[PATCH 4 of 4] mq: introduce mq.check setting

[Patrick Mézard]

Le 12/05/12 17:56, Angel Ezquerra a écrit :
> On Sat, May 12, 2012 at 5:02 PM, Idan Kamara <idankk86 at gmail.com> wrote:
>> On Sat, May 12, 2012 at 5:55 PM, Angel Ezquerra <angel.ezquerra at gmail.com>
>> wrote:
>>>
>>>
>>> On May 12, 2012 4:49 PM, "Patrick Mézard" <pmezard at gmail.com> wrote:
>>>>
>>>> Le 12/05/12 16:13, Patrick Mézard a écrit :
>>>>> Le 12/05/12 15:00, Angel Ezquerra a écrit :
>>>>>> On Sat, May 12, 2012 at 11:00 AM, Patrick Mézard <patrick at mezard.eu>
>>>>>> wrote:
>>>>>>> Le 12/05/12 10:37, Angel Ezquerra a écrit :
>>>>>>>> On May 12, 2012 10:09 AM, "Matt Mackall" <mpm at selenic.com> wrote:
>>>>>>>>>
>>>>>>>>> On Sat, 2012-05-12 at 01:12 +0200, Patrick Mezard wrote:
>>>>>>>>>> # HG changeset patch
>>>>>>>>>> # User Patrick Mezard <patrick at mezard.eu>
>>>>>>>>>> # Date 1336774770 -7200
>>>>>>>>>> # Node ID ed81cb27341e285d539617ba961f48c69dd18135
>>>>>>>>>> # Parent  f4da2aeb000408aa54f59829acb092ec85914475
>>>>>>>>>> mq: introduce mq.check setting
>>>>>>>>>
>>>>>>>>> Nice, these are queued for default.
>>>>>>>>
>>>>>>>> Patrick,
>>>>>>>>
>>>>>>>> I'm curious, why did you choose "check" as the name for this
>>>>>>>> option? It
>>>>>>>> does not seem obvious to me...
>>>>>>>
>>>>>>> I have not given much thought to it. --check makes it sound like
>>>>>>> qpush/qpop will closely "check" the local changes before deciding to bail
>>>>>>> out or not. The setting is named after the command line option. Maybe this
>>>>>>> is a mistake, if you have something better to suggest, feel free to submit a
>>>>>>> patch, that is exactly the right time to change it.
>>>>>>
>>>>>> There is a --check (-c) option for update. Its definition is"
>>>>>>
>>>>>> -c --check     update across branches if no uncommitted changes
>>>>>>
>>>>>> It does not sound anything like this option (IMHO). In fact, it seems
>>>>>> as if where quite the oposite. On the hg help update text it says:
>>>>>>
>>>>>> "    2. With the -c/--check option, the update is aborted and the
>>>>>> uncommitted
>>>>>>        changes are preserved."
>>>>>>
>>>>>> That is, with the -c option, the update will be aborted if there are
>>>>>> uncommitted changes, which is in fact the opposite of what this does,
>>>>>> isn't it?
>>>>>
>>>>> You may be right.
>>>>>
>>>>>> what about --nocheck (-n)? After all this does the opposite thing of
>>>>>> what --check does for update (if I undestand things correctly).
>>>>>
>>>>> [-n] is used for --dry-run so not this short option. Obviously, it is
>>>>> difficult for me to consider --no-check, since the code actually checks more
>>>>> things than the regular version (but is more lenient in its findings). Also,
>>>>> --no-check feels a bit like "I am not checking anything, all bets are off, I
>>>>> hope you had backups". But maybe.
>>>>>
>>>>> What other people think?
>>>>
>>>> Idan: --disjoint
>>>> Me: --keep-changes or --keep
>>>>
>>>> --
>>>> Patrick Mézard
>>>>
>>>
>>> I like --keep. I don't really understand --disjoint.
>>
>> It describes more accurately what's happening: allow
>> qpush if patched files and dirty files in wd are disjoint.
>>
>> But --keep might be simpler and more familiar.
> 
> I see. That makes sense, but as you say --keep is probably easier to
> understand even if it is not as accurate.
> 
> Do you guys want me to send a patch or will you change it?

I will wait a bit then rename to keep if nobody else complains.

--
Patrick Mézard