Re: Minor bug: git config ignores empty sections

On Mon, Aug 15, 2016 at 11:28:20AM -0700, Junio C Hamano wrote:

Jeff King [off-list ref] writes:

quoted

On Mon, Aug 15, 2016 at 07:34:50PM +0200, Andreas Schwab wrote:

quoted

On Aug 15 2016, Jeff King [off-list ref] wrote:

quoted

And implicit in your test is the other bug, which is that deleting the
last key in a section leaves the empty header. I think it's related to
the same issue.

Indiscriminately removing empty section headers may break comments that
have been put there on purpose.

I know, but we do not even do so discriminately.

I notice that we have thought about all the issues when we last
discussed it in 2013.  Refining a message from the earlier thread,
as it illustrates tricky cases in which we have to be careful.

Thanks for digging up the threads that I was too lazy to find.

I agree with most everything here, though I would be happy if somebody
even wrote a patch to handle the "easy" cases.

So a comment outside [section "name"] is tricky; it needs some
mechanism (or convention) to tell us if it is about the particular
section, or it is about the location in the configuration file.

Keep in mind that even "outside" is hard, because sections do not
explicitly close.

So in:

  [core]
  foo = bar

  # here are my remotes

  [remote "github"]
  url = ...

How do we know that the comment is "outside" and not part of [core]?

We can perhaps guess so because there are no keys after it in the
section, though there are some special cases, like:

  [core]
  foo = bar
  # This isn't necessary anymore because...
  # xyzzy = false

or even:

  [core]
  foo = bar # needed because of xyzzy

You can probably make reasonable cases based on heuristics around
newlines, but that is even further into "convention" territory.

-Peff