Re: Build failure tooling improvements

[Greg Troxel <gdt%lexort.com@localhost>]

Rob Whitlock <rwhitlock22%gmail.com@localhost> writes:

> I was thinking something that would just start the program, quit
> the program, and check for a core file. But even that might be
> challenging to do for graphical apps, which incidentally is where
> most of my concern lays.

We cannot rely on the build environment having an X server.
That doesn't mean that a test that needs one is never useful, but it
makes things harder.

>> but:
>> 
>>  - it's a lot of work to write these
>
> Yes, the more I think about it, the more I think it would be really
> easy to slide down the slippery slope of implementing a more-or-
> less full test suite framework. But the intention is to have a
> dirt-simple per-package implementation and a simple pkgsrc
> infrastructure implementation. If this cannot be done, it would
> probably be better to not have a smoke-test target and just
> have regular tests from upstream.
>
>>  - I don't think upstreams are going to want to add these
>>  - it won't work with cross builds
>>  - it will be testing the binaries in the build tree, using the
>>    package's mechanisms to find the built (and not installed shlibs)
>
> Good points.

So for me, the current pkgsrc doctrine of wrapping the upstream tests
remains sound.

I do understand what you are wanting, but I think it's too hard for the
benefit.   There's nothing stopping you from adding a smoke_test: target
to any given Makefile and seeing if you find it useful, though.   But I
would guess that really it is best to install the package and run that.