This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author giampaolo.rodola
Recipients giampaolo.rodola, josiah.carlson, josiahcarlson, loewis, michael.foord, pitrou, r.david.murray
Date 2010-05-08.11:15:20
SpamBayes Score 0.00015307454
Marked as misclassified No
Message-id <1273317326.07.0.710924994582.issue8490@psf.upfronthosting.co.za>
In-reply-to
Content
I have adjusted the failing tests in r80895 and r80930 which now pass on Solaris. 
It seems Solaris has two problems:

 - OOB data (aka handle_expt) doesn't work, but I'm not too worried about this as it's very rarely used. In past there have been proposals to remove its support, so I'd be for leaving it as it is.

- connect() connects immediately, even before starting the poller.
Now, this may be a problem when writing clients which implement some logic in their handle_connect() method.
Since handle_connect() gets called immediately if the user tries to, say, send some data over the socket, the dispatcher will immediately be closed as the socket is actually still disconnected.

I'd be for starting to commit the patch in attachment which skips such tests on SunOS, then if someone could give me SSH access over the Solaris box used for buildbots I could try to debug the connect() problem and see if I can fix it.
History
Date User Action Args
2010-05-08 11:15:26giampaolo.rodolasetrecipients: + giampaolo.rodola, loewis, josiahcarlson, pitrou, josiah.carlson, r.david.murray, michael.foord
2010-05-08 11:15:26giampaolo.rodolasetmessageid: <1273317326.07.0.710924994582.issue8490@psf.upfronthosting.co.za>
2010-05-08 11:15:24giampaolo.rodolalinkissue8490 messages
2010-05-08 11:15:23giampaolo.rodolacreate