Author giampaolo.rodola
Recipients akuchling, djarb, facundobatista, forest, giampaolo.rodola, gvanrossum, intgr, jafo, josiahcarlson, kevinwatters, markb
Date 2009-04-01.23:06:32
SpamBayes Score 6.66134e-15
Marked as misclassified No
Message-id <1238627195.21.0.435665486515.issue1641@psf.upfronthosting.co.za>
In-reply-to
Content
I don't know what are the problems experienced by the Zope folks (is
there a place where this is discussed?) but I can guess that they're
having problems with asynchat rather than with asyncore, since the
latter hasn't changed too much between 2.5 and 2.6 except for low level
connection related bug fixes.

The greatest difference in the new asynchat is that the producer_fifo
attribute is no longer an asyncore.fifo() instance but a deque().

Python 2.5:
> self.producer_fifo = fifo()

Python 2.6
> self.producer_fifo = deque()


Although they're quite similar the old code relying on the fifo() API
can't obviously work anymore.
This could have been a bad choice and there are probably other changes
that might have caused the problem (one other change that comes to my
mind is the different readable() writable() implementation).

An alternative to completely reverting asynchat.py to the 2.5 version,
which is somewhat too drastic IMO, could be identifying what are the
changes that caused the incompatibility, and reverting those parts only
for 2.6.2, in a way that no one (2.5 and 2.6 users) is affected.

If there's a place where this is discussed I could contribute in some
way since I've been working on asynchat/asyncore for a long time now.
History
Date User Action Args
2009-04-01 23:06:36giampaolo.rodolasetrecipients: + giampaolo.rodola, gvanrossum, akuchling, facundobatista, jafo, josiahcarlson, forest, kevinwatters, djarb, markb, intgr
2009-04-01 23:06:35giampaolo.rodolasetmessageid: <1238627195.21.0.435665486515.issue1641@psf.upfronthosting.co.za>
2009-04-01 23:06:33giampaolo.rodolalinkissue1641 messages
2009-04-01 23:06:32giampaolo.rodolacreate