Message162543
I point is that it shouldn't break at all if possible, although path_stats does potentially provide a way to deal with this (as does using keyword-only arguments for what needs to be verified).
I'll have to think about this. One issue I have with this is I don't love exposing bytecode files too much as people should not typically muck with bytecode files, and if they do they should really know what they are doing. I honestly view them as an implementation detail and not something for people to play with.
I also don't want people to get to into bytecode files as they are not necessarily the best way to store bytecode. E.g. a sqlite DB storing bytecode really shouldn't store all of that info in a single column but instead in separate columns and exposing a function as proposed kind of pushes against that idea. |
|
Date |
User |
Action |
Args |
2012-06-08 18:16:26 | brett.cannon | set | recipients:
+ brett.cannon, jcea, ncoghlan, Ronan.Lamy |
2012-06-08 18:16:26 | brett.cannon | set | messageid: <1339179386.14.0.0880911635747.issue15031@psf.upfronthosting.co.za> |
2012-06-08 18:16:25 | brett.cannon | link | issue15031 messages |
2012-06-08 18:16:25 | brett.cannon | create | |
|