Kamil Dudka wrote:
> Hello,
>
> I finally got some time to review the recent Guenter's patch for lib/nss.c. It
> looks good to me and it works fine with my test-suite [1]. I've just built it
> for rawhide.

What do you mean?

>
> While doing review of the patch, I spotted some nasty things in computing
> timeouts within Curl_nss_recv() and Curl_nss_send(). I wonder how this is
> supposed to work.

The timeout value is ignored in non-blocking mode by NSPR.

> If the socket works in non-blocking mode, why do we bother to compute some
> timeouts at all? If the socket works in blocking mode, we want perhaps to
> switch it to non-blocking mode.
>
> Any idea?
>
> Kamil
>
> [1] git clone http://kdudka.fedorapeople.org/git/nss.git

rob

-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html