[MLton] should Date.{fromString,scan} allow space instead of zero?

Daniel C. Wang danwang@CS.Princeton.EDU
Mon, 06 Sep 2004 22:13:11 -0400


Stephen Weeks wrote:
{stuff deleted}
> 
> A single flag alleviates the complexity somewhat.  Although this would
> be our only use of it so far (except for maybe -ieee-fp, but I think
> that deserves a separate flag).

I agree, as a general principle however we should dump a certain class of 
deviances from the Basic spec into one flag. Lets call that class of 
deviances "improvements to the spec". :) It should be on by default!

Given that the Basis spec is not a perfect beast yet. I suspect that there 
will always be a few places where what the spec says is not exactly "the 
right thing". We should lump all these things into one flag, and by default 
MLton should do "the right thing". However, if you are in the mood to be a 
language/basis spec lawyer there should be a way to get strictly conforming 
behavior.

There's no point in checking that my code is strictly conforming to some 
random subset of the spec. So having lots of flags for each small 
deviation/improvement seems not worth anyone's trouble.

I guess, you also should run regression testing once with the flag on and 
once with it off.. so maybe this is a reason not to do this... if the extra 
testing doesn't bother you this is the way I'd go.