[Stats] Add a test for long input-path bug (0e5b982d) #12279
Conversation
| // RUN: mkdir -p %t/very-long-directory-name-that-contains-over-128-characters-which-is-not-enough-to-be-illegal-on-its-own-but-presents | ||
| // RUN: cp %s %t/very-long-directory-name-that-contains-over-128-characters-which-is-not-enough-to-be-illegal-on-its-own-but-presents/a-problem-when-combined-with-other-long-path-elements-and-filenames-to-exceed-256-characters-combined-because-yay-arbitrary-limits-amirite.swift | ||
| // RUN: touch %t/main.swift | ||
| // RUN: %target-swiftc_driver -o %t/main -module-name main -stats-output-dir %t %t/main.swift %t/very-long-directory-name-that-contains-over-128-characters-which-is-not-enough-to-be-illegal-on-its-own-but-presents/a-problem-when-combined-with-other-long-path-elements-and-filenames-to-exceed-256-characters-combined-because-yay-arbitrary-limits-amirite.swift |
There was a problem hiding this comment.
Why %target-swiftc_driver rather than %target-build-swift?
There was a problem hiding this comment.
I'm not clear on the difference! Does it matter here?
There was a problem hiding this comment.
I guess it probably doesn't. %target-build-swift includes things like "pass -O when testing optimized", and that's not so relevant for this test. I mostly asked because I just default to %target-build-swift and wondered why you didn't.
| @@ -0,0 +1,16 @@ | |||
| // REQUIRES: OS=macosx | |||
| // RUN: rm -rf %t && mkdir -p %t | |||
| // RUN: mkdir -p %t/very-long-directory-name-that-contains-over-128-characters-which-is-not-enough-to-be-illegal-on-its-own-but-presents | |||
There was a problem hiding this comment.
I assume it doesn't matter that the total path at this point may or may not be over 256?
There was a problem hiding this comment.
nope. we're just trying to force it to be >256 when combined with the file and mangled into a string that has no path separators, so we need at least two big segments in the combined path. Extra big segments in %t will just make it fail more-thoroughly.
| // RUN: rm -rf %t && mkdir -p %t | ||
| // RUN: mkdir -p %t/very-long-directory-name-that-contains-over-128-characters-which-is-not-enough-to-be-illegal-on-its-own-but-presents | ||
| // RUN: cp %s %t/very-long-directory-name-that-contains-over-128-characters-which-is-not-enough-to-be-illegal-on-its-own-but-presents/a-problem-when-combined-with-other-long-path-elements-and-filenames-to-exceed-256-characters-combined-because-yay-arbitrary-limits-amirite.swift | ||
| // RUN: touch %t/main.swift |
There was a problem hiding this comment.
Why have a main file at all?
There was a problem hiding this comment.
Need to get the frontend invoked with a designated primary-file (not just 'all'), and this is the nearest to how it happens in the field: a multi-file compilation.
|
@swift-ci please smoke test and merge |
|
@swift-ci please smoke test Linux platform |
|
@swift-ci please clean smoke test Linux platform |
Add a testcase for the bug in 0e5b982 where we were forming over-long stats-dir output filenames.