Re: [PATCH 4/5] Keep '*' in pattern refspecs

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

Daniel Barkalow <barkalow@iabervon.org> writes:

> In order to do anything more capable with refspecs, the first step is
> to keep the entire input. Additionally, validate patterns by checking
> for the ref matching the rules for a pattern as given by
> check_ref_format(). This requires a slight change to make it require
> the '*' to be at the beginning of a path component.

I had a brief "Huh?" moment wondering about this "slight change", but at
this stage it does not change the rule at all ("/*" still must happen at
the end of the string), so there actually is no change.

> diff --git a/remote.c b/remote.c
> index 93fd03d..d0ce4c6 100644
> --- a/remote.c
> +++ b/remote.c
> @@ -722,10 +716,10 @@ int remote_has_url(struct remote *remote, const char *url)
>  static int name_fits_pattern(const char *key, const char *name,
>  			     const char *value, char **result)
>  {
> -	size_t klen = strlen(key);
> -	int ret = !strncmp(key, name, klen);
> +	size_t klen = strchr(key, '*') - key;
> +	int ret = !strncmp(name, key, klen);

Any particular reason why the first parameters to strncmp() were swapped?

>  	if (ret && value) {
> -		size_t vlen = strlen(value);
> +		size_t vlen = strchr(value, '*') - value;

We would want protection from programming error here, to catch keys and
values without any asterisk.  This comment also applies to [5/5].

>  		*result = xmalloc(vlen +
>  				  strlen(name) -
>  				  klen + 1);