Joe Mistachkin wrote:
Please review and/or run tests against the test suite changes on the
"testerCleanup" branch. It's designed to give each test file its very
own isolated set of test directories (one for each call to repo_init).
It's also designed to cleanup after itself. Finally, it is designed
so that it no longer matters if it's run from an open checkout. For
now, I've left the "no open checkout" restrictions in place for the
specific test files that have it; however, technically it no longer
matters.
Thanks for doing all this, Joe. This is great.
I ran the full test suite on Windows, and it errored out on the
test_cleanup call at the end of th1-repo.test. Testing just that file by
itself worked fine. Huh. Figured out it was because I didn't have any of
the optional TH1 features compiled in, so th1-docs.test was exiting
after test_setup had already been called but without calling
test_cleanup. An error gets thrown in the next script that calls
test_cleanup, which in this case was th1-repo. Swapping the test_setup
and feature support checks (as they are in th1-hooks.test) fixed this.
One thing I thought about is having test_cleanup always return with
"-code return". Then you can just call test_cleanup to exit a test
script at any time, and you don't have to care whether you've already
called test_setup or not. I'll let you or Richard decide if that's too
cute or not, though.
Thanks again!
_______________________________________________
fossil-dev mailing list
[email protected]
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/fossil-dev
