Version française
Home     About     Download     Resources     Contact us    

This site is updated infrequently. For up-to-date information, please visit the new OCaml website at

Browse thread
[Caml-list] How can I lend developers to the Caml team?
[ Home ] [ Index: by date | by threads ]
[ Search: ]

[ Message by date: previous | next ] [ Message in thread: previous | next ] [ Thread: previous | next ]
Date: 2004-04-22 (08:22)
From: Alex Baretta <alex@b...>
Subject: Re: [Caml-list] How can I lend developers to the Caml team?
As I haven't seen my former post come back to me through the list, I
infer my mail client has consigned it into oblivion. If you have
actually received it, please, ignore this post.

Xavier Leroy wrote:
> The only mail I got from you recently (last Tuesday) was about
> possible topics for a student internship, which isn't quite what
> you're describing in this message.  So, maybe I missed something.

I'll come to this in a second.

> As for the "very busy" part: the Tuesday mail arrived while I was
> connecting flights at Helsinki airport on my way to the "Applied
> Semantics" workshop, from which I returned yesterday night, so you see
> that I was hard-pressed to reply any earlier :-)

No problem :) I figured you must have been away and busy. I sent a mail
to the list to let the other Caml breeders in on this project and have
their feedback, too.

> We'll discuss this need in private -- the first thing is probably to
> identify what itch you'd like to scratch.  Generally speaking, there
> is a lot to be done in the Caml world (as many discussions on this
> list point out), but mainly outside the "core compiler and runtime"
> fragment.  

Ok, here's the essence of my project: we use ocaml to make a living. As
I have been asked by a fellow camler about my business, let me very
briefly say that we are a start-up developing GPLed business
applications with a set of domain specific tools/languages built on top
of Ocaml. There were three of us in October 2002; there are 12 of us
now, and we plan to grow consistently throughout 2004. This explains why
we feel very strongly about participating in the development of the
core. It's a sort of life insurance, you see. If we find a bug which is
critical to our process, what would we do if the Caml team does not have
time to solve it immediately? Besides, bug fixes are at best made
available in the CVS, which for other stuff might not be stable. In
conclusion, since there is no service level agreement between INRIA and
my company, I need to guarantee that we can stick our nose into the code
by ourselves if need be. The only way to achieve the competence to do
this is to consistently devote some development resources to the Ocaml
core, and thus develop an internal Caml-breeding team.

Since this project aims mainly at developing our competence on the Ocaml
core, it is sensible to acquire new resources in the form of interns to
be trained for a future Ocaml job in our company. I have one such person
whom I would like to use on the Caml core in the next several months. I
would also devote at least one other person at "runtime", say a few 
months from now, once che cooperation protocol with Inria will have been 

> I hope this isn't going to turn in an experiment to determine how much
> "Cathedral" we INRIA folks are (answer: very much so).  Additional

This is indeed not a problem. The Bazaar vs. Cathedral thread has 
essentially shown that the community effort will tend to condense around 
Gerd's GODI, which seems very promising.

> manpower and external contributions are most welcome, but as I
> indicated above, the demand is mostly on 3rd-party libraries and tools

There are also some major aspects of the core system which still require 
a good deal of effort: native code Dynlink, ocaml shared libraries, JIT 
compiler, code unlinking...

> Some parts of the core OCaml distribution could use outside help
> (e.g. Windows support, reorganization of the Camlp4 sources, Makefile
> reengineering), provided copyright and maintenance issues can be
> resolved.  But the compilers and core runtime system don't need
> additional manpower, in my opinion.

Now, you know better than I that industrial languages are developed by 
(too) many people. Camls are bred by the some of coolest hackers on the 
scene, but it's too small a team to handle everything. A few more 
herdsmen can only help, if adequately trained and guided. If you'll 
allow me, I'd like to staff specific projects within the core. I'll 
accept your priorities so long as you accept to put my requests 
somewhere on the TODO list. I am also willing to donate copyright to 
INRIA for all code included in the Ocaml standard distribution, so long 
as INRIA in willing to guarantee that Ocaml will remain free software. 
All I need is a little cooperation in terms of some supervision so as to 
achieve a cooperative development effort as opposed to a fork, which is 


To unsubscribe, mail Archives:
Bug reports: FAQ:
Beginner's list: