Re: [RFC] Instruct git-completion.bash that we are in test mode

[Jeff King <peff@peff.net>]

On Mon, Jan 21, 2013 at 03:32:29PM -0800, Junio C Hamano wrote:

> "Jean-Noël AVILA" <jn.avila@free.fr> writes:
> 
> > At least, "it works for me".
> 
> I suspect that your approach will still not fix the case in which
> you build a branch with a new command git-check-ignore, and then
> check out another branch that does not yet have that command without
> first running "make clean".
> 
> Does the following really pass with your patch?
> 
> 	git checkout origin/next
>         make
>         git checkout origin/maint
> 	git apply your_patch.mbox
>         make
>         cd t && sh ./t9902-completion.sh

I really hate to suggest this, but should it be more like:

  if test -z "$FAKE_COMMAND_LIST"; then
          __git_cmdlist() {
                  git help -a | egrep '^  [a-zA-Z0-9]'
          }
  else
          __git_cmdlist() {
                  printf '%s' "$FAKE_COMMAND_LIST"
          }
  fi

That gives us a nice predictable starting point for actually testing the
completion code. The downside is that it  doesn't let us test that we
remain compatible with the output of "help -a". But we could potentially
add a single, more liberal test (without $FAKE_COMMAND_LIST, but ready
to expect extra output) that checks that.

> > +	__git_cmdlist () { git help -a| egrep -m 1 -B1000 PATH | egrep '^  [a-zA-Z0-9]'; }
> 
> 'egrep' is not even in POSIX in the first place but grep -E ought to
> be a replacement for it, so I'll let it pass, but "-m1 -B1000"?
> Please stay within portable options.

If I recall correctly, egrep is actually more portable than "grep -E"
(and it is already in use, so I think we are OK). I agree on the rest,
though. :)

-Peff