Teach package-root to package iOS sandbox profiles
Created attachment 340712 [details] Patch
I think that this is right for correctness tests and QA, but would not be acceptable for at least some of the perf tests.
See also: rdar://problem/40233366
(In reply to Alexey Proskuryakov from comment #2) > I think that this is right for correctness tests and QA, but would not be > acceptable for at least some of the perf tests. Sure, probably not. Are those tests run with roots generated by package-root? My understanding is that this is just a convenience script for engineering builds.
Not sure, would need to grep through internal code to check (not near a Mac today). I think that this script is used in infrastructure, but maybe not where performance is important.
(In reply to Alexey Proskuryakov from comment #5) > Not sure, would need to grep through internal code to check (not near a Mac > today). I think that this script is used in infrastructure, but maybe not > where performance is important. The only hits for 'package-root' in Internal and OpenSource are ChangeLogs and one internal script that names an output file 'package-root.txt' (but isn't running package-root itself).
Comment on attachment 340712 [details] Patch Attachment 340712 [details] did not pass win-ews (win): Output: http://webkit-queues.webkit.org/results/7726400 New failing tests: http/tests/preload/onload_event.html
Created attachment 340753 [details] Archive of layout-test-results from ews202 for win-future The attached test failures were seen while running run-webkit-tests on the win-ews. Bot: ews202 Port: win-future Platform: CYGWIN_NT-6.1-2.9.0-0.318-5-3-x86_64-64bit
I believe package-root was originally intended for engineers. If any test infrastructure is using it, it could have its own variant but I don’t suspect that is the case.
I'm fine with the idea of having this script package sandbox profiles. The implementation seems a bit like a hack though - I would expect a "package" script to only operate on the build directory, never on source. Of course, sandbox profiles aren't currently going to the build directory, but maybe they should. In fact, package-root already copies everything in /usr.
(In reply to Alexey Proskuryakov from comment #10) > I'm fine with the idea of having this script package sandbox profiles. > > The implementation seems a bit like a hack though - I would expect a > "package" script to only operate on the build directory, never on source. Of > course, sandbox profiles aren't currently going to the build directory, but > maybe they should. In fact, package-root already copies everything in /usr. Maybe the sandbox profiles should be copied somewhere as part of the build process, just like we do with header files and resources?
(In reply to Brent Fulgham from comment #11) > (In reply to Alexey Proskuryakov from comment #10) > > I'm fine with the idea of having this script package sandbox profiles. > > > > The implementation seems a bit like a hack though - I would expect a > > "package" script to only operate on the build directory, never on source. Of > > course, sandbox profiles aren't currently going to the build directory, but > > maybe they should. In fact, package-root already copies everything in /usr. > > Maybe the sandbox profiles should be copied somewhere as part of the build > process, just like we do with header files and resources? Sounds like a better plan. I'll do that.