Re: [Jim Blandy <jimb@redhat.com>] RFA: Add tests for lookup_symbol_aux bug

[David Carlton <carlton@math.stanford.edu>]

On Sat, 21 Dec 2002 00:59:35 -0600, Michael Elizabeth Chastain <mec@shout.net> said:

> It feels a little strange to have a test which will PASS almost all the
> time, even if gdb is broken.

For what it's worth, this test has already caught an unrelated bug, so
it's having surprising benefits.  I think probably the test suite
doesn't pound on the partial symbol table as much as it could, so even
narrowly targeted partial symbol tests like this one are useful.

> (For instance, we could add some maint commands to gdb to force
> specific algorithmic paths in the symbol table reader so that test
> scripts can exercise them).

That sounds like an interesting idea.  I do think that unit tests are
useful, and unfortunately some parts of the symbol table (especially
partial symbols) are buried a bit too deep in GDB's innards to be able
to easily do unit testing on via the test suite.  Maybe maint commands
would help with that.

It would take some thinking to do well, though...

David Carlton
carlton@math.stanford.edu