[pacman-dev] [PATCH v2] pacman: print error when -Fx is given invalid regex

[Allan McRae]

On 28/11/19 3:45 am, morganamilo wrote:
>     When processing the targets for -Fx, compile all the regex ahead of
>     time, printing an error for each that failed to compile. Then, if they all
>     compiled successfully, continue with printing files.
> 
>     Signed-off-by: morganamilo <morganamilo at archlinux.org>
> 
>     ---
> 
>     v2:
>             Add comment about why we don't free targ
>             Fix whitespace error
> 
>     I have vim set to display trialing whitespace so normally I catch it.
>     Not sure why I didn't this time.
> 
>     Also I agree that running all the regex against each target before
>     moving on to the next is a good improvement. This patch is already kinda
>     long though, so I'll do it another time.
> 
> diff --git a/src/pacman/files.c b/src/pacman/files.c
> index 3b6dc23b..6fcc6e9b 100644
> --- a/src/pacman/files.c
> +++ b/src/pacman/files.c
> @@ -94,17 +94,27 @@ static void print_match(alpm_list_t *match, alpm_db_t *repo, alpm_pkg_t *pkg, in
>  	}
>  }
>  
> +struct filetarget {
> +	char *targ;
> +	int exact_file;
> +	regex_t reg;
> +};
> +
> +static void filetarget_free(struct filetarget *ftarg) {
> +	regfree(&ftarg->reg);
> +	/* we don't own ftarg->targ so don't free it */

Patch looks good.

I changed that comment to:

do not free ftarg->targ as it is owned by the caller of files_search

This does a few useful things:
1) gets the "do not free" right at the start of the comment
2) specifies where the memory is owned
3) does not use contractions - apparently the use of "not" make the
negative clearer for non-native speakers, although I am not sure what
sort of evidence is behind that is claim...

Allan