time critical: issue 7902 and 2.6.6rc2
For 2.6.6rc1 a patch was committed that addressed bug 7902: http://bugs.python.org/issue7902 http://svn.python.org/view?rev=81381&view=rev While this fixes a legitimate bug in Python 2.6, it also changes existing behavior in the 2.6 series. As 2.6.6 is the last planned maintenance release, I do not think we've had enough testing of this change and am not sure it should stay in. Note that we've already seen one fallout from this. Issue 9600 was fixed post rc1. While its use of the syntax was incorrect, and the fix was easy and legitimate, I wonder how many other cases in unknown third party code there might be. Antoine in IRC suggests it will be rare, and that the multiprocessing case was probably due to an incomplete code reorganization. I agree, but still. http://bugs.python.org/issue9600 Either way, I think we need a 2.6.6rc2 (a thought that's been invading my dreams all weekend :). This would push back 2.6.6 final to September 12 or 13. I'd like your opinion on whether the fix for 7902 should stay in 2.6.6 or not. Please respond asap, since *if* we revert this, I want that to go into 2.6.6rc2, and I'd like to tag that release in about 6 hours or so. Thanks, -Barry
Sounds to me like the fix is legit, and the bug it uncovered was a real bug and should have been caught. There really is no justification to consider it a feature -- the PEP is clear on that. So I think we should keep the fix. If it is a gray area, it is only gray because the implementation was imperfect, not because one could argue about whether it should be a feature. Anyone who wrote "from .os import walk" thinking it should work did not understand relative imports at all. But more likely such code only happened through some (semi-)automated editing and nobody thought much about it at all, and they would have preferred for it to fail in the first place. All in all it seems pretty clear. --Guido On Mon, Aug 16, 2010 at 8:54 AM, Barry Warsaw <barry@python.org> wrote:
For 2.6.6rc1 a patch was committed that addressed bug 7902:
http://bugs.python.org/issue7902
http://svn.python.org/view?rev=81381&view=rev
While this fixes a legitimate bug in Python 2.6, it also changes existing behavior in the 2.6 series. As 2.6.6 is the last planned maintenance release, I do not think we've had enough testing of this change and am not sure it should stay in.
Note that we've already seen one fallout from this. Issue 9600 was fixed post rc1. While its use of the syntax was incorrect, and the fix was easy and legitimate, I wonder how many other cases in unknown third party code there might be. Antoine in IRC suggests it will be rare, and that the multiprocessing case was probably due to an incomplete code reorganization. I agree, but still.
http://bugs.python.org/issue9600
Either way, I think we need a 2.6.6rc2 (a thought that's been invading my dreams all weekend :). This would push back 2.6.6 final to September 12 or 13.
I'd like your opinion on whether the fix for 7902 should stay in 2.6.6 or not. Please respond asap, since *if* we revert this, I want that to go into 2.6.6rc2, and I'd like to tag that release in about 6 hours or so.
Thanks, -Barry
_______________________________________________ Python-Dev mailing list Python-Dev@python.org http://mail.python.org/mailman/listinfo/python-dev Unsubscribe: http://mail.python.org/mailman/options/python-dev/guido%40python.org
-- --Guido van Rossum (python.org/~guido)
On Aug 16, 2010, at 09:56 AM, Guido van Rossum wrote:
Sounds to me like the fix is legit, and the bug it uncovered was a real bug and should have been caught. There really is no justification to consider it a feature -- the PEP is clear on that. So I think we should keep the fix.
Sounds good to me. Thanks for weighing in. -Barry
On Tue, Aug 17, 2010 at 5:15 AM, Barry Warsaw <barry@python.org> wrote:
On Aug 16, 2010, at 09:56 AM, Guido van Rossum wrote:
Sounds to me like the fix is legit, and the bug it uncovered was a real bug and should have been caught. There really is no justification to consider it a feature -- the PEP is clear on that. So I think we should keep the fix.
Sounds good to me. Thanks for weighing in.
A bit late (since you already cut rc2), but the key point for me is that code that is fixed to work correctly on 2.6.6 (where this bug has been fixed) will still work on any 2.6.x release. Cheers, Nick. -- Nick Coghlan | ncoghlan@gmail.com | Brisbane, Australia
participants (3)
-
Barry Warsaw -
Guido van Rossum -
Nick Coghlan