[Biojava-l] BioInformatics toolbox.

Michael L. Heuer heuermh@acm.org
Mon, 8 Apr 2002 23:38:03 -0400 (EDT)


On Tue, 9 Apr 2002, Mario Gianota wrote:

> Correction, there is a web page with lofty ideas that hasn't been updated
> since August of last year. In other words, there is nothing for biologists
> in this area and I'm honestly not surprised: have you ever tried to strip
> NetBeans down to its bare minimum ?

Yes, actually I have on more than one occassion.

I was able to successfully pull out from netbeans some but not all of the
components I was interested in -- the property sheet and the Explorer API.
Unfortunately, the core implementation of the netbeans openapis is a very
messy bit of code, and it's nearly impossible to instantiate a part of
netbeans without pulling in the whole mess.

The netbeans team is making progress in this area, see

> http://netbeans.org/platform
> http://core.netbeans.org/platform/index.html

There is an issuezilla item to track progress

> http://www.netbeans.org/issues/show_bug.cgi?id=17815

but there are a couple of key bugs/issues standing in the way, most
significantly this one

> http://www.netbeans.org/issues/show_bug.cgi?id=18273


> Whilst IDEs may be considered harmful by programmers that were raised on
> make, a command line editor and a compiler there exist a substantial number
> of programmers that cannot work without an IDE. There is every indication
> that the majority of biologists, though capable of writing Java programs,
> cannot produce working Java code outside of an IDE because the setup and
> deployment is too complex. Result ? They'll re-use other people's bio-PERL
> scripts because that's much simpler to do.


The biojava library, javadocs, and example code integrate well into any of
a number of java IDEs -- netbeans, forte, codewarrior, jde, jbuilder &c.
as is.

If you're advocating beanification of all the biojava classes with drag
and drop of SimpleSVMClassifierModels and the like, I really don't see
that as the desired interface for working with bioinformatics data.  IDEs
work well for painting guis, not so well for algorithm development.


> In the component software industry the trend is towards supplying mini-IDEs
> with the components. For example, F1Java the spreadsheet component from
> www.actuate.com includes a designer bundled with the component. I would
> suggest that an API as complex as bio-Java could only benefit from a
> dedicated IDE, or at the very least, a collection of wizards to automate
> some common tasks.
>
> The moment, the _very_ moment that bio-Java  gets an IDE and an accompanying
> web page with some reassuring screenshots is pretty much when you'll see the
> user base expand to extremely healthy numbers. Until that day, bio-Java is
> for expert Java programmers only --which is a shame really, isn't it?


It's good for job security.  :)


Seriously though -- if you see areas in the API that need improvement,
clarification, or simplification, please point them out to the list.  The
biojava community is very receptive to useful suggestions.

And most receptive to source code patches or submissions.

Respectfully,

   michael