Re: Can't figure out some sense from the git-commit-tree man page

From: Daniel Hulme <st@istic.org>
Date: Aug 17, 2007, 16:55 UTC
Message-ID: <20070817165519.GA28614@istic.org>
In-Reply-To: <20070817163034.GA11151@glandium.org>

On Fri, Aug 17, 2007 at 06:30:34PM +0200, Mike Hommey wrote:

Show 18 quoted lines

> Hi,
> 
> The git-commit-tree man page reads:
>   A commit encapsulates:
> 
>   ·   all parent object ids
> 
>   ·   author name, email and date
> 
>   ·   committer name and email and the commit time.
>   If not provided, "git-commit-tree" uses your name, hostname and domain to
>   provide author and committer info. This can be overridden by either
>   .git/config file, or using the following environment variables.
>   (...)
> 
> The "If not provided" part doesn't make sense. If you're not setting the
> author and committer in your config or environment, how are you supposed
> to provide it ?

Makes sense to me. I interpreted it as:

  If not provided (by one of the ways I'm about to mention), the default
  is ... This default can be overridden by ...

I can understand how it might be read differently, though. Assuming my reading is correct, might something along these lines be clearer:

  To find author and committer info, "git-commit-tree" first looks at
  these environment variables:

...

  If they are not set, it tries the keys user.name and user.email from
  the .git/config file. Finally, it falls back to using your name,
  hostname, and domain.

Even clearer, though this might be a bit extreme, would be to not mention .git/config and say it reads the global or repository-specific keys using git-config.

-- 
"I like talking to Rabbit.  He talks  about sensible things.  He doesn't
use long,  difficult words,  like Owl.  He uses short,  easy words, like
'What about  lunch?' and  'Help yourself,  Pooh.'  I suppose,  really, I
ought to go and see Rabbit."     A. A. Milne, 'The House at Pooh Corner'