Issue1182
Created on 2008-11-24.17:14:36 by dmbaggett, last changed 2008-12-18.11:06:45 by amak.
File name |
Uploaded |
Description |
Edit |
Remove |
socket.py.diff
|
dmbaggett,
2008-11-24.17:14:35
|
Patch against trunk of 24NOV08 |
|
|
msg3845 (view) |
Author: Dave Baggett (dmbaggett) |
Date: 2008-11-24.17:14:35 |
|
This patch (against trunk as of 24NOV2008) corrects completely broken
behavior in SSL sockets.
|
msg3851 (view) |
Author: Alan Kennedy (amak) |
Date: 2008-11-26.12:34:27 |
|
Please can you elaborate on what is "completely broken" before we fix it?
A short code snippet illustrating defective behaviour would be enough.
|
msg3852 (view) |
Author: Dave Baggett (dmbaggett) |
Date: 2008-11-26.12:46:29 |
|
The socket code for ssl was creating a new
BufferedInputStream/BufferedOutputStream for each read or write.
If, for instance, you read one byte, it would actually read up to the
buffer size and return you the one byte, while throwing away the buffer
containing anything else that had been read. So any program that read
from an SSL socket would fail.
|
msg3854 (view) |
Author: Alan Kennedy (amak) |
Date: 2008-11-26.16:01:52 |
|
Yes, you're exactly right.
I will apply your patch very soon.
Thanks for submitting the bug report.
|
msg3960 (view) |
Author: Alan Kennedy (amak) |
Date: 2008-12-18.11:06:44 |
|
This is now fixed on trunk and release 22 maint.
Thanks for the report and the patch.
|
|
Date |
User |
Action |
Args |
2008-12-18 11:06:45 | amak | set | status: open -> closed assignee: amak resolution: fixed messages:
+ msg3960 |
2008-11-26 16:01:53 | amak | set | messages:
+ msg3854 |
2008-11-26 12:46:29 | dmbaggett | set | messages:
+ msg3852 |
2008-11-26 12:34:28 | amak | set | nosy:
+ amak messages:
+ msg3851 |
2008-11-24 17:14:36 | dmbaggett | create | |
|