[Dynamite] Philosophy, Telegraph and "object functional"

[Ian Holmes]

On Fri, 30 Jun 2000, Ewan Birney wrote:

> 
> Guys 
> 
> It was great to see telegraph so well defined with this XML. I love it
> already. I think the Object Functional framework is nice; but it does
> not impact the question "can I get this to do smith-waterman".

Not insofar as Smith-Waterman should be easy; but as with any package, it
soon becomes necessary to delve into the object model if you want to get
it to do cool stuff.

> Afterall Dynamite *doesnt* have an object-functional framework and *is*
> able to a fuck of alot of the stuff we want telegraph to do.

Possibly; I would argue that the design of Dynamite *is* somewhat object
functional, but without explicitly mentioning it. After all, it

 - generates functions (code) from templates to do well-defined algorithms
 - does flexible binding of parameters to resources
 - allows arbitrary parameter mappings via "calc" lines
 - has a formal mathematical description of what's going on
 - situates all this within an object model

All of these design features and more have made their way from Dynamite to
Telegraph. I don't think there's a lot that's new in Telegraph in this
respect, except acknowledging the persistent functional theme.

> But it *is nice* that someone has a framework where this all works.

I'm glad you find some aesthetic quality to it; that's what I was shooting
for.

I would certainly hope that a consistent framework is a good thing. In the
distant future I think we will see biologists happily firing off queries
that encompass heterogenous datasets, e.g. sequence + expression + disease
mapping. IMO, requiring the query algorithms to be well-defined from both
an OOP *and* a math viewpoint is the only sane way to get there! Again,
this is what Dynamite already does to a large extent.

> I think I have three roles here:
> 
> 
> 	a) help spot potential practical issues in the design of the
> system
> 
> 	b) be a great example of a "power-user" of telegraph
> 
> 	c) cheerlead and talk it up.
> 

Brilliant on all three. My respect for Dynamite increases the more we try
and emulate it; IMO your authority and experience from having created and
maintained Dynamite is crucial to winning over a user base to Telegraph. I
wouldn't want to do this without you being close to the development
process AND using the tools in anger (& of course "talking it up" ;-) ).

> 
> A couple of questions
> 
> 
> 	a) where is the CVS repository?
> 

See Guy's mail.

> 
> 	b) when do we "go public".
> 

If you mean when do we announce the CVS repository to the world, I'd say
perhaps a month before ISMB, i.e. end of July. If you mean "when's the
IPO", I can't tell you.

Ian

> 
> 
> -----------------------------------------------------------------
> Ewan Birney. Mobile: +44 (0)7970 151230, Work: +44 1223 494420
> <birney@ebi.ac.uk>. 
> -----------------------------------------------------------------
> 
>