[Prev][Next][Index][Thread]

Re: Evidence that Functional Objects is real?

Maury Markowitz wrote:
> 
> "Rob Myers" <robm@h2g2.com> wrote in message
> 3A0683C9.5D432F45@h2g2.com">news:3A0683C9.5D432F45@h2g2.com...
> > Or if a plane crashes on the office... Risk management is part of any
> > project.
> 
>   Sure Rob, but if you're claiming that using Dylan isn't a much higher risk
> than Java (for instance), I think you need to illustrate it a little better
> than "any technology". For instance I can see no circumstances at all under
> which a Java project would be put at risk in the same way, the
> infrastructure is way to big.

I did Java development from 1996-1999, some server-side, some
client-side. Lets focus on client side. Each release of the (IE 3) JVM
had its incompatibilities, often little floating point bugs that came
out on only 3 or 4 numbers across the whole range of int. Which doesn't
sound like much, but we were producing the front-door software for
several multi-million dollar companies. Each new release of IE 3 meant
new bugs and obviated fixes. 
Then along came IE 4. We didn't change anything, but IE certainly did.
At least one big client got locked in to IE 3 because the software
wouldn't run on IE4 without rewriting. They weren't happy.
We wouldn't have had these problems with Dylan, OS releases are less
frequent than browser releases and we wouldn't be at the mercy of VM
bugs. Yes, we'd have had other problems, but not the ones we had with
Java, which is meant to have fewer problems..
 
> With Dylan I have to train programmers in
> the language, find ones that want to work (we have people say no because of
> the _Java_tools_ we use - not JSP? no way!) on it without feeling
> marginalized, 

Java and Java developers are not a magic bullet (no technology is). If a
programmer has so little imagination that they feel "marginalized" on a
project with a new technology and a blank sheet for design then I'm not
sure why they've been hired. I've programmed JSP, and seen proprietary
servlet APIs. There are very few situations where ASP or JSP (or even
P*RL) wouldn't do or be better, unless chasing API versions is a major
goal of the project. I've seen big Java projects overrun by a year or
more (not in any company I've worked for I hasten to add), with no-one
willing to take the blame for a bad choice of language and environment.
I've never programmed Dylan commercially, but if I did I'd bring the
same imagination and learning ability that I did to Java and I'm
currently bringing to cross-platform C++. Motivation is not up to the technology...

> and have the ability to choose from one supported product and
> one unsupported product.

Gwydion Dylan is supported by a team of volunteers, much the same as
Perl, Apache or egcs. MS products are supported, but being able to
apportion blame isn't always as important as being able to get something
fixed. Open Source as a general principle is becoming more accepted.
Yes, there's less support for GD than the projects I mention, but it is
*high quality* and has been around enough years to be regarded as stable.

> > Java is hot, but imagine if MS dumped Java, didn't update the runtime,
> > didn't implement new APIs in it, and didn't update their compiler. Oh,
> > hang on... :-)
> 
>   Exactly.  And because Java is big, it had zero effect.

I wouldn't describe decimating Java on PC as zero effect. JavaSpaces,
JavaMedia, Jini et al. are *not* taking off, Applets aren't flavour of
the month anymore, and there are only a few GNU Java applications still
doing the rounds. Does IE5 run them? Do IE5 users care?

> > I agree there's risk to using any technology, but no more so with Dylan
> 
>   That sounds shockingly like appologizing.

It was meant to be a FUD check, but as ever, if I'm apologizing I'm
sorry. :-)

- Rob.

-- 
Rob Myers - http://www.robmyers.org/   H2G2 - http://www.h2g2.com/
MacOS wonderfulness for The Hitch Hiker's Guide to the Galaxy Game.
"Smash Global Capitalism! Spend less money!"
References: