After http://trac.webkit.org/changeset/100558 NRWT exits before testing if the WebKit isn't is SVN or GIT SCM: For more parallelism, run new-run-webkit-tests directly. Traceback (most recent call last): File "/home/oszi/WebKit/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py", line 426, in <module> sys.exit(main()) File "/home/oszi/WebKit/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py", line 419, in main host._initialize_scm() File "/home/oszi/WebKit/Tools/Scripts/webkitpy/common/host.py", line 112, in _initialize_scm self._scm = detector.default_scm(patch_directories) File "/home/oszi/WebKit/Tools/Scripts/webkitpy/common/checkout/scm/detection.py", line 60, in default_scm raise Exception("FATAL: Failed to determine the SCM system for either %s or %s" % (cwd, script_directory)) Exception: FATAL: Failed to determine the SCM system for either /home/oszi/WebKit or /home/oszi/WebKit/Tools/Scripts/webkitpy/common/checkout/scm It would be great if we can run NRWT again without copying ~5Gb sized repository for an embedded system. I don't think if NRWT should depends on SCM.
Created attachment 123124 [details] workaround It is only a workaround (partial revert of r100558) to make NRWT work without GIT or SVN SCM. Feel free to use it until a proper fix.
No. I don't think we should support this.
I think not having a checkout just makes things more complicated. I don't think there is a lot of need to support running the tests w/o a full development environment.
I really don't understand why should NRWT depends on SCM if it doesn't use it for running tests ... Copying ~5Gb sized full repository to an embedded system is _impossible_. We really need WebKit lib, DumpRenderTree, Tools/Scripts/webkitpy, LayoutTests/platform/my-favourite-platform and LayoutTests directory without other platform results for testing. It is maximum 2-300Mb, which can be copied simple to a device. But now this crazy dependency is the only blocker.
I just run across a problem that when only git is installed and svn is NOT installed. I get a crash like the one supplied below. It's pretty hard to find out from that crash that missing subversion is the problem. alex@alex8540:~/projects/webkit-desk$ python Tools/Scripts/new-run-webkit-tests Traceback (most recent call last): File "/home/alex/projects/webkit-desk/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py", line 452, in <module> sys.exit(main()) File "/home/alex/projects/webkit-desk/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py", line 444, in main host._initialize_scm() File "/home/alex/projects/webkit-desk/Tools/Scripts/webkitpy/common/host.py", line 112, in _initialize_scm self._scm = detector.default_scm(patch_directories) File "/home/alex/projects/webkit-desk/Tools/Scripts/webkitpy/common/checkout/scm/detection.py", line 53, in default_scm scm_system = self.detect_scm_system(cwd, patch_directories) File "/home/alex/projects/webkit-desk/Tools/Scripts/webkitpy/common/checkout/scm/detection.py", line 69, in detect_scm_system if SVN.in_working_directory(absolute_path): File "/home/alex/projects/webkit-desk/Tools/Scripts/webkitpy/common/checkout/scm/svn.py", line 93, in in_working_directory exit_code = Executive().run_command(svn_info_args, cwd=path, return_exit_code=True) File "/home/alex/projects/webkit-desk/Tools/Scripts/webkitpy/common/system/executive.py", line 397, in run_command close_fds=self._should_close_fds()) File "/home/alex/projects/webkit-desk/Tools/Scripts/webkitpy/common/system/executive.py", line 453, in popen return subprocess.Popen(*args, **kwargs) File "/usr/lib/python2.7/subprocess.py", line 679, in __init__ errread, errwrite) File "/usr/lib/python2.7/subprocess.py", line 1249, in _execute_child raise child_exception
Created attachment 148215 [details] Patch
I've updated this patch (just merged to head and added a ChangeLog) and am landing it. NRWT is explicitly not supposed to be required to have a checkout, so I agree w/ Ossy here. Note that if we are in an SVN checkout and are configured to upload the build results we will extract the revision number if we can; that code is in json_results_generator, and it should deal with a lack-of-svn fine, though (it will probably raise an error if there is a .svn dir but you don't have an svn binary in your path, but that shouldn't happen in practice).
Committed r120646: <http://trac.webkit.org/changeset/120646>
*** Bug 87020 has been marked as a duplicate of this bug. ***