[darcs-users] darcs patch: Re-implement haskell_policy in terms of hlint.

Trent W. Buck trentbuck at gmail.com
Sat Aug 8 10:38:53 UTC 2009


Petr Rockai <me at mornfall.net> writes:

> this is an alternative "hlint" patch, that actually replaces
> haskell_policy. It only catches and reports actual errors, so there's
> no issue with huge output. It also provides a ratification mechanism,
> so it is possible to make the test pass (we need to use the unsafe
> functions from time to time).

I note that the ratification technique Petr employs can only be used to
ratify module hiding-type errors.  For example, if we wanted hlint to
complain about only *new* code that didn't use camelCase, we wouldn't be
able to use it to ratify existing not_camel_case functions.

For that reason I'm a little unhappy about that particular technique,
but as 1) I don't have a better suggestion; and 2) it gives the same
functionality that we had in the old haskell_policy, I'll give my
approval of it.

I suggest filing a wishlist bug against hlint for some sort of
generalized ratification mechanism, since it seems like something that'd
be useful for ALL hlint users, not just Darcs.

> Nevertheless, even though the test passes now, it should probably wait
> till the hlint cpp issue [1] is fixed.
>
> [1]: http://code.google.com/p/ndmitchell/issues/detail?id=137

I had forgotten about this one.  Do you know offhand how many files in
the Darcs tree will bork current hlint in this way?  Ah, you say in the
ticket:

>> about 30 or so files fail to parse due to CPP.

Or, about one in five source files (19%).



More information about the darcs-users mailing list