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 rishi.maker.forum
Recipients BreamoreBoy, Chui.Tey, flox, hynek, ishimoto, orsenthil, pitrou, r.david.murray, rishi.maker.forum, serhiy.storchaka, teyc
Date 2014-11-08.14:44:24
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1415457868.02.0.288036673935.issue1610654@psf.upfronthosting.co.za>
In-reply-to
Content
Hi,
I have created a new patch with a small design change. The change is that in situations where I don't find the boundary instead of keeping the last x bytes in the buffer I simply drain the whole data and call a readline(). 
This seems like the right thing to do also. I managed to get rid of the two obfuscated helper functions keep_x_buffer and remove_x_buffer that I had and the code should look familiar (I hope) to a module owner.
This also helped me get rid of quite a few class member variables that I could move to locals of my main function(multi_read).
I still need to maintain an overlap, but only for the trailing CRLF boundary. 
Ran all the new and old tests and tested on apache with the ubuntu iso server image. Without the patch ubuntu iso server image took 93seconds .. with the patch it took 25seconds.
History
Date User Action Args
2014-11-08 14:44:28rishi.maker.forumsetrecipients: + rishi.maker.forum, ishimoto, orsenthil, pitrou, teyc, r.david.murray, flox, BreamoreBoy, hynek, serhiy.storchaka, Chui.Tey
2014-11-08 14:44:28rishi.maker.forumsetmessageid: <1415457868.02.0.288036673935.issue1610654@psf.upfronthosting.co.za>
2014-11-08 14:44:27rishi.maker.forumlinkissue1610654 messages
2014-11-08 14:44:27rishi.maker.forumcreate