RESOLVED FIXED 103699
webkitpy: autoinstaller is failing on windows
https://bugs.webkit.org/show_bug.cgi?id=103699
Summary webkitpy: autoinstaller is failing on windows
Dirk Pranke
Reported 2012-11-29 18:22:49 PST
webkitpy: autoinstaller is failing on windows
Attachments
Patch (1.62 KB, patch)
2012-11-29 18:23 PST, Dirk Pranke
no flags
Patch (6.28 KB, patch)
2012-11-30 16:45 PST, Dirk Pranke
no flags
fix ChangeLog (6.48 KB, patch)
2012-11-30 16:48 PST, Dirk Pranke
tony: review+
webkit.review.bot: commit-queue-
Dirk Pranke
Comment 1 2012-11-29 18:23:18 PST
Dirk Pranke
Comment 2 2012-11-29 18:24:14 PST
Dirk Pranke
Comment 3 2012-11-29 18:25:25 PST
Landing this unreviewed since the python win unit tests are failing, but I wouldn't mind a review anyway. I *think* this fix is correct and safe :).
Dirk Pranke
Comment 4 2012-11-30 16:43:17 PST
still broken with these patches .. the problem is worse than I had hoped.
Dirk Pranke
Comment 5 2012-11-30 16:45:07 PST
Dirk Pranke
Comment 6 2012-11-30 16:47:13 PST
Given that this is my third attempt at making this work, can someone double-check my logic? I had to modify the code so that nested modules could be specified portably (as "logilab/common") but be flattened either to a safe basename (e.g. "logilab_common" for the downloaded url markers or for scratch dirs) or to a host-specific subdir (e.g. "logilab\common"). Then I had to also handle the stupid write-protected files in the archives :(.
Dirk Pranke
Comment 7 2012-11-30 16:48:34 PST
Created attachment 177054 [details] fix ChangeLog
WebKit Review Bot
Comment 8 2012-12-02 04:51:15 PST
Comment on attachment 177054 [details] fix ChangeLog Attachment 177054 [details] did not pass chromium-ews (chromium-xvfb): Output: http://queues.webkit.org/results/15117027 New failing tests: inspector-protocol/debugger-terminate-dedicated-worker-while-paused.html
Tony Chang
Comment 9 2012-12-03 13:58:42 PST
Comment on attachment 177054 [details] fix ChangeLog I would probably just not allow / to be in the target_name, but whatevs.
Dirk Pranke
Comment 10 2012-12-03 14:27:34 PST
(In reply to comment #9) > (From update of attachment 177054 [details]) > I would probably just not allow / to be in the target_name, but whatevs. I suppose an alternative would be to have the target name use periods as separators (to mirror the python module naming) rather than forward slashes, but I imagine that would impose its own set of conversions ...
Dirk Pranke
Comment 11 2012-12-03 14:50:45 PST
Note You need to log in before you can comment on or make changes to this bug.