I would like to make a plea for a validation suite for Haskell.
One of the best specified procedural languages is Pascal, which is similar in
size to Haskell. The Pascal standard defines Pascal rather tighter than the
Haskell report defines Haskell, but in practice the limiting cases are defined
by a suite of several hundred (mostly pathological) test programs.
Such a suite serves three important purposes. First it helps define the
language: I believe I could come up with dozens of programs where I can
honestly say that I can't tell what the Haskell report intends, and probably
dozens more on which other people (particularly compiler writers) would
differ. Second, it helps to interpret a language standard: the tighter you
make a language standard, the more difficult it is to understand the
consequences, and a validation suite helps avoid misunderstandings. Third, it
provides tests for compiler writers. In my bitter experience, a compiler
which you think is right tends to fail about half of the tests in a suite.
I am generating and collecting examples for Brisk, but not very systematically
so far. I am willing for Bristol to become the archiver for such a suite, if
people agree. Send us your pathological programs, preferably short ones which
make a point.
Ian [EMAIL PROTECTED], Tel: 0272 303334
