Re: Change in .gitignore handling: intended or bug?

From: Duy Nguyen <pclouds@gmail.com>
Date: Mar 10, 2016, 00:26 UTC
Message-ID: <CACsJy8D=OjLBN-EXyKM4NnGhSGftYX+RAaR9fg8ML3QWAAb+iQ@mail.gmail.com>
In-Reply-To: <xmqqk2lbil7k.fsf@gitster.mtv.corp.google.com>

On Thu, Mar 10, 2016 at 1:02 AM, Junio C Hamano <gitster@pobox.com> wrote:

Show 34 quoted lines

> Duy Nguyen <pclouds@gmail.com> writes:
>
>> On Wed, Mar 9, 2016 at 1:10 AM, Junio C Hamano <gitster@pobox.com> wrote:
>>> diff --git a/Documentation/gitignore.txt b/Documentation/gitignore.txt
>>> index 3ded6fd..91d1ce2 100644
>>> --- a/Documentation/gitignore.txt
>>> +++ b/Documentation/gitignore.txt
>>> @@ -148,7 +148,43 @@ excluded, the following conditions must be met:
>>>     be in the same .gitignore file.
>>>
>>>   - The directory part in the re-include rules must be literal (i.e. no
>>> -   wildcards)
>>> +   wildcards and has to start with a `/`).
>>
>> Technically '/' can just appear anywhere in the pattern, except at the
>> end. But because the patterns in question must look like this
>>
>>     dir # or any pattern, even "*"
>>     !dir/someth*ng
>>
>> even if there is a slash at the end (and is ignored), we are still
>> good. Not sure how to phrase that though.
>
> Post 2.8, we'd be correcting this properly anyway, we should aim for
> the simplest-to-explain way to work around the limitation in the
> current code, that will still work once the bug is fixed.  I am not
> sure if "Technically it can be other things" helps, unless it makes
> it a lot easier to use.
>
> With that in mind, do you think we need to find a better phrase to
> loosen what I wrote that is overly-strict?  That is, is "must begin
> with '/' to anchor it to the level .gitignore appears" too strict to
> make it too hard to use?
>

If we can name rules from the syntax part, then we can simply say "patterns that meet rules x and y".

A bit off topic, but these two paragraphs may need rephrasing, I don't really understand what it's trying to say

 - If the pattern does not contain a slash '/', Git treats it as
   a shell glob pattern and checks for a match against the
   pathname relative to the location of the `.gitignore` file
   (relative to the toplevel of the work tree if not from a
   `.gitignore` file).

Not sure why "relative to the location of .gitignore file" matters. We basically just take `basename $path` out and try to match it.

 - Otherwise, Git treats the pattern as a shell glob suitable
   for consumption by fnmatch(3) with the FNM_PATHNAME flag:
   wildcards in the pattern will not match a / in the pathname.
   For example, "Documentation/{asterisk}.html" matches
   "Documentation/git.html" but not "Documentation/ppc/ppc.html"
   or "tools/perf/Documentation/perf.html".

Perhaps "Otherwise" can be stated explicitly that "if the pattern does contain any slash, besides the trailing one". The "wildcards not matching /" is also true for the "pattern does not contain a slash" rule above. We can make it a generic note about wildcards and remove this "otherwise" rule, maybe...

-- 
Duy