W dniu 12.09.2016 o 03:57, Junio C Hamano pisze:
> Jacob Keller <jacob.kel...@gmail.com> writes:
>> Yes, I'm actually confused by "git commit <files>" *not* usinng what's
>> in the index already, so I think that isn't intuitive as is.
> You are excused ;-)
> In ancient days, "git commit <pathspec>" was to add the contents
> from working tree files that match <pathspec> to what is already in
> the index and create a commit from that state.
That is, "git commit <pathspec>" meant --include, being equivalent to
"git commit --include <pathspec>".
> This ran against the
> intuition of many users who knew older systems (e.g. cvs) and we had
> to migrate it to the current behaviour by breaking backward
That is, "git commit <pathspec>" means --only, being equivalent to
"git commit --only <pathspec>".
But it was always about working tree version of <pathspec>.
And of course older version control systems didn't have the index...