Re: [Swarm-Support] development priorities (was Re: Membership in Swarm Developmen Group)

[Marcus G. Daniels]

glen e. p. ropella wrote:
> Marcus G. Daniels wrote:
>   
> > The PS3 is an interesting value.  You can just use the PS3 as a 9
> > processor Linux machine.  Check it out:
> >
> > http://www.terrasoftsolutions.com
> >     
>
> So what?  There are many gadgets that have interesting value.  But, that
> doesn't mean the SDG should spend the little energy it has on porting to
> them.
>   
One problem people have complained about is runtime.   It's certainly a 
problem I've had and one I care about.
> You're correct, however, that they aren't motivated by the return on
> their investment because the investment is very high.  And we don't know
> what their return might be because they can't invest enough to use it!
>   
I can go round and round with you one what really amounts to an foolish 
act of faith on your part.   A best effort was made by the Swarm project 
and dedicated users to make it usable and it's not something I ever saw 
a `pull' for. 
> My point was that there are some requirements for Swarm that it fails to
> meet.  And adding new features to the current software is a waste of
> money.  Hence, a better use of the SDG's resources is to either build a
> good reference implementation (not the current albatross) that meets the
> requirements set forth for Swarm _or_ abandon the idea of supporting a
> reference implementation altogether.
>   
Everything takes work.  Moving a legacy system forward takes work but it 
can and has been done with Swarm.  It could be done again.    I don't 
think Swarm is particularly unique in this respect compared to other 
packages.   The urge to forget and move on to the another ungrounded 
fantasy world on can be a strong one.

Why you are so fixated on original requirements of Swarm is unclear.   
All software projects evolve.  There's no requirement that requirements 
stay fixed.  I'd say that's a very bad sign, actually.
As far Swarm being a reference implementation goes, I think it achieves 
that.   It implements composable activities, concurrency controls, it 
has test cases for those features, and we made a longstanding effort to 
ensure it worked.  Not all the ideas got implemented, but there were 
multiple priorities in play, e.g. the Java support which you promised we 
would implement.   Even very arguable features like phases were endured  
and made to work in a language neutral way.

No, a foolish use of money would be to write yet another Swarm-like 
thing, especially in the name of `original requirements'.

Marcus