[BUG] --state used when it shouldn't be

Ovid publiustemp-tapx at yahoo.com
Fri Jan 4 14:29:29 UTC 2008


--- Andy Armstrong <andy at hexten.net> wrote:

> OK, how about a *little* more bloat...
> 
> prove <absolutely whatever> --only t/these-tests
> 
> Whatever tests are scheduled by the other options are filtered by -- 
> only. So while your .proverc or an alias might mean that

I still feel uncomfortable here because this means that the interface
to this very widely used program has changed in a backwards
incompaticle way (assuming the .proverc option is used).  As humans, we
can respond more robustly than, say, another software package, but this
means that we always have to remember the "extra step" we have to take
to avoid strange results.

The best way, in my humble opinion, to approach any API decision is
usually to come up with the absolutely best API possible and then
figure out to make that work.  I don't think adding the '--only' switch
is the best approach, so I expect that people will just not use the
.proverc.  If they *do* use .proverc, they'll forget they're using it
and we'll have a silent "action at a distance" confusing people -- it
certainly confused the heck out of me! :)

Of course, this is just what I and my colleagues are experiencing.  We
could be outliers here.

Cheers,
Ovid

--
Buy the book  - http://www.oreilly.com/catalog/perlhks/
Perl and CGI  - http://users.easystreet.com/ovid/cgi_course/
Personal blog - http://publius-ovidius.livejournal.com/
Tech blog     - http://use.perl.org/~Ovid/journal/


More information about the tapx-dev mailing list