Re: [PATCH] http.c: Use curl_multi_fdset to select on curl fds instead of just sleeping

From: Mika Fischer <mika.fischer@zoopnet.de>
Date: Oct 30, 2011, 07:19 UTC
Message-ID: <CAOs=hR+YuF+HP0n0132Ktm3RdeWsnVp0Bgt89LNn+VyT6W0mcw@mail.gmail.com>
In-Reply-To: <alpine.DEB.2.00.1110292230500.28196@tvnag.unkk.fr>

On Sat, Oct 29, 2011 at 22:33, Daniel Stenberg <daniel@haxx.se> wrote:

Show 12 quoted lines

>> +                       curl_multi_fdset(curlm, &readfds, &writefds,
>> &excfds, &max_fd);
>> +
>> +                       select(max_fd+1, &readfds, &writefds, &excfds,
>> &select_timeout);
>
> At times, curl_multi_fdset() might return -1 in max_fd, as when there's no
> internal socket around to provide to the application to wait for.
>
> Calling select() with max_fd+1 (== 0) will then not be appreciated by all
> implementations of select() so that case should probably also be covered by
> the 50ms sleep approach...

Actually, the 50ms sleep was also implemented using select(0, ...) before the patch. I tried to keep the previous behavior when curl does not give us any information. I assumed that the select(0, ...) was some portable way to sleep with microsecond granularity. Is there some other way to tell select not to check any fds, or should I just call select(1, ...)?

Best,
 Mika