Re: sort --version-sort

[Bruce Korb]

Hi Jim,

On Fri, Aug 15, 2008 at 1:30 AM, Jim Meyering <address@hidden> wrote:

> Thanks for the follow-up!
> Note that in coreutils I'm not using a ChangeLog file
> per se anymore.  Instead, it's generated from commit logs,
> as mentioned in HACKING.  Speaking of ChangeLogs, I prefer
> more (some would say "excruciating" ;-) detail, so rewrote that one:

Yep.  "excruciating" is a good description.  :)  For me, change
logs are for giving fuzzy notions about what sort of change was
made when.  If you want excruciating detail, that is what SCM
historical diffs are for.  But that is for my projects and this
is yours.

> A minor qualm:
>  - This change adds a short option: -V
>    Does any other vendor's sort program accept -V?

I don't know.  Both "-v" and "-V" stand for version, so it
seemed like a reasonable choice.  All of the sort mode options
have an associated flag character, so picking one seemed
reasonable, too.

> I like your cpp-based table-manipulation idiom that lets
> us eliminate the use of ARGMATCH_VERIFY.  Thanks!

I hate having to coordinate parallel tables.  You're welcome.

> Also, I've made the following adjustments that will
> be combined into your commit before I push.
> Since the commit has your name on it, I'll wait to hear
> back from you before doing this:

You needn't have wated, for future reference.  But also for
future reference:

> @@ -1822,19 +1822,19 @@ compare_random (char *restrict texta, size_t lena,

>  static int
>  compare_version (char *restrict texta, size_t lena,
> -                char *restrict textb, size_t lenb)
> +                char *restrict textb, size_t lenb)

That indentation is emacs magic.  I press "tab" and never look at
it again.  If there is a c-mode variable setting, ....

Cheers  - Bruce