[PATCH 2 of 2 V2] histedit: add experimental config for using the first word of the commit

Pierre-Yves David pierre-yves.david at ens-lyon.org
Fri May 27 10:50:50 EDT 2016



On 05/27/2016 04:45 PM, Augie Fackler wrote:
> On Fri, May 27, 2016 at 10:37 AM, Pierre-Yves David
> <pierre-yves.david at ens-lyon.org> wrote:
>>
>>
>> On 05/27/2016 03:28 AM, Augie Fackler wrote:
>>> On Tue, May 24, 2016 at 05:07:37PM -0400, timeless wrote:
>>>> On Tue, May 24, 2016 at 4:51 PM, Martin von Zweigbergk
>>>> <martinvonz at google.com> wrote:
>>>>> One problem with basing it on sha is that you can't histedit after rebase
>>>>> without having to the assignment manually. Maybe not very important.
>>>>
>>>> Since I was doing this manually, it didn't matter. But with evolution
>>>> markers, we could pick the successor ...
>>>
>>> I'm open to trying things as shas first, and if we can't figure that
>>> out trying it as a prefix match on other commits in the histedited
>>> set. Seems worth some experimentation.
>>
>> We could also use revset, a bit more verbose but much more versatile.
>> (not entirely convinced myself because of the verboseness).
>>
>> On the same topic, I'm not super fan of the "action!" notation it's a
>> bit black magic (and could lead to collision quite easily). I'll try to
>> think about something a bit clearer for an alternative.
> 
> I agree it feels a little magical, but it's the exact syntax used by
> git for this, so it'll have to be an extremely compelling argument to
> convince me to change things.

Ha, yes, that would be convenient. Maybe we recognize two syntaxes, one
for git people and one for human being.

-- 
Pierre-Yves David


More information about the Mercurial-devel mailing list