>> * remove message output in getUniqueDPatchName
>
> I'm not sure we do always print out the final file name (e.g.
> in darcs obliterate -O)
Yes. I plan to fix that, showing the actual file name is pretty
essential for usability.
>> * simplify an expression in applyCmdCommon
>
> Did you swap the arguments of saveToBundle just so you could partially
> apply it? I find the new order more confusing but it's no big deal.
(This is actually in patch
* fix lazy reading of inventories for obliterate with --output)
Yes, ahem, looks like this was the reason. I tried to re-construct why I
did it like that and I think my thinking went as follows:
- okay, result of readRepo is only ever used once
- giving a name to something that will be immediately consumed is bad
- so use a combinator, hmm, >>= seems to be the one
- oh, order of arguments doesn't fit, use flip?
- no, flip is a bit obscure; where is saveToBundle used?
- this is the only call site, let's just change its argument order
I am curious: would you consider keeping the original order of arguments
and instead using flip the better solution in this case? Or would you
rather give a name to the intermediate PatchSet?
|