[rc5] Next Crack

Richard Freeman rfreeman at netaxs.com
Mon Aug 25 13:00:00 EDT 1997


On Mon, 25 Aug 1997, Ryan Krueger wrote:

> 
> suggests that JIT compilers can be faster than C++.  Of course this will 
> vary greatly, but it's encouraging.  Compiled Java will never be as fast 
> as assembly, but is in the neighborhood.
>

I still think that the majority of the people who will run the client will
run it with or without Java - making the speed loss due to high-level
instructions significant - when you consider the benefits of
assembly-optimizing and how it often enables you to get large loops inside
the cache and things like this, I doubt Java will ever be competitive.
 
> Anyway, a Java based client is not impossible and the speed lost by not 
> going to assembly can be made up by more people running the client.
> 

This I might agree with - I think a better compromise would be making the
client available in both forms - so you get the best of both worlds.

> Another idea:  There is a lag in writing the different platform clients, 
> Java could be the bridge to start a new project on all platforms and then 
> swap in the native code when it gets done.
> 

By far the best idea - Java would let you get your foot in the door until
you get the real programming done - the high-level language would probably
make it a lot less buggy at first as well.  You wouldn't have as much
pressure to keep releasing poorly-tested code to include more
architechtures.

One thing people keep saying is that Java will bring in more people and
more platforms.  I haven't been really involved with the whole Java thing
but I do have an idea of the basics.  What exactly would somebody need to
install on their computer to run Java in the first place.  I understand
that MacOS 8 might help out the Apple owners, but the vast majority of the
idle CPU-time is running win95 - which does not run Java without 3rd-party
software.  I know that I don't use Java on my computer - to me I think
that a well-written native client would be a lot easier to install than a
cross-platform client which requires additional software to run.  The
native client could have more specific instructions (since it isn't
cross-platform you don't need generic instructions on how to run a program
or things like that), and could tap features of individual computers (like
running as an NT service, or hidden in win95).  

I think that Java is a good idea for those who are able to run it or for
those who would have an incentive to use it.  I know that if the install
program were well written, a win95 version would only require one
self-installing exe to download and run, and it could be written to be
very easy to use.  Plus if you do it right you could integrate it with teh
standard OS features for program installation (using the version manager
on IRIX machines, or the auto-remove feature on win95, etc...).  I believe
that Java is only a stopgap measure at best to try to squeeze in those
last unusual computers which are present in unusually large proportion
in areas where programmers hang out...  While many of us on this list may
have seen Java-based workstations and things like that, I doubt that the
average business-executive who has a Pentium 150 he is using an hour a day
as a typewriter would be privy to them.  I think that this is the area
where the largest number of idle CPU-time is available.  The developers
who have the fancy software and machinery are more likely to be using
their CPU-time in more profitable enterprises...

I don't mean to criticise anyone, I am only suggesting that aiming at the
typical-computer illiterate end-user will get you the most CPU-time, there
are only a relatively small number of people who have access to
workstation-farms, and this is the bulk of our processing force
rightnow...  I believe that this sort of user is best served by
well-written OS-specific software, with nice fancy help files and things
like that...

Still I have to tell everyone to keep up the good work - this is an
amazing volunteer effort!

-----------------------------------------------------------------
Richard T. Freeman <rfreeman at netaxs.com> - finger for pgp key
3D CB AF BD FF E8 0B 10 4E 09 27 00 8D 27 E1 93 
http://www.netaxs.com/~rfreeman - ftp.netaxs.com/people/rfreeman

----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list