[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Demexp-dev] Fwd: [Caml-list] The OCaml Summer Project
From: |
David MENTRE |
Subject: |
Re: [Demexp-dev] Fwd: [Caml-list] The OCaml Summer Project |
Date: |
Fri, 26 Jan 2007 16:09:31 +0100 |
Hi,
2007/1/26, ketty <address@hidden>:
I am not sure yet if i want apply. For one thing it would require me
to have fairly complete plan to begin with and then stick to it. I am
not good at that kind of things. :-)
:-) But that could be a good motivation to learn that kind of thing!
;-) But you are your only judge on this.
But if you want a mentor, I'm ready to advise you.
And i don't know what project is accepted. Maybe it should be fairly
self contained? (but interfacing/integrated with demexp)
Well, this is the extension-of-an-existing-project kind, isn't it?
If we have concrete project suggestions we could suggest them to
janestcapital andor universities with courses using caml?
Maybe. But I need to allocate some time on this and my time slots are
already fairly booked. :-)
(realistic ?) suggestions:
* caching/relaying (sub)server to minimize load on main server.
Not very interesting right now. Before optimizing, we need a real
analysis of congestion points and their cause.
* application/tools to filter/organize questions in useful manners.
(could then be reused by different clients).
That would be definitely useful. For example, I find the LISFS
technology very promising. They have a theoretical formalism and a
prototype (GNU GPL, in OCaml) to have a tag-based search technology
where you can freely mix google-like queries and unix-directory-like
search. That would be a big plus for demexp to have a classification
server using that technology.
http://lfs.irisa.fr/index.html
http://lfs.irisa.fr/demo-area/
http://lfs.irisa.fr/prototypes.html
Here is an example on how it could be applied to demexp:
http://www.mail-archive.com/address@hidden/msg00351.html
* various clients. (web, standalone, commandline, curses)
commandline client is already fulfilled by Thomas' Python binding.
As I said previously, a web interface would be definitely useful.
The standalone client could have some improvements, but I would really
prefer work on a web interface.
Best wishes,
d.