searched for it or taken the error msgs or error codes and searched for them. 8-)
@terdon centos 7 uses systemd now I believe
 
@slm So? I am not up to snuff on the init systems, does that mean that /etc/default/rcS won't exist? And yes, the OP states it's on systemd.
So am I though (I think at least) and I have that file.
 
slm
centos doesn't have that dir. The /etc/default is typically something different on RH distros. There are /etc/rc* dirs, and systemd can use those and act as init would in the past.
 
@terdon some systemd implementations still have sysvinit stuff laying around to either fallback on or act as some compatibility layer for the transition
 
4:14 PM
And is /etc/default a sysvinit thing?
I just saw in my README that at some point I added CONCURRENCY=shell to /etc/default/rcS and that was supposed to allow services to be started in parallel. I was assuming that there is something similar going on on the OP's system.
 
slm
so far w/ systemd you could do either directly through systemd or still use rc* dirs.
 
I commented to that user to add Requires=network.target into the units
my understanding is that After= will set ordering if multiple units are being started
and that Requires= triggers loading of the unit as a dependency
 
so its best to put the unit you depend on in both, to a) trigger it being loaded if it isnt and 2) to order itself after that unit
 
3) allow programs to fallback if there isn't network available and wait for it
 
4:28 PM
yep
 
Why are we suddenly getting questions like this unix.stackexchange.com/questions/165265/… ... seems like that belongs elsewhere. Stack Overflow, maybe.
 
@derobert No code.
Also, someone or other made a point on meta for making such things on topic:
1
A: Is the Unix C API still on-topic?

derobertSmall amounts of certain types of programming are reasonably within the scope of what a system administrator does. Uncontroversially, writing a shell script to start a daemon, check if its running, and stop it is within scope (i.e., an init script). So is a shell scripting to automate system adm...

 
@derobert I would go with too broad
 
that question is nowhere near unix C api
 
yeah, but if we migrate it to SO they will kick it back
 
4:32 PM
@casey True, but it is about configuring/modifying the kernel. I feel that should be on topic.
 
@terdon a kernel, and not specific to any OS
 
Still, I wallow in extreme ignorance about the topic though so if you guys disagree, I won't press it.
 
its a low level systems programming / OS design question
 
"explain this <several lines of code here> code to me" shouldn't be the kind of Q's we should cater
"what does -F in awk" is ok
 
@terdon Yep, but that didn't seem to gain consensus...
 
4:35 PM
Yeah, that kind of thing goes way over my head so I can't really weigh in.
 
Files should not be compared by their content by intention. What does the OP mean?
 
Hey @Gilles, I notice the AIX tag has been removed from my question. I was thinking it might make more sense to keep it. People googling for help replacing newlines in AIX will discover many ways to do it that won't work because of the peculiarities of AIX, and then subsequently include AIX in the search.
3
Q: In AIX, how can I replace a newline with sed?

BasilI've got text with whitespace (including a newline) that I need to replace. I'd like to use sed, but AIX doesn't seem to support the answers to sed newline questions I found. The version of bash I'm using is "GNU bash, version 3.00.16(1)-release (powerpc-ibm-aix5.1)" If I run a command that outp...

 
@Basil Your question isn't specific to AIX. It applies to any OS with sed.
We only use OS tags for questions that are specific to one OS
 
cue tag
 
4:37 PM
@terdon The first one is something you'd answer by pulling out an ix86 microprocessor manual (the very low level one, that tells you things like what all the opcodes do)... the second appears to be about programming a new syscall into the kernel.
 
There's no peculiarity of AIX here, only the lack of GNU utilities, which applies to everything except non-embedded Linux and Cygwin
 
I don't really know much about different kinds of unix, but my experience trying to find an answer that worked was pretty constant- if I left out AIX, I found people getting help that worked for them and not for me
 
@derobert we aren't suddenly getting them, we've been getting them for ages
 
if I searched for the same thing with AIX, I found just questions, no working answers
 
@Basil instead of mentioning AIX, mention in your Q the version of sed you are using
 
4:39 PM
@casey How do I get that?
 
@Basil Most of the solution were probably GNU stuff. Also, AIX is in the title and the body of the question, searches will find it just fine.
 
@casey *clap* *clap* *clap* *clap*
 
@Basil do say that you're using AIX in the body of your question, and even insist that this means that you don't have GNU tools. But don't use the tag.
 
@terdon Fair enough.
 
sed --version
 
4:40 PM
@casey The version of sed is “sed from AIX <AIX version>”
 
% sed --version
GNU sed version 4.2.1
 
@terdon no, that would only work for GNU sed
 
heh
 
Oh, well then, case in point :)
 
was about to post the error message
 
4:40 PM
man sed, find the right flag for your sed
 
doesn't seem to have the word "version" in the man page
 
@casey why do you assume there is one?
the version of sed is the version of AIX.
 
meh, just go the regex rute
 
@Braiam If you're talking about my problem, the only way to use regex that I know of in AIX is sed, perl, or slightly in grep
 
@Gilles its been a few years since I used AIX regularly and perhaps my memory fails me in that assumption. No access to test currently.
 
4:42 PM
@casey Count your blessings :\
 
@Basil since this is AIX, can we assume you don't even have perl installed?
 
@terdon Actually I do. It's an older version, but we do have it
Of course, I could solve this with perl very easily. I was hoping for a bash solution
 
@slm, did you hear back from Linus?
 
