Re: Consensus on "Git"

From: Jeff King <peff@peff.net>
Date: Nov 11, 2009, 09:36 UTC
Message-ID: <20091111093656.GB17422@sigill.intra.peff.net>
In-Reply-To: <20091111092731.GA17422@sigill.intra.peff.net>

On Wed, Nov 11, 2009 at 04:27:31AM -0500, Jeff King wrote:

Show 14 quoted lines

> For a fun (ab)use of blame, you can see who wrote each of David's
> suspect lines with:
> 
>    git grep -n "\bGIT\b" |
>    while IFS=: read file line junk; do
>      git blame -f -L $line,$line $file
>    done |
>    less
> 
> Many of them are variables (e.g., \b gets rid of "_", but we still have
> "-", so GIT-VERSION and such are still there). Many others are the title
> of the "GIT" section of the manpages. But there are some legitimate
> uses, too. Many of them blame to quite a long time ago, though (e.g.,
> Documentation/everyday.txt has two uses by you in 2005).

BTW, here is a much better grep (still a few false positives, but most lines are meanginful):

   git grep -n -E "(^|[^A-Z_-])GIT($|[^A-Z_-])" |
   grep -v ':GIT$' |
   while IFS=: read file line junk; do
     git blame -f -L $line,$line $file
   done |
   less

You can really see that most of the uses of GIT are ancient. I wonder if we should downcase the few that are in user-visible error messages. I will let somebody else spend their time on such a conversion if they want, though. I've already spent enough on this. :)

-Peff