Re: keep on patchin'

On Tue, 18 Aug 2020 at 04:10, Jim Warner <[email protected]> wrote:

> That was a big Oops on my part.
>
It's probably one of all the first things we learn in C and yet, we all
still do it.

C takes no prisoners.


> However, your commit goes way too far in extending these tests to
> include successful 'new' and 'unref' calls rather than simply
> **surviving** the totally independent enum/table validations.
>
It depends on how you view the new() and unref() calls. If they are part of
the tests, then they should return fails if they fail.
If they are not part of the test but merely test infrastructure (the real
tests being inside the library) then they shouldn't return fails.


> Anyway, here is a fix for your fix that once again allows for including
> the slabtop test.
>
That's now been pushed.

p.s. Do you think it will be possible to extend 'make check' to include
> this new test?
>
make check already includes it because its in the check_PROGRAMS
definition.  When it was dying with the _new() that was me running the make
check.

However, at the moment we are not defining ITEMTABLE_DEBUG so its not
really doing much.

I think the simplest way would be if you run make check then it makes sure
the library is configured with that define set, that would do what's
needed, wouldn't it?

Just need to work out how to do that.

 - Craig