gildor changed the topic of #ocaml to: Discussions about the OCaml programming language | http://caml.inria.fr/ | OCaml 3.12.1 http://bit.ly/nNVIVH
dsheets has quit [Remote host closed the connection]
dsheets has joined #ocaml
Submarine has joined #ocaml
Submarine has quit [Changing host]
Submarine has joined #ocaml
gasche_ has quit [Remote host closed the connection]
gnuvince has joined #ocaml
asmanur_ has quit [Ping timeout: 244 seconds]
asmanur has joined #ocaml
Reventlov has quit [Ping timeout: 244 seconds]
Ptival_ has joined #ocaml
Ptival has quit [Ping timeout: 240 seconds]
patronus has quit [Ping timeout: 244 seconds]
patronus_ has joined #ocaml
contempt has quit [Ping timeout: 244 seconds]
Reventlov has joined #ocaml
Haseo has quit [Ping timeout: 244 seconds]
snarkyboojum has quit [Ping timeout: 244 seconds]
Reventlov is now known as Guest9164
snarkyboojum has joined #ocaml
contempt has joined #ocaml
milosn has quit [Ping timeout: 240 seconds]
Haseo has joined #ocaml
asmanur has quit [Ping timeout: 244 seconds]
asmanur has joined #ocaml
milosn has joined #ocaml
diml has quit [*.net *.split]
mfp has quit [*.net *.split]
nimred has quit [*.net *.split]
alxbl has quit [*.net *.split]
hcarty has quit [*.net *.split]
lusory has quit [*.net *.split]
srcerer has quit [*.net *.split]
diml has joined #ocaml
mfp has joined #ocaml
nimred has joined #ocaml
alxbl has joined #ocaml
hcarty has joined #ocaml
lusory has joined #ocaml
emmanuelux has joined #ocaml
nimred has quit [Max SendQ exceeded]
nimred has joined #ocaml
nimred has quit [Changing host]
nimred has joined #ocaml
emmanuelux has quit [Remote host closed the connection]
eikke has joined #ocaml
eikke has quit [Ping timeout: 246 seconds]
Submarine has quit [Ping timeout: 265 seconds]
Znudzon has quit [Ping timeout: 248 seconds]
Znudzon has joined #ocaml
ivan\ has quit [Quit: ERC Version 5.3 (IRC client for Emacs)]
Znudzon has quit [Ping timeout: 265 seconds]
Znudzon has joined #ocaml
ivan\ has joined #ocaml
ankit9 has joined #ocaml
ankit9 has quit [Ping timeout: 248 seconds]
ankit9 has joined #ocaml
cdidd has joined #ocaml
Znudzon_ has joined #ocaml
eni has joined #ocaml
thomasga has joined #ocaml
pango is now known as pangoafk
thomasga has quit [Quit: Leaving.]
Ptival_ is now known as Ptival
edwin has joined #ocaml
jaar has joined #ocaml
djcoin has joined #ocaml
cago has joined #ocaml
mika1 has joined #ocaml
ftrvxmtrx has quit [Quit: Leaving]
thomasga has joined #ocaml
mcstar has joined #ocaml
Cyanure has joined #ocaml
ankit9 has quit [Ping timeout: 265 seconds]
silver has joined #ocaml
edwin has left #ocaml []
<mcstar> ive got this profiling on a little ocaml program
<mcstar> it generates 10M random numbers by 1000 elements(in an array) from the gsl library
<mcstar> and sums up the elements
<mcstar> it is totally unbelievable to me, that summing up an array takes more time than generating its contents
<mcstar> the vertical stuff on the right side is the rng generation
<mcstar> the other side is from the summing
<adrien> it doesn't say much, and without the code it says nothing
<mcstar> it says much actually
<mcstar> and ofc i can provide the code
<mcstar> if i sum up the array by different means, i get slightly slower speed
<adrien> why do you use recursivity when dealing with arrays?
<adrien> of course you can but it's not the best way here
<f[x]> mcstar, where is that profile output from?
