Re: [PATCH/RFC] WIP: log: allow "-" as a short-hand for "previous branch"

[Junio C Hamano <gitster@pobox.com>]

Siddharth Kannan <kannan.siddharth12@gmail.com> writes:

> @@ -158,6 +158,51 @@ static void cmd_log_init_finish(int argc, const char **argv, const char *prefix,
>
>  	if (quiet)
>  		rev->diffopt.output_format |= DIFF_FORMAT_NO_OUTPUT;
> +
> +	/*
> +	 * Check if any argument has a "-" in it, which has been referred to as a
> +	 * shorthand for @{-1}.  Handles methods that might be used to list commits
> +	 * as mentioned in git rev-list --help
> +	 */
> +
> +	for(i = 0; i < argc; ++i) {
> +		if (!strcmp(argv[i], "-")) {
> +			argv[i] = "@{-1}";
> +		} else if (!strcmp(argv[i], "^-")) {
> +			argv[i] = "^@{-1}";
> +		} else if (strlen(argv[i]) >= 4) {
> +
> +	...
> +		}
> +	}
> +
>  	argc = setup_revisions(argc, argv, rev, opt);

"Turn '-' to '@{-1}' before we do the real parsing" can never be a
reasonable strategy to implement the desired "'-' means the tip of
the previous branch" correctly.  To understand why, you only need to
imagine what happens to this command:

    $ git log --grep '^-'

Turning it into "git log --grep '^@{-1}'" obviously is not what the
end-users want, so that is an immediate bug in the version of Git
with this patch applied.

Even if this were not a patch for the "log" command but for some
other command, a change with the above approach is very much
unwelcome, even if that other command does not currently have any
option that takes arbitrary string the user may want to specify
(like "find commit with a line that matches this pattern" option
called "--grep" the "log" command has).  That is because it will
make it impossible to enhance the command by adding such an option
in the future.  So it is also adding the problems to future
developers (and users) of Git.

A correct solution needs to know if the argument is at the position
where a revision (or revision range) is expected and then give the
tip of the previous branch when it sees "-" (and other combinations
this patch tries to cover).  In other words, the parser always knows
what it is parsing, and if and only if it is parsing a rev, react to
"-" and think "ah, the user wants me to use the tip of the previous
branch".

But the code that knows that it expects to see a revision already
exists, and it is the real parser.  In the above snippet,
setup_revisions() is the one that does the real parsing of argv[].
The code there knows when it wants to see a rev, and takes argv[i]
and turns into an object to call add_pending_object().  That codepath
may not yet know that "-" means the tip of the previous branch, and
that is where the change needs to go.

Such a properly-done update does not need to textually replace "-"
with "@{-1}" in argv[]; the codepath is where it understands what
any textual representation of a rev the user gave it means, and it
understands "@{-1}" there.  It would be the matter of updating it to
also understand what "-" means.

A correct solution will be a lot more involved, of course, and I
think it will be larger than a reasonable microproject for people
new to the codebase.

I didn't check the microproject ideas page myself; whether it says
that turning "-" unconditionally to "@{-1}" is a good idea, or it
hints that supporting "-" as "the tip of the previous branch" in
more commands is a reasonable byte-sized microproject, I think it is
misleading and misguided.  Can somebody remove that entry so that we
won't waste time of new developers (which would lead to discouraging
them)?  Thanks.