Re: [PATCH 1/2] drop length limitations on gecos-derived names and emails

[Jeff King <peff@peff.net>]

On Mon, May 14, 2012 at 05:02:25PM -0400, Jeff King wrote:

> Ick, there is something very wrong with this patch. While testing a
> completely unrelated bug, I noticed that it set my name to "Jeff
> KingJeff KingJeff King". Which, while a wonderful ego massage, is
> probably excessive.
> 
> I'm sure the problem is the switch to strbuf's appending semantics
> rather than strlcpy's overwriting semantics. I thought we were careful
> not to bother re-run the gecos code if we had already gotten a name, but
> obviously that is not the case in some code paths. I'll investigate.

Ah, I see. The problem is here:

> --- a/config.c
> +++ b/config.c
> @@ -767,7 +767,7 @@ static int git_default_user_config(const char *var, const char *value)
>  	if (!strcmp(var, "user.name")) {
>  		if (!value)
>  			return config_error_nonbool(var);
> -		strlcpy(git_default_name, value, sizeof(git_default_name));
> +		strbuf_addstr(&git_default_name, value);
>  		user_ident_explicitly_given |= IDENT_NAME_GIVEN;
>  		return 0;
>  	}
> @@ -775,7 +775,7 @@ static int git_default_user_config(const char *var, const char *value)
>  	if (!strcmp(var, "user.email")) {
>  		if (!value)
>  			return config_error_nonbool(var);
> -		strlcpy(git_default_email, value, sizeof(git_default_email));
> +		strbuf_addstr(&git_default_email, value);
>  		user_ident_explicitly_given |= IDENT_MAIL_GIVEN;

where we are not careful. The fix is trivial. However, while examining
fmt_ident, I notice there is another potential spot there that needs
further investigation (I think it may actually be unreachable code, but
I need to look closer).

I'll re-roll the series with the fixes after investigating fmt_ident.

-Peff