<mcstar> f[x]: you mean the data or the picture?
mort___ has joined #ocaml
<mcstar> adrien: there shouldnt be any difference from a simple recursion or an iteration
<mcstar> the recursion pattern is iterative
<f[x]> the data and the picture :)
<mcstar> f[x]: callgrind + kcachegrind
<f[x]> that's what I thought
<adrien> mcstar: there will since you're at least checking the bounds each time
<f[x]> mind that rng generation is done in C code
<mcstar> no, its fortran
<f[x]> whatever
<mcstar> :)
<f[x]> it is optimized
<f[x]> and no allocations
<adrien> and does it in batches
<mcstar> ofc, but it is much more work than summing up an array
<mcstar> and the GC isnt involved, there is only 1 array
<f[x]> caml_modify and page_table_lookup means that acc probably was not hoisted in register
<adrien> you're not simply summing the array, you're doing more work than that
<f[x]> need to look at asm
<adrien> yeah, plus using a reference wasn't a good idea
<mcstar> adrien: what exatly am i doing then?
<adrien> if you use a ref, use a for-loop
<adrien> if you use recursion, pass the value as an argument to the recursive call
<mcstar> that wasnt faster
<mcstar> i tried it before
<adrien> a for-loop wasn't faster?
<mcstar> no, using arguments in the recursion instead of reference
<mcstar> ill try a for loop then
<f[x]> hm, it allocates on each iteration
<mcstar> adrien: it is noticeably slower with for loops
<mcstar> 0.8 -> 1.0 sec
<mcstar> roughly
<mcstar> oops sorry
<mcstar> i compiled with -g accidentally
<mcstar> 0.453
mort___ has quit [Ping timeout: 248 seconds]
<mcstar> but how can there be such a big difference between iteration and recursion?
<f[x]> -g shouldn't influence speed!
<f[x]> mcstar, more optimizations kick in
<f[x]> try also -unsafe
<f[x]> the rule of thumb : iteration for imperative style, recursion for immutable
<mcstar> this is how it looks like
<mcstar> it is still too slow
<mcstar> f[x]: -unsafe -noassert already applied
<mcstar> -inline 1000 -ccopt -O9 -unsafe -noassert
<f[x]> -ccopt gives you nothing
<mcstar> w/ -g real 0m0.907s real wo -g 0m0.463s
<mcstar> f[x]: -ccopt -O9
<mcstar> thats 1 option
<f[x]> I know
<mcstar> and i read from an optimization guide, that it helps
<f[x]> it is useless - as you don't have C bindings
<mcstar> i see
<mcstar> i thought, ocamlopt might compile to c or something
<f[x]> and the only C code used during compilation is some runtime initialization
<f[x]> no, it doesn't
<mcstar> ok, 10ms difference consistently
<adrien> by passing another argument to the recursive calls instead of using a reference, timing went from 2.65s to 1.75s
<adrien> (not using -unsafe in any case)
<mcstar> adrien: 1 recursion for generating the arrays, and 1 for summing it up?
<adrien> no
<adrien> millions of mutations
<adrien> and possibly boxing (no idea about that but it's not impossible)
<adrien> that's what I've avoided
<mcstar> ok, but how
<mcstar> maybe paste your code
<nicoo> adrien: References are boxed, IIRC, so it's not surprising that arg-passing is faster
beckerb has joined #ocaml
<mcstar> adrien: why did you say no to: "1 recursion for generating the arrays, and 1 for summing it up?"
<mcstar> thats exactly what you are doing
<mcstar> adrien: your code runs at the same speed as my double for
<mcstar> with a reference!
<adrien> well, the same as your code
<adrien> my code runs twice as fast as the one you've first posted
<mcstar> yes
<mcstar> but since you told me to use for
<mcstar> so i did
<mcstar> still, as you can see from the profiling, too much time is spent on summing the array
<mcstar> very disappointing, really
<mcstar> whoa
<mcstar> adrien: the proportions in the profiling of my for loops and your recursion to the rng generation is very different
<adrien> also, your profiling lies
<adrien> you're profiling the ocaml code and not the C one
<adrien> so the ocaml code takes a huge (at least a factor of 2) performance hit
<adrien> while the C code takes nothing
<mcstar> why is that a lie?
