Re: th/remote-usage

[Tim Henigan <tim.henigan@gmail.com>]

Hi and thanks for your review.

On Wed, Nov 18, 2009 at 6:48 AM, Jonathan Nieder <jrnieder@gmail.com> wrote:
> Junio C Hamano wrote:
>> [New Topics]
>>
>> * th/remote-usage (2009-11-16) 1 commit.
>>  - git remote: Separate usage strings for subcommands
>
> Glancing at pu^2, I had two small nitpicks: [<options>...] is five
> characters longer than strictly necessary

I based my patch on what I found in other builtin functions (such
as push and diff).  That being said, I don't think that either my
original patch or your updated version is completely correct.
The choices seem to be:
  (1) [<options>...]:  My original based on my interpretation of
      IEEE 1003.1. [1]
  (2) [options]: Your proposal, which drops both the '<>' and '...'.
  (3) <options>:  Used in builtin-diff.c.  Which does not show
      that the options are -- optional.
  (4) [<options>]: What I now believe is correct (based on the
      current implementation of builtin-push.c).  This drops the
      '...' which IEEE 1003.1 defines as allowing multiple options
      to be specified, but it conforms to the conventions in other
      commands.

There does not (yet) seem to be consistency in how options
are presented.  My current plan is to change the patch to
use choice #4, but if Junio has a chance to comment, I will
of course defer to his decision.

I will send an updated patch that implements choice #4 as
soon as I can (should be within the next 12 hours).


> and the argument to git
> remote set-head is not actually optional.

This was obviously an oversight on my part.  I will include the
fix in the next version.


...and from your second email:
> Another option would be to make the strings into static
> variables.

Thanks for the analysis, but I don't plan to include this change
unless specifically requested.


[1]: http://www.opengroup.org/onlinepubs/009695399/basedefs/xbd_chap12.html