is this the common denominator on makefiles?

@terdon ok

@Braiam that doesn't look like a simple Makefile to me

ok, this is morronic, how I'm supposed to append non-standard path for libraries, binaries, etc. to pass configure?

and they are hardcoded, wtf

@Braiam are you using autotools? And this this your project, or 3rd party?

What is hardcoded, and where?

@FaheemMitha I think they are using them... at least is required to have it installed, and no, apache's

@FaheemMitha the paths for dependency libraries of apache that are part of apache in the configure script

@Braiam I see. Well, autotools can't include everything. I suppose you could patch configure.ac.

And then create configure and whatnot.

Why are you looking for things in non-standard places, anyway?

And for that matter, why are you trying to install apache from source?

Damn, too late to flag the spam.

We seem to be getting more of those recently. Or am I wrong?

@FaheemMitha because I'm not going to install the 6 dependencies on my system

and I was checking if I could build old stuff with newer stuff... well, you can't

@Braiam I'm not sure what you mean. Dependencies of Apache? And why not?

@FaheemMitha dependencies of the C++ framework of this stuff uima.apache.org, and why should I?

@Braiam Oh, not the web server then. And if it is available in Debian, it's less hassle than installing from source.

But I'm not sure exactly what you are trying to do...

@Braiam if it is autotools, configure should respect environment vars. You can try setting CFLAGS and LDFLAGS to pass options to the compiler and linker

e.g. CFLAGS=-I/some/nonstandard/include LDFLAGS="-L/path/to/lib -lblah" ./configure

@casey That sounds like a good idea. But you want to add it, not overwrite it.

@FaheemMitha those will get added to whatever the generated makefile would want to add

@casey nope

CFLAGS=-I/home/braiam/src/uimacpp/build/ LDFLAGS="-L/home/braiam/src/uimacpp/build/" ./configure
checking for Xerces C++ Parser... no
configure: WARNING: XERCES not found
The Xerces C++ library cannot be found.
Please install XERCES on this system and supply the appropriate
--with-xerces option to 'configure'
configure: error: no suitable XERCES found
CFLAGS=-I/home/braiam/src/uimacpp/build/: command not found

actually, it works, just not for those libraries/dependencies

@Braiam have you tried --with-xerces=/path/to/xerces ?

@casey it's not pretty ./configure --prefix /home/braiam/src/uimacpp/build --with-xerces=/home/braiam/src/uimacpp/build --with-apr=/home/braiam/src/uimacpp/build --with-icu=/home/braiam/src/uimacpp/build --with-activemq=/home/braiam/src/uimacpp/build --with-apr-util=/home/braiam/src/uimacpp/build

and yes, that's the only way it works, because hardcode

and xerces is populated into /home/braiam/src/uimacpp/build/lib,/bin,/include (or whatever relative paths it uses)?

yup

what is in the configure log? One of its files will have the log of the check and the program test that failed

just grep for "Please install XERCES" and you should find the failed test

I forget what file it'll be in offhand

configure:15516: Xerces C++ library configuration
configure:15721: checking for Xerces C++ Parser
configure:15801: result: no
configure:15811: WARNING: XERCES not found
configure:15817: error: no suitable XERCES found

what are you trying to build?

the C++ framework itself

you have a link to the specific thing you are working with? I see a lot of acronyms and java, but I'm missing the C++ stuff

i found it

@casey It does? Ok.

@Braiam why not just build and install those other things in standard places? then you won't have a problem. depends what you are trying to do, but making Debian packages for this sort of stuff is not hard.

@FaheemMitha is adhoc, I'm not trying to install stuff on my system, just figuring out if a newer library can be used instead of the old one

@Braiam I see. Ok.

@Braiam did you config.log have lines like

This is to answer a question, then?

there should be the contents of conftest.cpp along with the error if it wasn't found (I was hoping for that, but apparently I do have xerces)

Yes, I see it is.

@casey O_O

@casey and nope

@Braiam well, you can install different versions side by side if they are set up to allow you to do so. But I've never done that, so don't know the details.

@FaheemMitha they need different package name, otherwise apt will install the later (unless it breaks something else)

@Braiam Yes, they need different names. Also, the library files need to be different, of course. I guess the sonames need to be different or something.

177                 AC_LANG_PROGRAM(
178                     [[
179 @%:@include <xercesc/util/XercesDefs.hpp>
180 @%:@include <xercesc/util/PlatformUtils.hpp>
181 #if (_XERCES_VERSION >= 20200)
182 XERCES_CPP_NAMESPACE_USE
183 #endif
184                     ]],

@Braiam that is the text program that gets run

@casey I... have no idea then why -I doesn't work...

