01:57 <vincenz> any good live isos?

01:57 <vincenz> for 64bit

02:10 <picard> is there a way to create a block around a for loop? I want to make a local variable and manipulate it inside the for loop, then return it... but the return type of the for loop is ()

02:12 <Smerdyakov> I think most of consider for loops highly deprecated.

02:12 <Smerdyakov> s/most of/most of us

02:12 <picard> what is wrong with them?

02:19 <Smerdyakov> They're imperative.

02:19 <Smerdyakov> Which means harder to understand.

02:20 <vincenz> besides

02:20 <vincenz> that's easily done with a ref

02:20 <vincenz> let x = ref 0 in for loop; !x

02:20 <picard> ok thanks

02:26 <Smerdyakov> picard, I'd encourage you to consider how to do this better with higher-order functions and no use of "for."

02:29 <vincenz> tail-recursion

02:29 <Smerdyakov> vincenz, naturally, you'd like to avoid writing your own recursive functions whenever library functions can do the recursion for you.

02:33 <vincenz> how so?

02:33 <vincenz> oh

02:33 <vincenz> foldl and such

02:36 <picard> Smerdyakov: I will look into that, I am just trying to multiply to matrices, and I always did it with for loops before

02:39 <Smerdyakov> You can do slightly better with iter functions over arrays.

02:39 <picard> is it faster?

02:40 <Smerdyakov> It's more compact. It makes little difference in this case.

02:41 <picard> well I will definitely check it out, thanks

09:54 <Nutssh> Use whichever is more clear.

10:34 <det> vincenz, Ubuntu has an AMD64 live cd

10:34 <det> vincenz, and it is basicly debian sid

15:45 <hum> hello

15:49 <hum> I've a complety off-topic question: Is there an (ex-)student from the ENS's MMFAI (a french master degree in Paris) here ?

15:49 <hum> (i think it is a good place to find one !)

16:11 <gl> hum : parfois il y a pnou qui passe

16:14 <hum> je suis étudiant en province, j'espère être admis (sur dossier) au MMFAI, c'est pourquoi je cherche un logement sur Paris.

16:14 <hum> Je me demande donc s'il y a des facilités d'accès au logement pour les "auditeurs" (les non-normaliens, ou je dois simplement passer par le crous ...)

16:15 <gl> aucune idée

16:15 <hum> :/

16:40 <Snark> hum: le mieux est probablement de téléphoner au secrétariat de l'École...

16:43 <hum> C'est ce que gl m'a conseillé, je vais faire ça demain, merci !

16:58 <picard> I need to work with matrices for a program that I am writing. I am new to OCaml so I'dd like to know whether I should make a class or use a module

16:58 <Smerdyakov> Most of us never use classes.

16:59 <picard> any reason why?

17:00 <picard> the problem that I am having is that I would like to be able to use floats and ints interchangably as elements of the matrix, but then I would have to switch between using + and +., how should I take care of that?

17:02 <picard> I'm sorry if these questions seem really stupid, but I am new to using OCaml

17:03 <Smerdyakov> Functors

17:03 <Smerdyakov> Even better, grab a matrix library that's already written.

17:05 <picard> well that would be easier, but since I am still getting comfortable with OCaml, I think I should write my own anyway

18:54 <Nutssh> Classes in other languages are used to handle scoping, modularity, encapsulation, recursive types, and other duties. OCaml has an excellent module and sum-type functionality, making classes unnecessary for those purposes. I have used classes, but only when I wanted dynamic dispatch based on the runtime type. (and occasionally for very complicated fold-over-AST functions)

21:29 <ulfdoz> good eveninng.

21:30 <ulfdoz> Is there a way to specify a finalizer for an object. I want to close some filedescriptors, when an object isn't needed anymore.

21:36 <Smerdyakov> I think if you look at the standard library documentation, you will find it.

21:38 <ulfdoz> Gc.finalise sounds good. Sorry, if this was a stupid question, I'm probably thinking to much in java, where a finalizer belongs to the object.

21:42 <ulfdoz> thx

22:20 <ulfdoz> The strange thing is, that it seems to register the finalizer, i.e. it runs the initializer without any errors, but it seems not to apply the finalizer-function, or i've a bug in my debug output.

22:22 <Smerdyakov> You're sure a GC is occurring and that the object is really unreachable?

22:23 <ulfdoz> It's at program exit. I wrote a really simple test-app.

22:23 <ulfdoz> It opens a networking-connection, sends a string and terminates

22:24 <Smerdyakov> Did you interpret the documentation as saying that all finalizers are called at program exit?

22:25 <ulfdoz> No, I just assumed, because it seemed logical to clean up on exit.

22:27 <Smerdyakov> Now that I think of it, I'm surprised that OCaml supports finalizers at all, since I thought the GC never explicitly deallocated anything.

22:28 <Smerdyakov> I guess they need an extra mechanism to figure out when an object is unreachable.

22:28 <ulfdoz> But programm-termination would be safe point, I think.

22:30 <Smerdyakov> Just because it's safe doesn't mean that it is cost-free to implement it.

22:30 <Hadaka> Smerdyakov: yeah, finalizers and their ordering is kind of a special case - and is not really related to actually deallocating the objects

22:30 <Smerdyakov> They need a special case to run finalizers at GC time, and maybe they didn't want to also implement a special case for program exit time.

22:34 <ulfdoz> I tried to register the finalizer in the initializer of the object, so it's not definitely the program exit time it is called, but I can do a call to the Gc in main for now.

23:10 <ulfdoz> Seems to work now, but quite lowlevel.

23:26 <ulfdoz> cya