sponge45 changed the topic of #ocaml to: Discussions about the OCaml programming language | http://caml.inria.fr/
mnemonic has quit [Read error: 145 (Connection timed out)]
CissWit has joined #ocaml
<CissWit>
Hello
<CissWit>
whouhou very talkative lol
<CissWit>
des frenchis ptete ?
<Smerdyakov>
Please don't use color anywhere on Freenode where you haven't explicitly verified that channel policy allows/encourages it.
<mattam>
oui
<CissWit>
ops sorry
<mbishop>
Smerdyakov: perhaps the channel should be +c?
<CissWit>
thats what i wondered
<CissWit>
la forme mattam?
<Smerdyakov>
mbishop, well, first, am I wrong that color is inappropriate on Freenode by default?
<mattam>
euh oui
<mbishop>
I'm not sure, but I know most channels, and indeed most people, don't like it
danly has joined #ocaml
<mbishop>
is there an official rule? *shrug*
<Smerdyakov>
There could be situations where it's useful, but making all your text blue by default is not one of them.
<mbishop>
probably not, since they leave it up to the channel to decide (with the +c mode)
<CissWit>
well im sorry lol i am usually on an other server, i didn't unload my script
<CissWit>
but still, colors make a chan a little more easy to read i think .. even if i guess that here, there are not even 20 people talking together ^^
<CissWit>
*often not even stupid me >_<
<CissWit>
well, lets talk about ocaml, has anyone ever done anything in ocam to log on an irc server ?
<mattam>
not sure it helps anyway :)
<mattam>
there was an IRC client example in some version of lablgtk i think.
<CissWit>
oh ben moi j'arrive mieu a lire un ecran ou j'ai des couleurs pour me reperer dans les gens qui parlent, qua quand cest tout gris ou noir
<Smerdyakov>
CissWit, y'know, it's not very nice to do that for people who don't read French.
<CissWit>
he is the only who answered my question, its not very nice not to answer me btw :/
<CissWit>
for the moment you told me : 1 that i shouldn't use color, 2, that im not very nice
<CissWit>
u didn't even answer to my hello
<CissWit>
is that very nice Smerdyakov ?
<Smerdyakov>
IRC etiquette does not deem it impolite to avoid answering "hello."
<Smerdyakov>
If everyone did that, 95% of channel traffic would be hello-acknowledgements.
<Smerdyakov>
And I don't have a helpful answer for your question about OCaml, so the nicest thing to do is stay silent if anyone else at all has answered.
<CissWit>
stop complaining five minutes :/ you should have a rest :/
<CissWit>
are all channel such cold on freenode ?
<CissWit>
is that an irc etiquette ?
<Smerdyakov>
Very few things are considered as rude as using colors on Freenode. The conversation doesn't usually have this tone; that's what set it off.
<CissWit>
i apologized for colors :/
<Smerdyakov>
Yes. And I was never rude or cold about that or anything else.
<CissWit>
do i need to kiss your feet for beeing so rude ?
<Smerdyakov>
No. You're the one perpetuating this line of conversation, not I.
<CissWit>
anyway, how are u Smerdyakov ? ^^
<Smerdyakov>
Fine, thanks!
<CissWit>
im fine too thanks
<CissWit>
Smerdyakov ? can u tell me where i can find those irc etiquette which you talked about ?
