curl-and-python
Re: bug in pycurl 7.15.4 / curl_multi_fdset ?
Date: Tue, 27 Jun 2006 21:45:31 +0200
hi,
i'll commit a fix for the pycurl issue, thanks for the bug report.
regards,
- kjetil
On 6/27/06, K.S.Sreeram <sreeram_at_tachyontech.net> wrote:
> [pycurl-7.15.4/curl-7.15.4/c-ares-1.3.1 winxp]
>
> Note: I've also posted to curl-library & c-ares, because it could be
> some internal bug in curl_multi_fdset.
>
> When I run 'tests/test_multi4.py', I get the following error:
>
> pycurl.error: (0, 'curl_multi_fdset() failed due to internal errors')
>
> The culprit seems to be the following code in pycurl.c (line 2211)
>
> res = curl_multi_fdset(self->multi_handle, &self->read_fd_set,
> &self->write_fd_set, &self->exc_fd_set,
> &max_fd);
> if ((res != CURLM_OK) || (max_fd < 0)) {
> CURLERROR_MSG("curl_multi_fdset() failed due to internal
> errors");
> }
>
>
> In particular, curl_multi_fdset returning a max_fd of -1, is *not* an
> error. It just indicates that the fdset is empty. Just removing the
> check (max_fd < 0), fixes the problem.
>
> But, this seems to indicate some deeper problem in libcurl/c-ares
> itself. Why are there no sockets being returned by 'curl_multi_fdset'
> when an operation is still in progress???
>
> In fact the same code used to work curl 7.15.2 (and c-ares 1.3.0), but
> only with the latest version i'm facing this issue.
>
> Regards
> Sreeram
>
>
>
> _______________________________________________
> http://cool.haxx.se/cgi-bin/mailman/listinfo/curl-and-python
>
>
>
>
_______________________________________________
http://cool.haxx.se/cgi-bin/mailman/listinfo/curl-and-python
Received on 2006-06-27