@Basil perl -pe 's/\n//' file
@Basil The best is tr -d '\n' of course.
 
just that in that case is replacing something with newline, not newline with something
last post is the most important of all
 
4:45 PM
@terdon That gets rid of all the newlines- I was hoping to just remove the first one (and possibly the whitespace before and after it)
 
> Early versions of 'sed' (and implementations that copied them) did not originally accept escaped characters in the Right-Hand-Side (RHS).
 
Yes, sorry, just read your whole question. Gimme a sec, I'll post an answer.
 
I've only used linux a handful of times, but man, I wish AIX was more like that...
 
@Braiam: It's not relevant, since when @Basil had run the sed version, and the newline was replaced
@Basil: If you can use perl, you can use:

perl -pe 's/\n//; END {printf "\n"}'
 
@cuonglm yes, I know how to solve it in perl. I am looking to stick with bash, though- only my LPAR has perl installed, and I would like to know how this is done when you can't arbitrarily change the AIX installation
bash or ksh would be ok
 
4:50 PM
@Basil command | while read i; do printf "%s " $i; done; echo
 
@cuonglm he's asking how, I'm presuming he's trying but is not getting the desired results
 
@Braiam: But the reason he doesn't get the desired results, because sed had replaced all newlines, instead of skip the last one. I only mean the one you said "Early versions of 'sed' (and implementations that copied them) did not originally accept escaped characters in the Right-Hand-Side (RHS)." is not the reason.
 
@terdon That has the same effect as tr- I'm trying to do what would be a sed replace without the /g, just replacing the first newline. This printf replaces all of them :)
@cuonglm Just to clarify, I can't get sed to replace any newlines :)
Oh wait, I can actually!
@Gilles's answer showed me how to do that :) sed -e ':1' -e 'N' -e '$!b1' -e 's/\n/ /g'
 
@Basil That's why I added the echo to the end.
 
@Basil edit your question with what you tried and the result, so we can know exactly where the problem lies
 
4:55 PM
@Basil: of course sed will, it's only strange that the $!b1 part seems not to applied with the last line
 
slm
@Ramesh nope
 
@Braiam Done, thanks for the reminder
@terdon it adds a single newline to the end of the stream, but I can get an arbitrary number of aliases back from the command.
 
@Basil Ah, that's completely different then. It sounds like you want to remove the first of each pair of newlines then, is that correct? If so, please edit and make it clear.
Damn, too used to writing [edit].
 
@terdon Just did, I realize now that it was unclear. Sorry :)
Is there a way to make that series of sed -e commands match a tab-newline-tab instead of just a newline?
 
5:12 PM
@Basil: What the result when you run this:

ssh localhost "echo -e 'a\nb'" | sed -e ':1' -e 'N' -e '$!b1' -e 's/\n/ /g'
 
I want to add system calls in the linux-3.13.3 kernel. Help me
 
@cuonglm -e a b
@cuonglm I think the main issue now is how to only take the first newline
 
Try this:

sed -e ':1' -e 'N' -e '$!b1' -e 's/\n/ /'
 
@cuonglm also -e a b
@cuonglm but when I tried that with the actual (not localhost echo) command, it only replaced the first newline and none of the rest of them
 
@Basil: Do you have od
Can you run:

ssh localhost "echo -e 'a\nb'" | od -c
 
5:18 PM
0000000 - e a \n b \n
0000007
 
Did you run it with server or localhost?
 
@cuonglm localhost
 
Can you try this with server?
 
@Basil updated my answer. How about:
    command |
  while IFS="\n" read i; do
    let c++; [ "$(expr $c % 2)" -eq "0" ] && echo "$i" || printf "%s " "$i";
  done
Advantages: pure bash, should be completely portable. Disadvantages: getting kind of complex.
 
@terdon that worked :)
 
5:27 PM
Yay! Ugly thing though.
 
heh, I'll take it
"ugly but it works" should be on the t-shirts IBM hands out to the sysadmins here...
 
5:39 PM
im so happy the ibm powered supercomputer I use has linux on it rather than AIX.
all my stuff "just works", which surprisingly includes my python 3 scripts which I was dreading having to backport to 2.6 or 2.7
 
@casey That's their go-to tool for some things, but probably not SAP...
 
though battling with the LSF job management software can be fun at times....
oh, my job is finally running after waiting for hours! oh wait, something wasn't right in the submission script and it aborted...
 
@casey Worse: My job crashed in the middle after running for5 days. Error message: segfault.
 
5:55 PM
@terdon oh don't even get my started on having to submit to the "interactive" queue (billed at 1.5x !) to even be able to debug an MPI program. That is the major reason I took the pains to setup a full MPI environment on my box so I can do mess with optimizing, profiling and debugging...
 
Yeah, I had a partitioning algorithm that would cover a graph with a system of classes that sometimes (not always) threw a segfault half way through. The damn thing took days to finish and usually worked if I just relaunched it. I ended up making my job command || command.
 
haha, nice
though on the system I'm on means you'd have to declare 2x wallclock time to ensure it could run before being killed and then having to wait for a window that big to get your runtime
I'm constantly weighing the issue "should I ask for 4 nodes @ 4:00 or 32 nodes @ 0:30"
 
00:00 - 18:0018:00 - 00:00

« first day (1483 days earlier)      last day (3454 days later) »