[PATCH evolve_ext V2] fold: take an explicit list of revisions (BC)

[Pierre-Yves David]

On 06/26/2014 07:05 PM, Jordi Gutiérrez Hermoso wrote:
> On Thu, 2014-06-26 at 08:41 -0400, Greg Ward wrote:
>> On 24 June 2014, Jordi Gutiérrez Hermoso said:
>
>>> fold: rehaul handling of multiple and single revisions (BC)
>>          ^^^^^^
>>
>> "overhaul"
>
> Oops, how did I think that "rehaul" was a word?
>
>> *IF* we accept Pierre-Yves' judgement that "specified revision and all
>> of its descendants up to ." should be the default, then there is no
>> need to accept multiple revisions.
>
> There is, actually, because one thing you might reasonably want to do
> is `hg fold -r 'draft()'` or other revsets that might result in
> multiple revisions.

+1 on what JordiGH says. The old behavior of histedit (who insisted in 
picking the head of the revset as a the value) was a major pain. The 
current behavior (without --rev).

As the benefit to be very versatile:, fold with ancestors, fold with 
descendants, fold with both, fold with revset (as long it contains ".").

>>> +    If specifying multiple revisions, use --exact for folding those
>>> +    revisions while ignoring the parent of the working directory. In
>>> +    this case, the given revisions must form a linear unbroken chain.
>>
>> No. Yuck. Confusing. Do not like.
>
>  Sadly, we can't afford usability studies to decide which way to
> go about this.

Can't we? Facebook have a ##### engineers and we are logging every 
single action they do with Mercurial. Not enough data on fold yet, but 
we'll be able to use those those data before freezing the UI.

-- 
Pierre-Yves David