<mcstar> the rng code takes the same time
<mcstar> and it doesnt change
<adrien> yes
<adrien> it doesn't get slower because it's not profiled
<adrien> but the ocaml code does
<mcstar> i dont see a problem
<mcstar> anyway, your recursive version, doesnt get slower when i compile with -g
<adrien> -g doesn't incur a noticeable performance hit
<adrien> -p does
<adrien> but only for the ocaml code
<mcstar> adrien: the for loope version gets a 2x performance hit from -g
<mcstar> as i explained before
<adrien> the ocaml code gets slower, the C code doesn't get slower during profiling
<adrien> so your profile doesn't show what actually happens
<f[x]> mcstar, what ocaml version?
<adrien> because for 3.12.1 at least, -g changes _nothing_
<mcstar> 3.12.1
<mcstar> what do you mean?
<f[x]> adrien, using pmp for profiling I get 50-50 (generation vs summing) for original code
<mcstar> am i imagining this you think?
<f[x]> mcstar, no, we just try to notice strange things
<f[x]> because that looks like a bug
<mcstar> 0.449 vs 0.998 seconds, wo, and w/ -g
<mcstar> i removed all the optimization flags
<mcstar> the 2x difference still holds
<f[x]> I get 12s -> 7s when using recursion
<f[x]> recursion get rids of caml_modify
<mcstar> looks better indeed
<mcstar> but i dont get an improvement over the 'for loop' version
<mcstar> in performance
<f[x]> that's expected
<f[x]> they should be identical in speed :)
Snark has joined #ocaml
<mcstar> well
<mcstar> actually it is indeed consistent
<mcstar> if with -g my code slows down 2x
* f[x] tries 4.00 now
<mcstar> then it explains the difference in the proportions in the profiling wrt the rng generation
Znudzon_ has quit [Ping timeout: 265 seconds]
<mcstar> so, adrien 's code shows, whats more close to the real code performance
<mcstar> 70% in gsl, 30% in ocaml
<mcstar> the problem is, that if i add other operations on the random numbers, they will slow down again
<adrien> you cannot compare the two
<adrien> not with profiling
<adrien> with printf, yes, with profiling no
<mcstar> printf?
<mcstar> its a damn shame, that profiling has such a great impact on a perfectly normal code
<mcstar> but whatever
<f[x]> use pmp
<flux> pmp?
<flux> maybe oprofile could be used with ocamlopt compiled programs?
<mcstar> whats pmp?
<adrien> poor man's profiler
Guest9164 has quit [Quit: leaving]
Reventlov has joined #ocaml
<f[x]> I don't see any significant difference when compiling with -g and without
<f[x]> mcstar, what is your OS and cpu?
<f[x]> can you share asm files for cases w/ and wo/ -g ?
<mcstar> latest linux kernel, x86_64 amd x2
<mcstar> f[x]: if you tell me how to get them..
<f[x]> ocamlopt -S
<mcstar> 3.4.4-2-ARCH
<mcstar> f[x]: http://sprunge.us/GLFd
<mcstar> w/ and wo
<mcstar> 2nd one starts at 184
Progster has joined #ocaml
mmajchrzak has joined #ocaml
ftrvxmtrx has joined #ocaml
avsm has joined #ocaml
eni has quit [Ping timeout: 246 seconds]
eikke has joined #ocaml
<mcstar> f[x]: have you found out anything?
<f[x]> I don't get difference for -g and no -g
<f[x]> second - I remove gsl out of the equation
<f[x]> and I get consistently original code and for loop 5 times slower than recursion
<f[x]> at this step I've got no more time to dig further
<f[x]> kept the asm to investigate
chambart has quit [Ping timeout: 252 seconds]
<mcstar> yeah, i meant the asm
<mcstar> if something struck you
<mcstar> well, thanks all for the attention
<f[x]> btw -inline also doesn't give me speedup but it makes asm harder to search
<mcstar> f[x]: btw, what is your platform?