johnnowak has quit [Read error: 110 (Connection timed out)]
johnnowak has joined #ocaml
johnnowak_ has quit [Read error: 110 (Connection timed out)]
diffbavis has quit [Read error: 104 (Connection reset by peer)]
CaptainBanana_ has joined #ocaml
diffbavis has joined #ocaml
CaptainBanana has quit [Read error: 110 (Connection timed out)]
_JusSx_ has quit [Read error: 104 (Connection reset by peer)]
CaptainBanana_ is now known as CaptainBanana
swater has joined #ocaml
smimou has joined #ocaml
malc_ has joined #ocaml
malc_ has quit ["leaving"]
diffbavis has quit [Read error: 104 (Connection reset by peer)]
ikaros has joined #ocaml
ikaros has quit [Read error: 104 (Connection reset by peer)]
ikaros has joined #ocaml
Smerdyakov has joined #ocaml
_JusSx_ has joined #ocaml
malc_ has joined #ocaml
malc_ has quit ["leaving"]
diffbavis has joined #ocaml
pants1 has joined #ocaml
_JusSx_ has quit [Client Quit]
_JusSx_ has joined #ocaml
ikaros has quit ["Leaving"]
ikaros has joined #ocaml
stevan__ has quit ["Leaving"]
johnnowak has quit []
stevan_ has joined #ocaml
Shimei has quit [Remote closed the connection]
_JusSx_ has quit [Client Quit]
_JusSx_ has joined #ocaml
hcarty has quit [Remote closed the connection]
swater has quit ["Quat"]
pango__ has joined #ocaml
pango__ is now known as pango
hcarty has joined #ocaml
pango_ has quit [Remote closed the connection]
pango has quit [Remote closed the connection]
pango__ has joined #ocaml
diffbavis has quit [Read error: 104 (Connection reset by peer)]
mnemonic has joined #ocaml
_JusSx__ has joined #ocaml
<mnemonic>
hi
benny_ has joined #ocaml
_JusSx_ has quit [Read error: 113 (No route to host)]
love-pingoo has joined #ocaml
benny has quit [Read error: 110 (Connection timed out)]
pango__ has quit [Remote closed the connection]
pango__ has joined #ocaml
slipstream has quit [Read error: 110 (Connection timed out)]
slipstream has joined #ocaml
malc_ has joined #ocaml
_JusSx_ has joined #ocaml
_JusSx__ has quit [Read error: 110 (Connection timed out)]
pango__ is now known as pango
malc_ has quit [Read error: 110 (Connection timed out)]
dark_light has joined #ocaml
<dark_light>
if i create two connected sockets with Unix.socketpair, how can i trigger a "exceptional condition" in one socket to be handler at the other point? (like the one captured third parameter of in Unix.select)
<tsuyoshi>
I don't think unix sockets have exceptional conditions
<tsuyoshi>
iirc that's just a tcp thing
<tsuyoshi>
er.. tcp/udp
<dark_light>
ah, no, it's not a unix socket
<dark_light>
it's just created from the Unix module
<tsuyoshi>
huh? doesn't socketpair give you a unix socket.. hmm
<tsuyoshi>
oh you can specify the protocol
<tsuyoshi>
in that case you use the MSG_OOB flag with sendmsg
<dark_light>
Hmmmm
<dark_light>
nice, thanks :)
<dark_light>
just to say, my problem is: i am a thread that waits for two conditions: reading from a socket or reading for a signal "kill yourself" from another thread. i am polling this with Event.poll to read the signal and Unix.select with no wait to read the socket
<dark_light>
it's working, but it's very slow, because there are no blocking (it is constantly changing from Unix.select to Event.poll, trying to catch something)
<tsuyoshi>
why polling?
<dark_light>
tsuyoshi, because i don't know which thing will come first
<dark_light>
if i just do a recv in the socket, it will block.. and i can receive a "kill itself" while blocking, which is bad
<tsuyoshi>
never used Event before.. where is that from
<Smerdyakov>
dark_light, so much nicer to use CML than UNIX sockets. :)
<tsuyoshi>
oh I see
<Smerdyakov>
dark_light, or any sockets, I mean.
<dark_light>
Smerdyakov, but i must use sockets, because, ugh.. it's a telnet server
<tsuyoshi>
well... in that case
<Smerdyakov>
I think you can implement a CML interface to TCP without much work.
<tsuyoshi>
the easiest thing might be to have a separate thread which selects on the socket and then sends an event when it's ready
<Smerdyakov>
You would have to keep your own extra buffers to make it work properly, though.
CaptainBanana has quit ["killed"]
<dark_light>
:o
<dark_light>
i was thinking two threads for each connection would be a big overhead, tsuyoshi..
<tsuyoshi>
how many connections are you expecting?
<dark_light>
and, Smerdyakov, i have no idea on *how* implement networking using events
<Smerdyakov>
dark_light, CML threads probably have less overhead than sockets.
<dark_light>
tsuyoshi, 20-40, but the thing might scale
<Smerdyakov>
dark_light, that is, it costs more to create and use a new socket than to create and use a new thread.
