Re: [PATCH] rebase -i: fix has_action

Sverre Rabbelier [off-list ref] writes:

Heya,

On Thu, Aug 4, 2011 at 21:34, Junio C Hamano [off-list ref] wrote:

quoted

       has_action () {
         test -s "$1"
       }

quoted

       has_action () {
         sane_grep -v -e '^#' -e '^[   ]*$' "$1" >/dev/null
       }

I think the former more correctly checks what the function name
implies, is there any downside to that which makes you suggest this
second approach?

I vaguely recall the original reason we didn't do the most straightforward
thing was something like what J6t said already.

As we are not interested in _adding_ new feature, I would say that,
strictly speaking, this *should* become a two-patch series whose first one
uses

	sane_grep -v -e '^#' -e '^$' "$1" >/dev/null

that is, "do we have anything aside from comments and blanks?", which is
the original semantics, with Noe's "safety" change as the second patch in
the series that uses

	sane_grep -v -e '^#' -e '^[	 ]*$' "$1" >/dev/null

to say "let's count a line that solely consists of whitespaces also as a
blank".

But of course in practice it can and should be just a single patch that 
squashes these two "conceptually separate" steps.