On 22 July 2015 at 08:07, Robert Collins <robertc@robertcollins.net> wrote:
On 22 July 2015 at 05:08, Larry Hastings <larry@hastings.org> wrote:
On 07/21/2015 06:35 PM, Robert Collins wrote:
Cool. http://bugs.python.org/issue21750 is in a bad state right now.
I landed a patch to fix it, which when exposed to users had some defects. I'm working on a better patch now, but need to either roll the prior patch completely back, or get the new one landed before the next beta. I hope to have that up for review later today {fingers crossed} - will that be soon enough, or should I look up how to easily revert stuff out with hg?
If you want to undo it, "hg backout" is the command you want. In general it's best to not check in broken stuff.
Thanks. And yes, naturally - we didn't realise it was broken. Passing tests != fit for purpose.
21750 is now sorted out in the cpython repo. I have a separate question for you - issue2091 has a good patch on it, but would you like it added to 3.5? It makes a broken combination of file modes - rU+ - a clean error, and tweaks the existing exception text for U + writing modes. -Rob -- Robert Collins <rbtcollins@hp.com> Distinguished Technologist HP Converged Cloud