01:06 <fax> I wonder how hard it would be to build basic ObjC and cocoa bindings for OCaml

01:06 <mbishop> don't they already have Cocoa bindings for ocaml?

01:07 <fax> not that I have been able to find

01:09 <mbishop> CocOcaml isn't what you are looking for?

01:09 <fax> no

01:09 <fax> thats just some cocoa program which gives you an ocaml repl afaict

01:10 <mbishop> http://alan.petitepomme.net/cwn/2005.02.08.html#6

01:10 <mbishop> page is in french, but *shrug*

01:12 <fax> oh wow cheers :D

01:12 <fax> looks like there is no way o download the bindings only a sample app :[

07:00 <zeeeeee> is there a version of Marshal that produces human-readable serializations?

07:04 <zeeeeee> (a la 'show'/'read' in haskell)

07:20 <pango> check the Hump for sexplib, json-wheel,...

07:22 <zeeeeee> pango: thanks

07:25 <zeeeeee> how do i make a class obj containing a Set of obj?

07:25 <zeeeeee> it's like the chicken and egg problem - if i first instantiate the module, obj is undefined, but if i first define the class, i haven't instantiated the Set yet

08:37 <flux-> I think you need to go via class types: class type base = object .. end module ObjSet = Set.Make (struct type t = base let compare = compare end) class actual_class = object val s = ObjSet.empty .. end

08:37 <flux-> make that actual_class : base

09:31 <cpfr> howdy

09:31 <love-pingoo> yo

09:31 <cpfr> got a quick q

09:31 <cpfr> im building a program for someone who doesnt have ocaml

09:32 <cpfr> how do i statically compile my program if it uses libs

09:32 <cpfr> like unix.cma etc

09:32 <love-pingoo> ocaml libs or C libs ?

09:32 <cpfr> ocaml libs

09:32 <love-pingoo> it's always static

09:32 <cpfr> oh well thats nice

09:33 <love-pingoo> the only thing you have to worry about is not to build an executable requiring ocamlrun if you're compiling to bytecode

09:33 <love-pingoo> make sure to use -custom

09:34 <cpfr> im compiling to machine code

09:34 <love-pingoo> no worry, then

09:34 <cpfr> excellent

09:36 <cpfr> so if i am using an ocaml lib that is a binding to a c lib

09:37 <cpfr> that will still be dynamically linked

09:37 <cpfr> or will that get statically linked

09:46 <love-pingoo> not sure about the piece of C code

09:46 <cpfr> can i run ldd to check?

09:46 <love-pingoo> it'll say that you don't need the stub C code

09:47 <love-pingoo> and I think it's right...

09:47 <love-pingoo> actually I run a custom bytecode executable without installing the C bindings on which it depends

09:48 <love-pingoo> yeah, the only dynamic stuff is the real C dynamic lib

09:48 <cpfr> well thanks this is all great news

15:19 <nucleos> hello

17:27 <nucleos> i've installed tuareg-mode, but i don't see syntax highlighting as i type in a buffer... Can someone help me?

17:48 <Lacrimosa> :)

17:50 <Lacrimosa> Can I ask something? Is there a mechanism in Ocaml for dynamic bytecode generation and execution, runtimely?

17:52 <Lacrimosa> You know, there are a lot of Java and .NET libraries for dynamic bytecode generation - what about Ocaml ?

17:56 <mbishop> er

17:56 <mbishop> I'm not sure what you mean, but perhaps you want MetaCaml?

17:57 <Lacrimosa> What's that ?

17:58 <mbishop> http://www.metaocaml.org/

17:58 <mbishop> MetaOCaml[4] is a multi-stage programming extension of OCaml enabling incremental compiling of new machine code during runtime. Under certain circumstances, significant speedups are possible using multi-stage programming, because more detailed information about the data to process is available at runtime than at the regular compile time, so the incremental compiler can optimize away many cases of condition checking etc.

17:58 <mbishop> says wikipedia

18:00 <Smerdyakov> Lacrimosa, for what applications do you want that functionality?

18:02 <Lacrimosa> Smerdyakov: code persistance in a BerkleyDB database.

18:03 <Smerdyakov> You want to store code in a database, you mean?

18:03 <Lacrimosa> Smerdyakov: Yeah, like stored procedures, but in a flat db file

18:04 <Smerdyakov> And simple dynamic loading isn't good enough for you?

18:04 <Smerdyakov> I know OCaml bytecode has dynamic linking.

18:04 <Lacrimosa> Smerdyakov: well i have to compile that...i need a faster solution

18:04 <Smerdyakov> There just isn't any special support for compilation, that I know of.

18:04 <Smerdyakov> Why do you need a faster solution? Creation of new stored procedures is in your critical path performancewise?

18:06 <Lacrimosa> well i could dynamically generate machine code but..i thought there might be a proven solution already

18:06 <Smerdyakov> The OCaml compilers are used a lot. I would consider any solution that depends on them "proven."

18:08 <Lacrimosa> but they're relatively slow for that specific kind of procedures...

18:09 <Smerdyakov> Why does it matter have compilation is slow?

18:09 <Smerdyakov> s/have/if

18:11 <Lacrimosa> well ...it's an OLTP database... about 40 transactions per second per CPU

18:11 <Lacrimosa> itanium2

18:11 <Lacrimosa> linux

18:13 <Smerdyakov> So? Do transactions create new stored procedures?

18:15 <Lacrimosa> yah its possible...there will be situations where most of the transactions will do exactly that - modify existing stored procedures

18:15 <Smerdyakov> OK, thank you. Now I understand.

18:15 <Lacrimosa> and that must be done realtime u know

18:16 <Smerdyakov> So you really have situations where "regular users" are modifying stored procedures frequently, not just relatively rare batches of changes performed by "admins"?

18:18 <Lacrimosa> well it's a component-based SOA system, so RPC calls really

18:18 <Smerdyakov> That doesn't answer my question.

18:19 <Lacrimosa> from time to time

18:19 <Smerdyakov> What can a "regular user" do to modify a stored procedure?

18:19 <Lacrimosa> ~25

18:19 <Lacrimosa> %

18:20 <Lacrimosa> no regular users or something...that isnt really directly connected to the end user

18:20 <Smerdyakov> Well, I propose you try doing it with ocamlc and dynamic linking. Run some experiments to see if performance is acceptable.

18:22 <Lacrimosa> it isnt at all, what i need is a lightweight library ...well something like LLVM's C++ API

18:24 <Lacrimosa> thank you anyway

18:59 <joelr1> good day

18:59 <joelr1> i'm doing something simple that's not working

18:59 <joelr1> let symtab : (string, sym_typ * bool) Hashtbl.t = Hashtbl.create 64

18:59 <joelr1> let clear =

18:59 <joelr1> Hashtbl.clear symtab;

19:00 <joelr1> the above is in symtab.ml but calling Symtab.clear does not clear my hash table for some reason

19:00 <joelr1> any suggestions?

19:07 <jlouis> hrmf

19:07 <jlouis> Joel!

19:09 <jlouis> let clear () = ...

19:09 <jlouis> or clear becomes a value which is evaluated at start for its side effect only

19:10 <jlouis> I guess he figured

19:18 <mbishop> everyone who mentions that question/answer seems to disconnect...*hides*