Bug 37555
| Summary: | new-run-webkit-tests --results-directory should be an absolute path | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Eric Seidel (no email) <eric> |
| Component: | Tools / Tests | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED FIXED | ||
| Severity: | Normal | CC: | abarth, cjerdonek, dpranke |
| Priority: | P2 | ||
| Version: | 528+ (Nightly build) | ||
| Hardware: | PC | ||
| OS: | OS X 10.5 | ||
Eric Seidel (no email)
new-run-webkit-tests --results-directory should be an absolute path
old-run-webkit-tests --results-directory takes an absolute path (or at least supports them), and that seems to make more sense.
The existing build.webkit.org bot config scripts are used to being able to set this to an absolute path (or having it be CWD relative) instead of it being build-directory relative.
If we change this we need to be careful not to break the chromium bots which likely depend on the current behavior.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Eric Seidel (no email)
Actually, we can change how we interpret this argument for just the webkit port. See webkit.py:
def results_directory(self):
# Results are store relative to the built products to make it easy
# to have multiple copies of webkit checked out and built.
return self._build_path(self._options.results_directory)
Eric Seidel (no email)
Hmm.. and seems there is already code for absolute paths in run_webkit_tests.py:
if options.results_directory.startswith("/"):
# Assume it's an absolute path and normalize.
options.results_directory = port_obj.get_absolute_path(
options.results_directory)
Maybe I was just calling the script wrong.
Dirk Pranke
I think this has always worked, but last night in r69065 I actually checked in some unit tests to ensure this is working. Closing ...
http://trac.webkit.org/changeset/69065