[darcs-users] [patch65] remove commutex

Ganesh Sittampalam ganesh at earth.li
Wed Nov 25 05:36:38 UTC 2009


On Tue, 24 Nov 2009, Jason Dagit wrote:

> On Sun, Nov 15, 2009 at 8:15 AM, Ganesh Sittampalam <ganesh at earth.li> wrote:
>
>> On Sun, 15 Nov 2009, Jason Dagit wrote:
>>
>>  toCommute and fromCommute are not the best function names, but I think
>>> a future refactor can get rid of both of them or give them better
>>> names.  I've also tried to make their types sufficiently general such
>>> that they may be able to transform the type of commuteFL, although I
>>> haven't actually tried it.
>>>
>>
>> A little while ago I started adding some general infrastructure for
>> handling commutes between different types of patches in
>> Darcs.Patch.Permutations - in particular there's
>>
>> type CommuteFn p1 p2 = FORALL(x y) (p1 :> p2) C(x y) -> Maybe ((p2 :> p1)
>> C(x y))
>>
>> I think it would make sense to combine this stuff with your toCommute and
>> fromCommute code, and put it in an appropriate module (perhaps a new one).
>>
>
> Ganesh,
>
> It's unclear to me:  Were you going to apply this?  I see it hasn't made it
> into darcs.net yet, and the status is "amend-requested"

it's "review-in-progress" - I've had a quick look but not finished yet.

> but I thought you
> were commenting on future work above.  Did you feel that I need to include a
> usage of CommuteFn at the same time?

No, but I didn't really like the toCommute/fromCommute names at all, so 
was erring towards asking that they be changed before it's applied. Hadn't 
thought of good alternatives yet which is part of the reason I haven't 
finished the review yet.

> It looks to me like there is a bit of thinking to be done about unifying the
> above and refactoring it properly.  I agree it needs to happen, but I'd
> prefer it to be independent of the removal of commutex (mainly because that
> bit has already been done, and the refactor of the types would delay it
> getting into darcs).
>
> What do you think?

Yes, I agree with that part.

Ganesh


More information about the darcs-users mailing list