Re: standard libraries [was: 1.8.2 - what can we do to help?]

On Thu, 28 Oct 2004, Dave Thomas wrote:

>
> On Oct 27, 2004, at 10:52, James Britt wrote:

>> So I wonder if there is a similar need for a Ruby Library Project, where a 
>> deliberate, public effort is made to ensure that a) the std-lib code is 
>> complete and maintained, with an assurance that the code is just as 
>> supported as the base language, and b) (maybe) push to improve or acquire 
>> 3rd-party libs such as are common in Perl & Python?

I think this is laudable,

>
> I don't think so. The problem with the RDP, and I anticipate with any RLP, is 
> that the folks doing the work are not the folks writing the code. This often 
> leads to disconnects between what is done and the intention of the library 
> writer. It also means that we're always playing catch-up.

but agree that these problems exist.
>
> I would much rather see an atmosphere created where everyone understands that 
> libraries must meet certain standards: versioning, testing, documentation,

See point 1 below...

> and so on, and the have the library writer put together the necessary

See point 2 below...

> resources to make their libraries compliant. In future, no library should be 
> added to the core until it meets these standards.

This would be the ideal, but *I* don't see how we can achieve it,
because:
>
> One thing I like about gems is that it goes part way towards this. However, 
> gems is for external libraries. I'd like to think that our internal ones 
> could be as good.

(1) Gems, and Test:Unit give us the versioning and the testing
scaffolding, but the problem of documentation raises its head again,

The rise and rise of Rdoc shows that "the ruby code is the
documentation" and "Use the source, Luke" are somewhat out of step with
reality.  It takes a while to be "strong in the source", and we have
been discussing how beginners are put off.   Yes, there are people
who only need code posted to the lists, and barely need English,
French or Japanese :-), but many of us will never reach such dizzy
heights, alas.

(2) What about the "and so on"?  What other resources do we need to
create to help writers make their libraries compliant with the
ideals one would want?  The only thing I can think of is a
refactoring tool, and I think there was some work towards that with
Ride.  I've not kept up with that, though.
>
> Support for this would need to come from the top, which really means from 
> Matz and a few others in Japan. If they got behind this, I think we'd see a 
> rapid and dramatic improvement.
>
and we run into the language barrier again.  How I wish I could just
self.extend(Japanese); in seconds!  Translation is expensive,
so should we try to find a way to raise money to facilitate
this communication?

Does anyone know of good ways to meet the conflicting aims of:

(a) keep the documentation with the code
(b) provide it in many languages
(c) don't put too many constraints on the type of editor you need to
     edit the code

because I don't, at least, not before Unicode is ubiquitous.

I'm not trying to pour cold water on this: any negativity is due to
my lack of foresight in this area.  I'm trying to focus on aspects
of the problem so that it can be tackled piecewise.

Oh, and I am very aware that this is turning into aspects of 
[Ruby-talk:66114]. Ouch.  Especially as I was voice 6 at one point!

>
> Cheers
>
> Dave
>
         Hugh