Sat, 5 Jul 1997 12:20:36 +0100
Michel Lavaud writes:
> > In my view, a project to construct a test suite for every package on
> > CTAN is doomed to failure. (For some reason, my fingers kept wanting
> > to type `dodo' where I meant `doomed' ;-)
> My proposal was not to construct a test suite for every package.
> It was: IF somebody is able to give an opinion on one given package,
> then he has written a test file (otherwise, could he have any
I'm readily capable of giving opinions on many packages I've used, but
I'm not about to give away my (often private) documents that
constitute the only tests I've made of them.
> So, instead of giving an opinion, rather provide the test
> file. Or, if he is enthusiast about it, maybe a demo file that
> illustrates some useful aspects of the package?
I would far rather have an informed opinion than a `test program'.
Frankly, I don't know many people who _do_ produce test programs in
the course of deciding that a package is a `good thing': I certainly
don't -- I read the source (if I have time) and then start
constructing my document that uses it.
I really do not believe that an exercise to produce test suites/demo
programs for packages is effort well spent.