[Haskell-cafe] Re: GSoC: Improving Cabal's Test Support

[Ivan Miljenovic]

On 7 April 2010 10:28, Thomas Tuegel <ttuegel at gmail.com> wrote:
> Ok, this is the bottom-line that I didn't understand after our first
> exchange, but I think now I do: I should entirely scrap the second
> aspect of my proposal and focus exclusively on making Cabal build and
> run test programs.

Just as a food for thought: how would you deal with test programs that
deal with user input?  Should "cabal test" only run those that are
automatic?

The reason I'm asking is that I'm developing a second test executable
for graphviz that checks if it can parse "real world" Dot files.  Now,
I could have it do a search for all .dot and .gv files on the machine
in question and try those, except not all of those might be Dot code
(just something with the same extension); as such the user has to
explicitly pass a list of filenames to the program.  This is thus more
of a guided test program rather than something that should be run
automatically.

Also, I'm not sure if this is come up yet in your proposal: what about
installation?  One thing that Sergei (aka trofi) is trying to add in
the Gentoo ebuilds for Haskell packages is something similar to this
where tests are optionally built and run; the problem is that if the
tests reside in an optional executable then the installation stage
will also install these test executables because Cabal has no way of
saying "build these but don't install them".  This is probably going
to be the case for your proposed project as well: should test
executables be installed?

-- 
Ivan Lazar Miljenovic
Ivan.Miljenovic at gmail.com
IvanMiljenovic.wordpress.com