[MLton] Windows ports and paths

[Andreas Rossberg]

Wesley W. Terpstra wrote:
> 
> Instead of taking mkCanonical of the path arguement, our version just
> compares the canonName form of the paths. If we make it canonical like
> Andreas suggested, several regressions result, including some explicitly
> listed in the standard (eg: "/a/b/" + "/a/c/" -> "../b/" not "../b").

Indeed, that is a more conservative fix.

> I also special-cased concat (_, "") to behave as before.

Mh, why is that preferable, when it deviates from the spec?

   - Andreas

-- 
Andreas Rossberg, rossberg@ps.uni-sb.de

Let's get rid of those possible thingies!  -- TB