<f[x]> x86_64, 3.12.1, debian testing
mort___ has joined #ocaml
Fnar has quit [Ping timeout: 246 seconds]
Fnar has joined #ocaml
Fnar has quit [Changing host]
Fnar has joined #ocaml
_andre has joined #ocaml
mfp has quit [Read error: Connection reset by peer]
chambart has joined #ocaml
mfp has joined #ocaml
chambart has quit [Ping timeout: 244 seconds]
eni has joined #ocaml
avsm has quit [Quit: Leaving.]
chambart has joined #ocaml
eni has quit [Quit: Leaving]
chambart has quit [Ping timeout: 248 seconds]
chambart has joined #ocaml
chambart has quit [Ping timeout: 248 seconds]
chambart has joined #ocaml
avsm has joined #ocaml
mort___ has quit [Quit: Leaving.]
sepp2k1 has joined #ocaml
sepp2k has quit [Ping timeout: 265 seconds]
mmajchrzak has quit [Ping timeout: 265 seconds]
gnuvince has quit [Ping timeout: 250 seconds]
sepp2k1 has quit [Read error: Connection reset by peer]
zorun has quit [Ping timeout: 246 seconds]
Anarchos has joined #ocaml
smondet has joined #ocaml
gnuvince has joined #ocaml
Ninju has quit [Ping timeout: 240 seconds]
thizanne has joined #ocaml
Ninju has joined #ocaml
ftrvxmtrx has quit [Quit: Leaving]
Ninju1 has joined #ocaml
Ninju has quit [Ping timeout: 248 seconds]
mika1 has quit [Quit: Leaving.]
zorun has joined #ocaml
ankit9 has joined #ocaml
<f[x]> mcstar, indeed, 3.12.1 with -g would box float ref used in a loop. It is fixed in 4.00
<mcstar> f[x]: how come you didnt experience a difference?
<f[x]> I do see a difference on a small example
<f[x]> maybe on a bigger example this optimization didn't kick in due to something else
silver has quit [Read error: Connection reset by peer]
<mcstar> f[x]: these other standard libs, like janestreet, or batteries, would make a difference in performance?
ankit9 has quit [Quit: Leaving]
<mcstar> im just asking in what way are they better?
<thelema_> mcstar: not on a function that doesn't use them
<thelema_> mcstar: they offer lots of convenience
<f[x]> what thelema_ says
<mcstar> do they replace arrays?
* f[x] afk
<thelema_> they offer replacements, but don't replace standard arrays
<thelema_> for example, batteries has Vect, which stores array elements in a tree structure for fast concat, insert, etc.
<mcstar> o.O
<thelema_> And core has a resizable array
<mcstar> why call it a vect
<thelema_> mcstar: that's what the original author called it - "Rope" and "Vect"
<thelema_> rope for strings, vect for arrays of arbitrary types
<mcstar> ok, so these libs provide convenience, but not performance enhacement
<thelema_> well, there's certainly possible performance enhancements if you're doing lots of appends to large arrays.
<thelema_> but they're algorithmic performance enhancements, not faster implementations of stdlib functions
<thelema_> (although both offer tail recursive versions of the stdlib functions that aren't, so that you don't accidentally blow your stack.)
<mcstar> thats good news
<thelema_> the reason they don't do this is that the builtin implementations are already damn efficient.
<thelema_> so there's no room for improvement.
<mcstar> wrt what?
<mcstar> ok, just nitpicking
tumdum has quit [Disconnected by services]
Submarine has joined #ocaml
Submarine has quit [Changing host]
Submarine has joined #ocaml
robthebob has quit [Quit: Leaving]
avsm has quit [Quit: Leaving.]
Znudzon has quit [Remote host closed the connection]
zorun has quit [Ping timeout: 252 seconds]
djcoin has quit [Quit: WeeChat 0.3.2]
fds has quit [Ping timeout: 248 seconds]
joelr has joined #ocaml
<joelr> good day
Xizor has joined #ocaml
joelr has quit [Quit: joelr]
pangoafk is now known as pango
jaar has quit [Ping timeout: 248 seconds]
beckerb has quit [Ping timeout: 245 seconds]
smondet has quit [Read error: Connection reset by peer]
ulfdoz has joined #ocaml
ulfdoz has quit [Ping timeout: 246 seconds]
thomasga has quit [Quit: Leaving.]
smondet has joined #ocaml
eni has joined #ocaml
ocp has joined #ocaml
<Ninju1> Hi, I need to package SDL libs with my app so I can distribute a binary to my client but I haven't had any luck with it. Does anyone know the right flags to use because I don't seem to be making any progress since adriens help yesterday.
