Re: Where we stand in regard to the future platform / desktop technology

jamie <jamiemcc <at>> writes:

> On Sun, 2004-03-28 at 15:11, Havoc Pennington wrote:
> > On Sun, 2004-03-28 at 06:29, jamie wrote:
> > > 
> > > An open source high performance JVM is years away so you would have to
> > > wait for Gnome 4 or Gnome 5 for this option unless someone can persuade
> > > SUN to let go of its "precious" or alternatively get them to donate code
> > > for JITs and other missing bits. Running a JVM without such JITs as a
> > > short term solution is not a good idea (would be lot slower than
> > > python).
> > 
> > That's certainly not the case. We have IKVM, gcj, and Jikes among others
> > and open source Java is perfectly usable.
> At various levels, yes, but there is no current open source java + JVM
> that is close to complete. Classpath will require more time to complete,
> GCJ does not fully use ClassPath yet and AFAIK none of them can run
> Eclipse in its entirety. It may be usable or a partly implemented java
> might be usable, yes, but its a far cry from a high performance version
> that a desktop solution would require - my estimate is several years
> away based on the current slow pace of development.

Your completeness argument is bogus. Let me fire off a few rhetorical questions
to set that "argument" straight:

Is GNU tar "complete"? Is it as "complete" as Solaris tar? Is gcc "complete"? As
complete as Microsoft VC++? Is mono "complete"? As complete as Microsoft's .net
SDK [1]? Will GNOME ever be as "complete" as Windows XP? :)

Finally, the most important question of all: does that matter? In the context of

You're comparing apples and oranges here. The free java runtimes and the
non-free ones have different qualities, goals, etc. While it would be good
material for a discussion or even /., I think quite
pointless in the context of GNOME development.

The more relevant question is: are the free java runtimes good enough for
developing GNOME & GNOME applications? My impression from following gnome-java
bindings mailing list is: yes, they are. They are also getting better and better
at running the other, non-GNOME stuff, like Eclipse, OfBiz, JBoss, and whatever
makes some people happy, but that's quite irrelevant here, I guess, since these
projects are not part of GNOME.

If you are dissatisfied with the performance of free java runtimes, you may be
inclined to help to improve it. One of the interesting integration projects for
a prospective, engaged and well-educated developer in runtime technology in
particular, pretty much just like you, would be to bring Kaffe and gcj together
by finishing the work that was done in this area, and merging in the 'better
than hotspot' jit, gc and threading performance improvements from the latte
project back into Kaffe. Your fame would rival that of Miguel, at least :)

On a side note, I don't get why the Mono crowd is so scared of free java
runtimes, that they have to advocate their runtime by claiming how inferior
others are. If mono's JIT is so ueber-cool, it will be adapted and merged in by
some free java runtime project, like Kaffe or gcj, and that will be the end of
that argument. If the C# bindings are good enough for building GNOME apps, now
that's a great thing in my opinion, regardless whether C# becomes a part of the
GNOME core, or not.

I am really glad to see projects like mono and dotGNU provide more options for
free software developers. You won't see me pointing the finger at mono and
claiming that it is inferior to Kaffe because it's not as "complete" or
performant as Microsoft's .net implementation. 

So instead of advocating an "either mono or free java" solution, you may want to
think about a world where the free java and free C# camps don't play to the tune
of the marketing battles that the respective large companies play, but instead
work together[2] on advancing the state of the art of free software development. 

Just a thought from outside of GNOME.

dalibor topic

[1] Or whatever that's called nowadays in the non-free world.
[2] Or at least peacefully coexist :)

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]