<dark_light>
Smerdyakov, hmm
<dark_light>
tsuyoshi, must scale*
<tsuyoshi>
40-80 threads is no big deal
<tsuyoshi>
what is it going to scale to?
love-pingoo has quit ["Connection reset by pear"]
<dark_light>
i expect no more than 50 connections
<Smerdyakov>
Try some experiments if you need to be sure, but, in general, creating a small constant number of threads per TCP connection should have no noticeable performance difference from one thread per entire server for that many TCP connections.
malc_ has joined #ocaml
<tsuyoshi>
2 threads * 100 connections should not be a problem unless this is on some sort of embedded thing
<tsuyoshi>
I doubt you'll see any problems at all unless you get into thousands of threads
<dark_light>
Hmmmm... and I want to send things to each client in a synchronous way. I was thinking in do a select with a) the client socket, b) a write socket. reading from b) implies in sending to client. and, exceptionals conditions in b) implies in killing the connection
<malc_>
dark_light: only that error fdlist has nothing to do with kill the connection
<dark_light>
that model seemed good enough to don't use events..
<malc_>
killing
<dark_light>
malc_, ah, but, i would receive this and interpret as a signal to kill the connection..
<dark_light>
or.. i might receive a eof in socket to means this.. both cases i would not need events
<malc_>
dark_light: on sockets the fd will be market as set when OOB is available, and not when "errors" happen (on linux and bsds at least)
<dark_light>
fd?
<malc_>
socket
<malc_>
descriptor
<malc_>
whatever
<dark_light>
malc_, i am with two sockets created with Unix.socketpair, so i may signal the OOB whatever i want
<malc_>
dark_light: yes, but if you are under impression that particular socket will be in error fdset when "bad things" happen you are mistaken, that was my point
<dark_light>
hmmm.. yes
Smergo has quit [Read error: 110 (Connection timed out)]
<tsuyoshi>
damnit.. Netclient is not in debian
<tsuyoshi>
oh wait yes it is
<tsuyoshi>
just not in testing yet
<tsuyoshi>
nor in unstable
Smergo has joined #ocaml
_JusSx_ has quit [Client Quit]
gmh33 has joined #ocaml
smimou has quit ["bli"]
malc_ has quit ["leaving"]
mtomczak has joined #ocaml
<mtomczak>
Greetings all! I'm attempting to build the ocaml-ode project and I've run into some trouble... Could someone answer a generic question about file linking to help me narrow down the problem?
<dark_light>
well, ask it
<mtomczak>
So I'm attempting to use the Makefile that is included with the ocaml-ode project (though I've had to beat on the ocaml install I have... the DarwinPorts installs lack META files for a couple of packages)...
<mtomczak>
(Oop, sorry... that's the wrong buffer).
<mtomczak>
The command I'm using "ocamlfind ocamlopt -package sdl,unix,lablgl,lablgl.glut,extlib -c katamari.ml
<mtomczak>
As the optimizer chews on the file, I get the following error: "Unbound value GlDraw.viewport"
<mtomczak>
My understanding is that GlDraw should be defined in lablgl; I'm not sure why the ocamlopt process can't find it.
<mtomczak>
So how, in theory, should 'ocamlfind ocamlopt' know where GlDraw is defined? There aren't any explicit references to lablgl in the katamari.ml file, so I assume...
<mtomczak>
... that ocamlfind compiles the .ml file in the context where the libraries are in-scope?
<tsuyoshi>
it looks for a glDraw interface file
<tsuyoshi>
glDraw.mli or glDraw.cmi
<mtomczak>
That exists in the ocaml/site-lib/lablgl directory.
<mtomczak>
Should the ocaml/site-lib/lablgl/META file be pointing to it somehow?
<tsuyoshi>
yes
<tsuyoshi>
I think without the meta file, ocamlfind won't add that include path
<tsuyoshi>
you could probably just add the path manually
<tsuyoshi>
with the -I option
<mtomczak>
How does the ocamlopt mechanism know that glDraw.cmi should be consulted?
<mtomczak>
Other languages I've used require something morally equivalent to "#include myfile.h," but I see no open, include, using, etc. statements in my katamari.ml sourcefile that reference glDraw...