<adrien> well, you're trying to run the binary and failing I guess? which error message do you get?
<Ninju1> hold on a sec, im running it on my other machine to test
<Ninju1> bsaically it says "dyld: Library not loaded: /usr/local/lib/libSDL-1.2.0.dylib"
<Ninju1> thats the error i've pretty much got for everything i've tried
<Ninju1> I have the sdl libs in ./lib/sdl and the .so files in ./lib/so relative to my project
<Ninju1> but everything i've tried to get it to look there hasn't worked
<Ninju1> and im now compiling with ocamlopt instead of ocamlc btw
ousado_ has joined #ocaml
pheredhel has joined #ocaml
wormphle1m has joined #ocaml
dgfitch_ has joined #ocaml
<thelema_> Ninju1: by the "dyld", I guess that you're running on mac, right?
snarkyboojum_ has joined #ocaml
taruti_ has joined #ocaml
strlen_ has joined #ocaml
eni has quit [Ping timeout: 240 seconds]
beckerb has joined #ocaml
peddie_ has joined #ocaml
<Ninju1> thelema_: yep thats right
cdidd has quit [*.net *.split]
snarkyboojum has quit [*.net *.split]
lopex has quit [*.net *.split]
ssbr has quit [*.net *.split]
pheredhel` has quit [*.net *.split]
wormphlegm has quit [*.net *.split]
peddie has quit [*.net *.split]
jave has quit [*.net *.split]
strlen has quit [*.net *.split]
ousado has quit [*.net *.split]
dgfitch has quit [*.net *.split]
adrien has quit [*.net *.split]
taruti has quit [*.net *.split]
peddie_ is now known as peddie
ssbr- has joined #ocaml
jave has joined #ocaml
eikke has quit [Ping timeout: 250 seconds]
<thelema_> Ninju1: the output suggests to me that you're not actually statically linking libSDL into your executable
<thelema_> you say that you have the libs in ./lib/*, but why would the system's dynamic linker look there?
adrien has joined #ocaml
<Ninju1> yeah exactly - failing to link them statically i was hoping to try and link them dynamically instead and send them with the build, but I can't get either method working
cdidd has joined #ocaml
<mcstar> putting them next to the binary might work
<Ninju1> can't hurt to try
<Ninju1> i'll let you know in a sec
<mcstar> (thats how i deployed something on linux/windows)
ousado_ is now known as ousado
ousado has quit [Changing host]
ousado has joined #ocaml
lopex has joined #ocaml
<mcstar> i read as a strong point of ocaml that you can statically link anything with your binary
<mcstar> though, im not sure how is that different from the usual static linking
<thelema_> bool prompt_yn(string prompt) {
<Ninju1> I'm sure you can, but I'm not sure what I'm doing wrong because I've tried so many different things to try and get this working
<thelema_> }
<thelema_> int atoi(string s) { return atoi(s.c_str()); }
<thelema_> bool is_int(const string s) { return s.find_first_not_of("0123456789") == string::npos; }
<thelema_> oops, mispaste
<mcstar> seems not just me writing c++ from here
<thelema_> maybe section 18.1.3 from the manual will help
<Ninju1> i believe `sdl-config —static-libs` returns the flags you need to link sdl statically but even managing to get that into the compile command didnt work
<Ninju1> I read that earlier actually :-/
<Ninju1> i'll do it again in case I missed something i guess
<thelema_> so you're using the native code compiler and putting the libSDL.so files on the command line?
<mcstar> Ninju1: so it didnt work to put the libs beside the binary?
<Ninju1> mcstar: nope just did that and it didn't work any better (same error as previous)
strlen_ is now known as strlen
<adrien> Ninju1: well, the ocaml code is linked statically, as it is by default
<Ninju1> native code compiler and im not sure how to link the .so files but I've tried lots of different things
<adrien> the C one isn't
<thelema_> what's the command you're using to produce your executable?
