[PATCH] merge: output the complete options in the guidance text, not only the often used. Change it as well in the test case outputs.

[Mads Kiilerich]

On 06/27/2010 07:05 PM, Matt Mackall wrote:
> On Sat, 2010-06-26 at 13:33 +0200, rupert THURNER wrote:
>> # HG changeset patch
>> # User Rupert THURNER<rupert.thurner at gmail.com>
>> # Date 1277550559 -7200
>> # Node ID 74e62efe1bc867702df88e0f5c8b9bf1c18ca207
>> # Parent  8b452fe4bf506a1a08bbc7e1bac81aceda8f4d10
>> merge: output the complete options in the guidance text, not only the
>> often used. Change it as well in the test case outputs.
>
> For future reference, updating the translation files isn't your job -
> translators will do that as needed.
>
> I'm still not sold on this: up -C is dangerous and handing it to
> clueless users without caveats isn't a great idea. How about something
> like:
>
> diff -r b9a46acdfe1f mercurial/commands.py
> --- a/mercurial/commands.py	Sat Jun 26 12:26:03 2010 -0500
> +++ b/mercurial/commands.py	Sun Jun 27 12:04:41 2010 -0500
> @@ -2570,6 +2570,10 @@
>       head, the other head is merged with by default. Otherwise, an
>       explicit revision with which to merge with must be provided.
>
> +    To undo an uncommitted merge, use :hg:`update --clean .` which
> +    will check out a clean copy of the first parent, losing all
> +    changes.
> +
>       Returns 0 on success, 1 if there are unresolved files.
>       """

Perhaps we should add something with "and forget the second/merged parent".

The story of working directory parents vs content is technical and 
apparently hard to get right, but it is important in order to understand 
the difference between revert, debugsetparents and update --clean.

The trailing "." is easy to miss. Perhaps it should be mentioned 
explicitly, now when we are in the verbose friendly mode.

/Mads