On 06/29/2018 08:04 PM, David Malcolm wrote:
On Fri, 2018-06-29 at 13:19 -0400, Eric Gallager wrote:
On 6/29/18, Martin Liška <mli...@suse.cz> wrote:
Hi.

I would like to add some DejaGNU tests for completion option.

Ready for trunk?
Martin

Presumably the point of the DejaGnu tests is:
(a) to give us some integration testing, to complement the unit-tests
expressed via "selftest::" in the code
(b) in future, to allow for target-specific tests for the aspects of
completion (some target-specific options have non-standard arg-parsing,
iirc, so might need special-casing if the completion code is to be
perfect).  Also, it allows for other DejaGnu-based filters to be used.

Interesting idea to cover target specialities. I'll write that on my TODO
list. It's closely related to https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83193
where I plan some enhancements.


So, correct me if I'm wrong, but these tests would need to be updated
every time a new option starting like one of those is added, right?

You're correct.

If
so, can you think of any ways to ease the increased maintenance
efforts?

I think/hope Martin picked option prefixes for which the results are
relatively stable.

Exactly, it should be relatively stable over a time.

Martin


If it turns out to be a pain, other options can be chosen, but I guess
we can cross that bridge when we come to it.

Dave


Reply via email to