[issue32551] Zipfile & directory execution in 3.5.4 also adds the parent directory to sys.path

2018-05-15 Thread Petr Viktorin
Change by Petr Viktorin : -- resolution: -> fixed stage: commit review -> resolved status: open -> closed ___ Python tracker ___

[issue32551] Zipfile & directory execution in 3.5.4 also adds the parent directory to sys.path

2018-01-23 Thread Ned Batchelder
Ned Batchelder added the comment: I can confirm that 3.5.5rc1 fixes the problem I had. -- ___ Python tracker ___

[issue32551] Zipfile & directory execution in 3.5.4 also adds the parent directory to sys.path

2018-01-23 Thread Larry Hastings
Larry Hastings added the comment: New changeset 891c91d8d38848377a9f475242507510873eb9c3 by larryhastings (Nick Coghlan) in branch '3.5': [3.5] bpo-32551: Consistently configure sys.path[0] (#5197)

[issue32551] Zipfile & directory execution in 3.5.4 also adds the parent directory to sys.path

2018-01-19 Thread Nick Coghlan
Nick Coghlan added the comment: Thanks Larry - I've reduced the priority accordingly. While the NEWS entry sounds a bit complicated, the actual fix is nicely self-contained: it's all within Modules/main.c, and relates to how that manipulates sys.argv[0] during startup.

[issue32551] Zipfile & directory execution in 3.5.4 also adds the parent directory to sys.path

2018-01-19 Thread Larry Hastings
Larry Hastings added the comment: If you're certain it isn't a security bug, then please downgrade it from release blocker. I might permit a fix for it in 3.5.5 anyway, depending on how small it is, because nobody likes regressions. --

[issue32551] Zipfile & directory execution in 3.5.4 also adds the parent directory to sys.path

2018-01-16 Thread Nick Coghlan
Nick Coghlan added the comment: Good catch Eryk, I misdiagnosed what was going on, since the current directory and the parent directory were the same location in Ned's particular example. I double checked, and we resolve symlinks in path entries *before* performing the