On Fri, 08 Apr 2011 11:23:44 +0200, Stuart Henderson wrote:
> would anyone else find this (or a variation thereof) useful?
>
> example use: if you download a partial snapshot and would like to
> check files, this makes it easy to identify whether a file actually
> fails the checksum, or whether you just don't have it.
>
> Index: md5.c
> ===================================================================
> RCS file: /cvs/src/bin/md5/md5.c,v
> retrieving revision 1.52
> diff -u -p -r1.52 md5.c
> --- md5.c 27 Oct 2010 15:24:10 -0000 1.52
> +++ md5.c 8 Apr 2011 09:17:29 -0000
> @@ -655,7 +655,7 @@ digest_filelist(const char *file, struct
>
> if ((fp = fopen(filename, "r")) == NULL) {
> warn("cannot open %s", filename);
> - (void)printf("(%s) %s: FAILED\n", algorithm, filename);
> + (void)printf("(%s) %s: MISSING\n", algorithm, filename);
> error = 1;
> continue;
> }
Two messages are printed for the very same error. Would it make sense to
remove the printf() message altogether or, alternatively, integrate it
into the warn() message (e.g.
warn("(%s) %s: FAILED", algorithm, filename))?
