Re: Feature Request: gitignore recursion

From: Jeff King <peff@peff.net>
Date: Oct 7, 2013, 11:24 UTC
Message-ID: <20131007112411.GA5792@sigill.intra.peff.net>
In-Reply-To: <CACsJy8CHcyDrzieX1Ps3LU2UxCP+QJdv9+5BXZj9Mw2bMs5v3w@mail.gmail.com>

On Mon, Oct 07, 2013 at 05:26:17PM +0700, Nguyen Thai Ngoc Duy wrote:

Show 11 quoted lines

> On Mon, Oct 7, 2013 at 5:23 AM, AJ <alljeep@gmail.com> wrote:
> > I'm hoping to get the following feature implemented into git.
> >
> > Add the ability to recursively include using:
> > !/my_dir/**/*
> 
> You can do that since v1.8.2. Actually the pattern should be
> 
> !/my_dir/**
> 
> Checkout gitignore man page for more information.

I think that is a reasonable approach to doing what the OP wants.

However, one thing I expected to work but didn't is:

  echo '*' >.gitignore
  echo '!*' >my_dir/.gitignore

That _does_ work for attributes, like:

  echo '* foo=one' >.gitattributes
  echo '* foo=two' >my_dir/.gitattributes

where the more-specific file takes precedence. It works because we keep an attribute stack, and look from most-specific directory to least.

The directory traversal code, however, tries to avoid entering directories that are ignored. So if you have told it to ignore "my_dir", we will not even look at "my_dir/.gitignore".

This is arguably a bug, as the documentation says that more specific paths should take precedence over toplevel ones. But the optimization of pruning ignored paths is very important for performance; otherwise, we would end up enumerating entire ignored subdirectories on the off chance that they have a negative .gitignore entry buried deep inside them.

So I don't think it is worth fixing, but it might be worth mentioning in the documentation. Or is it mentioned somewhere that I missed?

-Peff