Re: [PATCH] RFC: help.autocorrect prompt

[Jonathan Nieder <jrnieder@gmail.com>]

(+cc: Dscho and Heiko)

David Barr wrote:

> A value of 'prompt' causes git to wait for confirmation
> before executing the assumed command.

Nice idea.

> ---

Sign-off?

> --- a/help.c
> +++ b/help.c
> @@ -267,8 +267,12 @@ static struct cmdnames aliases;
>  
>  static int git_unknown_cmd_config(const char *var, const char *value, void *cb)
>  {
> -	if (!strcmp(var, "help.autocorrect"))
> -		autocorrect = git_config_int(var,value);
> +	if (!strcmp(var, "help.autocorrect")) {
> +		if (!strcmp(value, "prompt"))
> +			autocorrect = INT_MAX;
> +		else
> +			autocorrect = git_config_int(var,value);

Any particular significance to INT_MAX rather than, e.g., -1 here?
(Just curious; it seems unlikely someone would use INT_MAX and
accidentally trip on this.)

Not a problem introduced by your patch: should we be checking for
out-of-range (e.g., negative) values?

> @@ -342,7 +346,10 @@ const char *help_unknown_cmd(const char *cmd)
>  			"which does not exist.\n"
>  			"Continuing under the assumption that you meant '%s'\n",
>  			cmd, assumed);
> -		if (autocorrect > 0) {
> +		if (autocorrect == INT_MAX) {
> +			if (strcmp("y", git_getpass("Continue? (y/n) ")))
> +				exit(1);

Funny. :)

It might be better to actually always write this prompt to the
terminal, rather than popping up a gui $GIT_ASKPASS if the user has
set that up.  Maybe something like Heiko Voigt's "mingw: make failures
to unlike or move raise a question" (9229029, 2010-02-21 from
4msysgit.git):

	if (!isatty(STDIN_FILENO) || !isatty(STDERR_FILENO))
		exit(1);
	fprintf(stderr, "Continue? (y/n) ");
	if (!fgets(answer, sizeof(answer), stdin))
		exit(1)
	if (*answer != 'y' && *answer != 'Y')
		exit(1);

> +		} else if (autocorrect > 0) {
>  			fprintf(stderr, "in %0.1f seconds automatically...\n",
>  				(float)autocorrect/10.0);
>  			poll(NULL, 0, autocorrect * 100);

Thanks for a pleasant read.
Jonathan