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

Gwern Branwen gwern0 at gmail.com
Fri Aug 28 05:52:24 UTC 2009


On Mon, Aug 10, 2009 at 1:44 AM, Petr Rockai<me at mornfall.net> wrote:
> But that would be wrong! HLint can already do that, but we have to ratify
> *instances* of the problem, not files. What you propose would disable the check
> for all of the file, and that's naturally undesirable.
>
> The fact the ratification mechanism is ugly is OK -- it we found ourselves
> ratifying too many things, it would mean that we are doing something wrong
> (i.e. that our policy is broken).
>
> Yours,
>   Petr.

Are there that many double or triple uses of the dangerous functions
in Darcs code? (Is the ratified module technique even foolproof? It
would seem as vulnerable to someone using the
Ratified.getContents/whatever 'just to get it to compile' as to
someone just reusing the function in a ignored module.)

Which is uglier, risking the small chance of missing a double use, or
supporting this ratified module stuff forever?

-- 
gwern


More information about the darcs-users mailing list