@Braiam try setting CPPFLAGS instead of CFLAGS

the test uses that

no joy either

if you supply a CPPFLAGS="blah" and --with-xerces=foobar, it'll put "$CPPFLAGS -Ifoobar/include -Ifoobar/include/xercesc" into the test

for the header check at least

according to the test you can also do

you can try that and explicitly give the include and library paths

yeah, but I also need to do the same with the other 4 dependencies, and it gets silly when autoconf should have been covering that

@Braiam autoconf is great when things are in your system search paths or /usr/local

when not, you need to give it hints

I do give it hints, but he apparently ignores them

it uses them, I'm not sure why its not working for you

see m4/find_xerces.m4

that is the file with the macro that does the test

i assume xerces is properly built and installed into the prefix you are using?

yes, it configure correctly with the --with thingy

where the macro test the environment variable?

134 CPPFLAGS="$CPPFLAGS -I$xerces_include_dir -I$xerces_include_dir2"

@casey nope

found why: it runs that if run_xerces_test is yes, run_xerces_test is yes only if xerces_prefix is not empty or --with-xerces is given, and xerces_prefix only tests /usr and /usr/local, so the test is rilled

@Braiam rilled?

@FaheemMitha set up to fail or success without an actual fair try

@Braiam not familiar with that word.

the elections have been rilled

or something like that

A more common word would be rigged. Is that what you meant, perhaps?

probably, through the definition don't seems to catch the meaning I'm trying to give

aren't build systems fun...

yep

so my assertion was correct?

I was just wrestling one that uses a non-standard homebrewed script named "configure" and uses ymake to regenerate makefiles on thy fly from a collection of files defining system specific crap. The Makefiles don't stop on error, so you have to carefully make sure things worked, and when they don't you have to figure out which obscure template file you need to put an undocumented define in that you only knew about because a default gets set in some other template, then rebuild it all

@casey that sounds like a massive PITA, and I only understood like half of the barbarities were there

once i figured out the peculiarities of the build system it was quite easy to do what I needed, but getting there was a challenge

@casey if that is some academic software, terrible engineering is the norm.

@FaheemMitha semi. probably half written by scientists half by CS people, but with lots of ancient legacy cruft

NCL/NCAR graphics

I've seen much, much worse though

visit or vapor (I can't remember which) is particularly bad

I don't understand why people don't just use one of the standard build systems.

@casey that's what version control is for.

@Braiam Yes, we all know xkcd. And yes, I've seen that one before.

@FaheemMitha the base templates are in source control, but there isn't much point to telling them to commit your local computer speicifc options into the trunk

But really there are only two standard things - autotools and cmake.

and when you get down to it, it is nice to set a variable in one file and have it cascade to the hundreds of makefiles below it

@casey Well, no, but I just mean they would be safe locally. If you committed them locally.

@FaheemMitha no, I'd commit them and when I ran make, they would be overwritten

the makefiles first regenerate themselves, then compile stuff

@casey Ok, but if you committed them, you could get them back. That was my sole point.

Your changes, I mean.

@FaheemMitha yes, but that wouldn't help

@casey Yes, that sounds nutty.

@casey In this case, apparently not.

Homegrown build systems are A Bad Idea.

@FaheemMitha this softwares origins date back to the mid 1960's

even autotools isn't that ancient

@casey wow

@casey True, it isn't.

But switching to a modern build system isn't that big a deal.

although "1.0" was in 1986

What is it called again? link?

@FaheemMitha ncl.ucar.edu/index.shtml

@casey thanks

I imagine sticking with ymake is a combination of inertia and supporting builds on Linux, Cygwin, Darwin, FreeBSD, DEC UNIX, AIX, Cray, SGI, Solaris, SystemV UNIX and more

I wonder how much it would cost to scrap all of that and using a more sane solution...

@casey Well, they could hire someone to switch it. Wouldn't be that hard.

eh, it works and probably 99% of users just install the binaries built for their enterprise distro

you probably aren't compiling it yourself unless you aren't running one of the bigger platforms or you want to link against something besides the reference BLAS/LAPACK implementation or want something non-standard

@casey Probably. But good enough is usually the primrose path to terrible software.

And most academic stuff is there.

I only build it because the binaries are built against redhat generally won't dynamically link on my system because my libraries are newer

well upon inspection they do have builds for a number of systems, just not mine :)

@casey oh the irony

@Braiam ?

@casey there's a flavor for everyone, but not the one you like

Writing good software is about endless attention to details. Academics think if they can make a plot with it, they're done.

@Braiam well, i run a source distro, so they do have my flavor :)

@casey A DIY flavor. Like those desserts made to order.

@Anthon ok now let's clear redundant comments

@Pandya Do you want to paste into the same terminal or somewhere else?

@Pandya in bash you can use emacs bindings to move/copy/paste

e.g. move to start of selection, C-SPC, move to end of selection, C-w to cut, M-w to copy, C-y to paste

@StéphaneChazelas - i really was in earnest when i asked you to edit that xor thing - i don't want to misrepresent it. I think the standard math expansion is too frequently discounted - or perhaps too little understood - or something. I do not desire to contribute to that, because, as i think, it definitely me far too long to understand its capability (if i even do yet). While the expression is immediately sensible to me, i'm willing to concede that's not saying much.

Or, if you just took that -a/-o quote out of mine, i could just delete it, i guess.

