Re: All's fair (in) love and the sdk list

On Tue, May 16, 2000 at 08:33:01AM -0600, Nathan Torkington wrote:
> Jonathan Scott Duff writes:
> > However, Graham's idea of a web/newsgroup survey can't hurt.
> > TMTOWTDI.  (Perhaps release 0.75 will be the survey)
> 
> Anything other than a release will delay it yet again.  Just release
> 0.75 and get feedback that way.  Post the list of modules, build a
> list of suggested inclusions and exclusions, weigh the pros and cons
> of each, make the final decision and go.

To that I would disagree.

> The more we wring our hands and worry that we don't have the best
> list, the smaller the chance becomes that it will ever happen.  One
> thing I learned while doing TPI work is that a perfect system which is
> never created is useless.  I think trying to find "the right list" is
> the wrong approach.  Find a set of modules that we'd be pleased to
> install en masse, call that 0.75.  If people don't like it when we
> remove modules before 1, that's no problem.

And what would be your criteria for removal ? I doubt very much that
users will ask for any given module to be removed. They are more likely
to be asking for additions.

But then it would seem that if someone asks for a module where the SDK
includes a similar, but not the same, module the answer to them will be
"no, the SDK already has xyz, use that"

This is WRONG, you are going to build the perception of a two class
CPAN. Which just happens to be one thing Andreas is trying to remove
by moving cpan ids around.

> 1.00 is what will be shipped by vendors,

And so will any other version. Don't under estimate vendors. Remember
we have had so called production Linux distribution which had "development"
releases of perl in.

> announced, and used by normal
> people rather than testers. 

Again, if it is on CPAN, people will install it.

> I can live with pre-1.00 breakage and
> whinage.  Getting it out the door should be priority one.

Just because it has taken two years to acyually get something going on this
subject does not mean it should be rushed out.

Graham.