On Thu, 20 Aug 2009, koettermarkus_at_gmx.de wrote:

>> libcurl doesn't respect SSL_ERROR_WANT_* at all, no, but the impact of not
>> respecting that is a short burst of too-much-looping and I've not yet seen
>> anyone badly damaged by that. This flaw has existed since forever in
>> libcurl.
>
> Recycling ssl connections without proper shutdown works?

Sorry, what is that referring to exactly? libcurl can re-use SSL connections,
yes, but that's not what you're asking about is it?

> Current openssl documentation is pretty complete to do all required actions,
> but it is a large, and easy to misunderstand.

I disagree. There are numerous undocumented or very scarcely documented
functions. Just look for example how the RAND_screen() issue appeared just a
short while ago: undocumented side-effects. But of course, it's just another
open source projects so it is nobody's and everybody's fault so we can all
help out and fix this and I can admit to not having done a lot about it.

>> Really? NSS and gnutls also provide fine non-blocking functionality.
>
> I've never seen gnutls nonblocking code working

Apart from the libcurl implementation limitations, what's not working
non-blocking in our current gnutls layer?

-- 
  / daniel.haxx.se