This reduces the bloat in the source tree and makes the tests more consistent.
Can you explain this a bit? I thought we were going to output them to the
lldb-test-traces folder, or whatever location was specified via -s to
dotest. I have a very strong dislike for introducing environment variable
dependencies into the test suite. And this seems to still ignore -s.
This environment variable is already used to communicate info about the session dir.
In dotest.py the variable is set (according to the -s option).
Then lldbtest.py takes it and uses it to output its own logs.
I just copied these two lines from there.
Potentially I could encapsulate it better, put it in a function or something.
Ahh, so just to be clear, both variables are already set, and you're just
reading them? If so that's probably ok.
What do the filenames of the test outputs look like? Is there anything in
the filename to indicate it's a remote test case?
Yes, I am reusing existing variables.
The files look like:
No indication of remoteness present.
The other files placed in the folder include the compiler used and architecture (eg. ExpectedFailure-x86_64-_usr_bin_clang-TestConcurrentEvents.ConcurrentEventsTestCase.test_signal_watch_dwarf.log), but still nothing remote.