Hi All,

After I saw John Wiegley's chat with Sacha Chua a few months ago I
decided to try writing some tests for Emacs and after going off on a
few tangents along the way I've come up with the following:

    - kmacro-tests.el, tests for kmacro.el
    - should-call, a context-creating macro which can temporarily
      add advice to one or more named functions around the execution
      of some test code, and its close relative, should-not-call
    - autocover.el, a batch-mode code coverage tool

Originally I thought it would be nice to see code coverage summarized
as percentages instead of having to eyeball pink and tan splotches,
then I got the idea to see if I could get it to run on all of Emacs's
tests in batch mode. A substantial proportion of the .el files loaded
by the test suite show buggy behavior under testcover, so I ended up
making a list of over 100 files to exclude from code coverage. I've
attached the results of my "make check-coverage" for the curious, but
the testcover-related bugs need to be found and fixed before this is
going to be useful for anyone. I have a couple of them narrowed down
to nice reproducible cases and reported, but there are definitely
more.

I wrote should-call to make writing tests for kmacro.el easier. Advice
is a really useful tool when writing tests, but making sure each
advice-add is matched by an advice-remove inside an unwind-protect is a
chore. I think other test writers would find should-call useful so I'd
like to propose it for inclusion in ert. Is sending it to bug-gnu-emacs
labeled WIP and PATCH like the recent lispy backtrace thread the best
way to do that?

Best regards,
Gemini