<thelema_> you may need a -ccopt -lSDL
<Ninju1> one sec
<adrien> I *think* you can *maybe* link the .dylib as static if you also have a .la file with a similar name
<mcstar> $LD_LIBRARY_PATH
<mcstar> $DYLD_LIBRARY_PATH
<mcstar> The process’s working directory
<mcstar> $DYLD_FALLBACK_LIBRARY_PATH
<mcstar> thats the dlopen search priority on osx
<Ninju1> its hard to give you the command because obviously I've had so many different ones (during the process of trying to get this to work), so I guess you could call this the base (where i haven't got anything in Makefile that tries to include the files in my lib dir)
<Ninju1> $ make
<Ninju1> ocamlfind ocamldep -I +sdl -I src/ src/main.ml > ._d/src/main.d
<Ninju1> ocamlfind ocamlopt -package extlib,sdl.sdlimage,sdl.sdlmixer,sdl.sdlttf,sdl,sdl.sdlgfx,threads -c -thread -I +sdl -I src/ src/main.ml
<Ninju1> File "src/main.ml", line 91, characters 8-10:
<Ninju1> Warning 26: unused variable dt.
<Ninju1> ocamlfind ocamlopt \
<Ninju1> -package extlib,sdl.sdlimage,sdl.sdlmixer,sdl.sdlttf,sdl,sdl.sdlgfx,threads -linkpkg \
<Ninju1> -thread -I +sdl -I src/ -cclib "-framework Cocoa" -ccopt -Lsrc/ -o build/project-x \
<Ninju1> src/main.cmx
<Ninju1> i tried setting LD_LIBRARY_PATH=lib/so (and various other things) but that didn't give me any better results
<mcstar> lib/so?
<mcstar> what would that mean?
<Ninju1> oh
<Ninju1> thats just where i keep all the *.so files
<mcstar> i see
<mcstar> im wrong
<mcstar> well, not really
<mcstar> no, its ok :)
<mcstar> btw
<mcstar> im not sure whether you need absolute pathnames
<mcstar> it would make sense if you do
<mcstar> Ninju1: try adjusting LD_ by using an absolute path
<Ninju1> hold on
<Ninju1> i've got a call about this exact project lol
ftrvxmtrx has joined #ocaml
<Ninju1> btw i have tried LD_LIBRARY_PATH=/absolute/path/to/my/project/lib/so
<emias> Anyone happen to know some RADIUS (client) implementation written in OCaml? Google seems to suggest there's no such thing ...
<mcstar> Ninju1: i suggest you read the dyld man page
<mcstar> there are interesting fact there
<mcstar> facts*
<mcstar> Unlike many other operating systems, Darwin does not locate dependent dynamic libraries via
<mcstar> their leaf file name. Instead the full path to each dylib is used (e.g. /usr/lib/libSys-tem.B.dylib). /usr/lib/libSystem.B.dylib).
<mcstar> tem.B.dylib).
<mcstar> hah, sry for the mispaste
<mcstar> not my fault
<mcstar> the osx docs page after hyphenation, continues the text, in white color, blends with the background
<mcstar> so i copied more...
Submarine has quit [Quit: Leaving]
<Ninju1> sorry, still on that call; I'll read the dyld man page after
<Ninju1> I just hope i can get it working because at this rate we may have to switch tech
<Ninju1> thanks for you help so far though
<Ninju1> *your
<adrien> well it's really a system issue
<mcstar> yeah, next time you will have an easier time to deploy on osx, if you understand the issue now
cdidd has quit [Quit: Leaving]
<gildor> anyone knows a little bit CDuce ?
_andre has quit [Quit: leaving]
<ousado> what would you recommend to pretty print source code from an AST, is there something more targeted / better suited than Format?
<ousado> (the language in question is not ocaml)
<ousado> smondet: thanks
<thelema_> gildor: seems like a terrible idea to me
<gildor> thelema_: what is a terrible idea ?
<adrien> the whole cduce approach?
osa1 has joined #ocaml
BiDOrD_ has joined #ocaml
BiDOrD has quit [Ping timeout: 252 seconds]
<thelema_> yup.
<thelema_> well, just the "using XML for programming syntax"
<thelema_> the rest I can't comment on
<gildor> thelema_: so I will suffer in silence ;-)
<gildor> nobody will help me !
