Re: [PATCH] fast-export: factor out print_oid()

From: Taylor Blau <me@ttaylorr.com>
Date: Aug 13, 2020, 15:18 UTC
Message-ID: <20200813151856.GE2244@syl.lan>
In-Reply-To: <1eb633f5-31ec-5d8d-69ac-35d0fb9772da@web.de>

On Thu, Aug 13, 2020 at 01:11:18PM +0200, René Scharfe wrote:

Show 25 quoted lines

> Simplify the output code by splitting it up and reducing duplication.
> Reuse the logic for printing object IDs -- anonymized if needed -- by
> moving it to its own function, print_oid().
>
> Signed-off-by: René Scharfe <l.s.r@web.de>
> ---
>  builtin/fast-export.c | 26 +++++++++++++++-----------
>  1 file changed, 15 insertions(+), 11 deletions(-)
>
> diff --git a/builtin/fast-export.c b/builtin/fast-export.c
> index 9f37895d4cf..49bb50634ab 100644
> --- a/builtin/fast-export.c
> +++ b/builtin/fast-export.c
> @@ -420,6 +420,14 @@ static const char *anonymize_oid(const char *oid_hex)
>  	return anonymize_str(&objs, generate_fake_oid, oid_hex, len, NULL);
>  }
>
> +static void print_oid(const struct object_id *oid, int anonymize)
> +{
> +	const char *oid_hex = oid_to_hex(oid);
> +	if (anonymize)
> +		oid_hex = anonymize_oid(oid_hex);
> +	fputs(oid_hex, stdout);
> +}
> +

The fact that this calls fputs makes this patch (in my own opinion) noisier than it needs to be. This is because of all of the factoring out of the other printfs. I'd expect that this looks something more like:

  -				       anonymize ?
  -				       anonymize_oid(oid_to_hex(&spec->oid)) :
  -				       oid_to_hex(&spec->oid));
  +				       anonymize_oid(anonymize, &spec->oid));

without moving around all of the other printf code.

Show 26 quoted lines

>  static void show_filemodify(struct diff_queue_struct *q,
>  			    struct diff_options *options, void *data)
>  {
> @@ -470,21 +478,19 @@ static void show_filemodify(struct diff_queue_struct *q,
>  		case DIFF_STATUS_TYPE_CHANGED:
>  		case DIFF_STATUS_MODIFIED:
>  		case DIFF_STATUS_ADDED:
> +			printf("M %06o ", spec->mode);
>  			/*
>  			 * Links refer to objects in another repositories;
>  			 * output the SHA-1 verbatim.
>  			 */
>  			if (no_data || S_ISGITLINK(spec->mode))
> -				printf("M %06o %s ", spec->mode,
> -				       anonymize ?
> -				       anonymize_oid(oid_to_hex(&spec->oid)) :
> -				       oid_to_hex(&spec->oid));
> +				print_oid(&spec->oid, anonymize);
>  			else {
>  				struct object *object = lookup_object(the_repository,
>  								      &spec->oid);
> -				printf("M %06o :%d ", spec->mode,
> -				       get_object_mark(object));
> +				printf(":%d", get_object_mark(object));
>  			}
> +			putchar(' ');

... That said, this transformation looks correct from a quick glance.

Show 16 quoted lines

>  			print_path(spec->path);
>  			string_list_insert(changed, spec->path);
>  			putchar('\n');
> @@ -724,12 +730,10 @@ static void handle_commit(struct commit *commit, struct rev_info *rev,
>  		else
>  			printf("merge ");
>  		if (mark)
> -			printf(":%d\n", mark);
> +			printf(":%d", mark);
>  		else
> -			printf("%s\n",
> -			       anonymize ?
> -			       anonymize_oid(oid_to_hex(&obj->oid)) :
> -			       oid_to_hex(&obj->oid));
> +			print_oid(&obj->oid, anonymize);
> +		putchar('\n');

As does this one.

Show 5 quoted lines

>  		i++;
>  	}
>
> --
> 2.28.0

So, I guess what I'm trying to say is that this patch doesn't look wrong, other than that it seems more invasive than I would have expected it to be.

Thanks, Taylor