Some of the test cases in test/file1.test require running inside an open
checkout, because the command "fossil test-tree-name" calls
file_tree_name() which requires an open checkout.
Before I touched anything, I see 22 errors out of 58 tests in file1.test.
I added a call to repo_init at the top of file1.test, and that
immediately drops to 17 errors out of 58 tests. Unfortunately, that
still includes nearly all of the cases that are based on "fossil
test-tree-name", both for relative and absolute file names.
It looks like these failures are a result of changes to what
file_tree_name() returns. I used fossil bisect to narrow down the point
where file1.test stops passing all tests. Checkin [74836bc8] passes 35
of 35 tests, and checkin [c56a387d] fails 22 of 58 as is, but whines
about needing a repository. Adding repo_init to its version of
file1.test calms that down to failing just 17 of 58 tests.
At this point I have ask whether the test cases in test/file1.test or
the code in src/file.c are correct?
My gut says that we all use fossil daily and no one is complaining about
strange side effects of incorrectly canonicalized file names, so it is
more likely the test cases that are wrong.
--
Ross Berteig r...@cheshireeng.com
Cheshire Engineering Corp. http://www.CheshireEng.com/
_______________________________________________
fossil-dev mailing list
fossil-dev@mailinglists.sqlite.org
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/fossil-dev
