Bill Janssen wrote:
> Jean-Paul Calderone <exarkun at divmod.com> wrote:
>
>> On Wed, 17 Sep 2008 10:40:01 PDT, Bill Janssen <janssen at parc.com> wrote:
>>> Ah, now I remember. It seems that sometimes when SSL_ERROR_WANT_READ
>>> was returned, things would block; that is, the "handle_read" method on
>>> asyncore.dispatcher was never called again, so the SSLSocket.recv()
>>> method was never re-called. There are several levels of buffering going
>>> on, and I never figured out just why that was. This (very rare) re-call
>>> of "read" is to handle that.
>>>
>> You certainly do need to call read again if OpenSSL fails an SSL_read with
>> a want-read error, but in asyncore, you don't want to do it right away,
>> you want to wait until the socket becomes readable again, otherwise you *do*
>> block waiting for bytes from the network. See the SSL support in Twisted
>> for an example of the correct way to handle this.
>>
>> Jean-Paul
>
> Yes, I understand, and that's how I started out. The bug we were seeing
> was that "handle_read" wasn't being called again by asyncore.
It's probably worth sticking a comment in the code explaining why we're
taking the risk of temporarily blocking on a non-blocking socket (i.e.
until someone figures out how to reproduce that bug reliably so that a
more correct fix can be devised).
Cheers,
Nick.
--
Nick Coghlan | ncoghlan at gmail.com | Brisbane, Australia
---------------------------------------------------------------
http://www.boredomandlaziness.org
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4