<thelema_> gildor: why do you want to know cduce?
<gildor> ocsoap is using it, I need to fix some bugs in it
<thelema_> gildor: okay, here's "why" #2, why do you need ocsoap to work for you?
ocp has quit [Quit: Leaving.]
<gildor> because forge.ocamlcore.org provides an API in SOAP
<thelema_> gildor: does anyone use that API?
<gildor> I use it in a Python script to upload stuff
<gildor> but this was an exploratory script
<gildor> the final step is to be able to transform archive name (foo.tar.gz) into their URL (http://forge.ocamlcore.org/download.php/frs/918/foo.tar.gz)
<gildor> I have used a direct access to PostgreSQL in the past to do that, but this is really ugly
thomasga has joined #ocaml
<Ninju1> mcstar: I had a look at the dyld man page, and DYLD _LIBRARY_PATH looks like what i need, but setting it made no difference
<Ninju1> if you have any other ideas I'd love to hear them or things I could search for/read etc
<mcstar> Ninju1: i think you have to apply some naming convention
<mcstar> as per the exact passage i imgure'd says
<mcstar> only then will you be able to relocate the binary with the libs
<mcstar> it must be a simple thing, but i havent done any development on osx
<Ninju1> sorry I didn't quite follow what you mean there
<mcstar> thats the image i pasted for different reasons
<mcstar> but i think your answer is in there
<Ninju1> ahh yeah, thats what i just tried a few minutes ago
<mcstar> Ninju1: no, you tried setting *_LIBRARY_PATH
<mcstar> thats not what it means afaik
<Ninju1> ohh
Snark has quit [Quit: Quitte]
<Ninju1> yeah you're right I did
<mcstar> in the binary, you have to change the actual references to the dlls
<mcstar> as for that, im not sure how to proceed
<mfp> Ninju1: have you tried using otool yet?
<Ninju1> yeah thats what i've been using to check references mostly
<Ninju1> do you want the output of that? not sure if it would help at all
<mfp> Ninju1: take a look at this http://macdylibbundler.sourceforge.net/
<Ninju1> ok thanks
<mfp> otool can be used to rewrite some magic vars in the exec used to find the dynlibs; the program I linked to automates this
ftrvxmtrx has quit [Remote host closed the connection]
<Ninju1> ahh ok, I'll read it through and let you know
ftrvxmtrx has joined #ocaml
Znudzon has joined #ocaml
hongboz` has quit [Remote host closed the connection]
gnuvince has quit [Ping timeout: 245 seconds]
osa1 has quit [Quit: Konversation terminated!]
eikke has joined #ocaml
mcstar has quit [Quit: mcstar]
smondet has quit [Quit: Leaving]
Cyanure has quit [Remote host closed the connection]
eikke has quit [Ping timeout: 264 seconds]
<ousado> smondet: wow, that's a powerful thing
eni has joined #ocaml
eni has quit [Client Quit]
sepp2k has joined #ocaml
beckerb has quit [Remote host closed the connection]
<Anarchos> Is it possible to do a grammar which is ambiguate whith ocamlyacc ?
<_habnabit> Anarchos, which is 'ambiguate' ?
<Anarchos> _habnabit i want to parse math formulas written in latex. And we all know that mathematicians can use much notations which are ambiguates
<_habnabit> Anarchos, did you mean ambiguous?
<Anarchos> _habnabit i want to return the ambiguous sentence to the user to ask him to disambiguate the situation
* Anarchos learnt the noun "amibguous"
<Anarchos> _habnabit no idea ?
<_habnabit> not offhand
<Anarchos> _habnabit well i guess it will be the trickiest part of my soft...
<_habnabit> could you give an example?
<Anarchos> _habnabit just think to all abuse of notations in your maths lessons...
<Anarchos> _habnabit A*B*C for cartesian product of sets
Xizor has quit [Ping timeout: 260 seconds]
thomasga has quit [Quit: Leaving.]
Anarchos has quit [Quit: Vision[0.9.7-H-090423]: i've been blurred!]
<thelema_> yes, it's possible to parse ambiguous grammars, generally by setting an associativity
Ninju1 has quit [Quit: Leaving.]