[darcs-users] Leading / stripped from local repository

[Jason Dagit]

On Sat, Aug 30, 2008 at 1:58 AM, Eric Y. Kow <eric.kow at gmail.com> wrote:

>  I'm not convinced that System.FilePath is a route we'd want to go.  We
>  already have code that is almost certainly better tested than
>  System.FilePath, and that we understand.
>
> This is true; our path handling code has been battle tested in a real
> world darcs setting, whereas System.FilePath not so much.  On the other
> hand, System.FilePath has been around for quite some time now -- notice
> that this comment was written in 2006 -- and was written by somebody
> that certainly understands Windows better than us and who also provided
> an extensive battery of QuickChecks.  The QuickChecks do not substitute
> for field testing, but they do give us a lot of extra confidence about
> this library.

In my mind, a switch like this is can only be good for us in the long
run.  The obvious drawback to switching is the introduction of
regressions.  I would like to think this issue comes down to one of
regression and risk management.

Would it be possible use QuickCheck and pit our homegrown functions
against System.FilePath to detect potential incompatibilities so that
we have a better understanding of how this change might introduce
regressions?

It might be a lot of work, but it would be kind of interesting to see
the results.

Jason