Hi
Romain Beauxis a écrit :
[ I don't know if balbinus is reading this mailing list, but I don't acutally
have his email.. ]
I am :P (and now you've got my mail address ;)
My first remark would be that it uses a *lot* of code from http_source.ml..
Indeed, a move to sources/ and a simple open Http_source and inherit allowed
to compile a harbor where only different things from http are used..
I don't know how it should be problematic, but I would recommend to use http's
code as it seems to work like a charm here..
Agreed.
On the other hand, any source connection is supposed to be infinite, so I
don't know if this is pertinent to use a single thread for each
input.harbor..
Mmm. I'd rather say that we should use one thread per client (which
translates indeed into one per input.harbor, but remember we're building
a server ;).
I also agree with your other proposals (splitting into two parts).
Cordialement,
Vincent Tabard
Radio Pytagor : http://www.radiopytagor.com/