@terdon Many upvoted comments under meta.stackexchange.com/a/3021/139866 suggest leaving/allowing "Thanks". FWIW I'll delete a signature or a greeting from a question, but tolerate a thank you (a harmless nicety).

@ChrisW 681 vs ~100+21 doesn't hold out

and in comments there are very compelling reasons to remove them

@Braiam The 600-odd votes were in favour of editing out "greetings and salutations".

@ChrisW I think that "along with other extraneous clutter" is inclusive of other pleasantries

Anyway I'm one of the people who tolerates this OP's "Thank you" on Buddhism.SE -- if you want to, you could consider taking less of a hard line against a closing courtesy (e.g. Thank you for your time) than you would against opening verbiage (Hello!) and/or a redundant signature (--Username) ... the meta-post you referenced would support that too, imo.

@ChrisW Hi, thanks for chiming in. I wanted to ask you about this earlier.

As far as I know, the general stance on all SE sites is that "thanks" is just noise. It distracts from the main point of the question and is just not needed. In this particular case, the user had rolled back three edits from 2 different users who removed the phrase.

The culture in most SE sites, certainly those I hang out in, is quite clearly against them. We even have a user here who considers it rude as he takes them as a sign that the OP ignored the site's guidelines.

Personally, I wouldn't go that far, but I agree that they don't give anything useful to a question.

I will rarely if ever edit just to remove those, but I will remove them from a post I'm editing anyway.

Oh, and I have absolutely nothing against the user in question. He had a different opinion, defended it with arguments and then accepted the community's guidelines. I wish more users were like him/her. I hope they didn't feel I was attacking them, it was certainly not my intent.

@ChrisW also, SE actually removes greetings automatically. Jeff's answer here suggests that the only reason they don't also remove thanks is that they're harder to identify.

@mikeserv I would update your answer for obvious technical, errors, typos or clarification. But here, it's about a choice of approach. We have different views (and that's a good thing, diversity is a good thing, a lot of your answers show us a different way to look at things) so I can't really apply my views in your answer.

@terdon also, somehow people manages to do atrocities with the language

where I was when this comment was posted?

@Braiam lol, yeah, I remember that one :)

so much time without cat pics on this room

@StéphaneChazelas - dang, man. Thanks. that means a lot. I'm good with it either way.

@StéphaneChazelas - For what it's worth, every time i get curious about some new thing, and start banging my head on it, inevitably i stumble on an sc somewhere while searching for information. It happens every time.

@terdon No that user didn't complain or anything; it was just me, who happened to notice your interchange because it was a "hot network question". I like that user too (and "wish more users were like him/her") and if they want to end their questions like that, that's OK, imo. I read the meta-topic you referenced and IMO it supports that (permitting "thank you") as a possibility (many in favour of "thank you"s, with the big votes being more explicitly against salutations and signatures).

@StéphaneChazelas - i guess what i mean is, if ive got weird ideas, it's probably at least a little bit your fault. and thanks for that, too.

> RR1.3 introduced panning. Without panning, the visible area of the CRTC is exactly defined by the mode, rotation, transformation and offset. Basically nvidia have only implemented half of the extension. They allow you to set panning, but don't report it back to the application.

well done Nvidia

oh, btw, don't upgrade chromium on testing, they introduced a bashism on a sh script

hey guys quick question. Is this the right place to ask a question about vim?

@AlanChavez if you want to know which is better between vim and emacs, the answer is obviously emacs

@Braiam lol :) I disagree. I had a question about an automatic search/replace in vim, but I think I spotted the mistake in my regex.

@AlanChavez vi.stackexchange.com , though you can probably get quick help on SO as well as here

ah that site is handy :) Thanks for pointing me out to the right place.

@Braiam evil-mode :)

@ChrisW It's not just that one. There are various posts on meta about this subject. The general consensus is quite clear. We want to build a repository of knowledge here. "Hi" and "Thanks" don't add to that. As I said though, I wouldn't edit only to remove them.

Personally, I also just don't like "Thank you for your time". It makes me feel as though this is a call center.

I was thinking I saw everything after chef, pickles and cucumbers... but reactive-banana?

@ChrisW oh, and check out the post again. There have since been 4 upvotes on my comment about thanks. I'm not alone :)

@terdon You're not, and I see that behaviour is justifiable. The post is empty of comments now.

Ah, good. I was planning to flag'em.

bugs.debian.org/cgi-bin/bugreport.cgi?bug=789310

Do people have problems with user switching on MATE/GDM?

> Be polite, but no salutations

@Braiam Is that directed at me?

@c0xc no

@c0xc on the other hand, instead of asking if people is having problems with it, try to describe your problem, find the logs of gdm (or whatever mate uses) and post a question on the site, it will be more efficient that way and, at the same time, you may get an answer

Well that's true, I wasn't very specific, was I. I've had GDM start a new session when trying to unlock a user who's already logged in. That's one. Another one would be that I'm asked for my password twice (when unlocking works), after unlocking in GDM, MATE asks for the password. Do other DE/dm combinations work better, without those issues?