Re: deprecate --error-output [was: traceon behavior in head]

[Gary V. Vaughan]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hey Eric!

On 19 Sep 2006, at 14:16, Eric Blake wrote:
> According to Gary V. Vaughan on 9/18/2006 6:23 PM:
> > Nice work!  Do you ever sleep? ;-)
>
> Is that rhetorical? :)

I'll take that as a `no' then :-p

> > One of the big things I wanted to push for with 2.0 was  
> > consistency between
> > the options and builtins, particularly where something was  
> > possible with
> > one
> > but not the other.  In respect of that mantra, --debugfile is the  
> > right
> > option.
> > Since we don't want to break backwards compatibility so  
> > gratuitously, lets
> > continue to support -o and --error-output too, but perhaps give a
> > deprecation
> > warning for the next several releases.
>
> For now, I left -o as is, added --debugfile as a synonym, and added a
> deprecation warning to --error-output.  But for consistency with other
> programs, perhaps a future release of m4 (post 2.1) should treat -o  
> like
> other GNU tools, where it redirects stdout (and is named --output);  
> but
> that would mean we need to start issuing a deprecation warning for -o
> right now in order to phase out existing uses first.

Agreed.  Consistency is worth striving for.  Please go ahead and  
deprecate
`-o' now, perhaps adding a comment in the code that reminds us why.

>   I don't know if
> - --debugfile needs a replacement short option, or if the long  
> option is
> good enough; but -f is still available (-d and -D are both taken,  
> and -o
> doesn't really mesh with --debugfile).  Thoughts?

I think it is used rarely enough that having to type the long option  
is not
a big deal.

> Also, I found that your move from global variables to a context  
> broke -o
> back in 2003.

Oops!  Just goes to show how little used it is :-)  Thanks for the  
regression
test... we can't have too many tests!

> 2006-09-19  Eric Blake  <address@hidden>
>
>         * m4/debug.c (set_debug_file): Fix regression in -o when stdout
>         and stderr are same file, introduced 2003-07-23.
>         * tests/options.at (--debugfile): New test, to catch the bug.
>         (deprecated options): Test --error-output.
>         * src/main.c (usage, main, long_options, ERROR_OUTPUT_OPTION):
>         Deprecate --error-output, and replace it with --debugfile.
>         * doc/m4.texinfo (Invoking m4, Debug Output, Dumpdef, Errprint):
>         Document this change.
>         * NEWS: Likewise.

Cheers,
        Gary.
- --
Gary V. Vaughan      ())_.  gary@ 
{lilith.warpmail.net,gnu.org},address@hidden
Research Scientist   ( '/   http://blog.azazil.net
GNU Hacker           / )=   http://www.gnu.org/software/{libtool,m4}
Technical Author   `(_~)_   http://sources.redhat.com/autobook




-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFFD+8fFRMICSmD1gYRAv+VAKDF41AsTlC4KbVa2cvYiae+sHRHEwCgt1kZ
hoZh+5usvg7UP8YFpb44W3A=
=7AZv
-----END PGP SIGNATURE-----