[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Pan-devel] g_message function calls

From: Guilhem Bonnefille
Subject: Re: [Pan-devel] g_message function calls
Date: Tue, 26 Dec 2006 13:49:13 +0100


I attached a patch to #372013 ( ).

This patch make the logger thread safe. It also replace all g_message
calls by a log. So, no more message on the console.

On 11/8/06, Charles Kerr <address@hidden> wrote:
I opened a ticket for this at
with some links to mingw-friendly implementations
of nonblocking connections, but I'm pretty busy.

If someone wants to make a patch, though, I'd be happy
to accept it.


robert mcdonald wrote:
> You're correct. And the windows variant of the code is
> extremely similar (trivial differences) and works
> equally well.

>> On 11/6/06, Charles Kerr <address@hidden>
>> wrote:
>>> The socket creation code is a special case there
>> -- it's the /only/
>>> part of Pan that uses threads, and it's a very
>> small task.  The
>>> purpose here is to make a connection in a
>> nonblocking way so that
>>> Pan doesn't freeze while connecting.
>> Humh... As I remember, it is possible to make a
>> socket connection
>> non-blocking without using threads. I have done such
>> thing with select
>> (or poll). The algo: create the socket and connect
>> it (O_NONBLOCK),
>> then register the file descriptor in select (poll)
>> and then, when
>> something occurs, it is probably a connection
>> success (or failure).
>> Look at
>> Perhaps it could be interesting to rewrite the
>> "connection" part, so
>> pan will be fully thread safe (which is more simpler
>> and safer than
>> multi-thread). I'm quite sure that such asynchronous
>> connection could
>> be made with the GLIB "event loop".
>> What's your opinion? Do you already try such
>> asynchronous connection?

Pan-devel mailing list

-=- #UIN: 15146515 JID: address@hidden MSN: address@hidden
-=- mailto:address@hidden

reply via email to

[Prev in Thread] Current Thread [Next in Thread]