RE: interpreted languages

Let me see if I understand this correctly. You agree that that programs should be machine and system independant, right? You don't agree that they should be interpreted, right?

If that's true then what is java? You need a JVM which executes java byte code which runs on the JVM, which runs on the machine. Isn't that interpreted?

Why Java? Why Java, which uses a rather slow AWT? Why not use Python/Perl which uses fairly fast C/C++ bindings to native toolkits, like gtk+ or Qt?

I agree that using an interpreted language would 'probably' have more of a battery drain. But none is going to want to put linux/bsd on these devices if there aren't any apps.

Using Python (I'm a pybigot, sue me.) with things like glade you could actually write the apps once and target them to multiple platforms with multiple screen sizes. All you would need to do is have some logic inside your program to either magically pick the proper glade file or have the user specify it on the command line (or in a twm menu).

Another reason to use interpreted languages (like Java, Python, Perl, Tcl/Tk) is that there are lots of people familiar with them. While a LOT of people are familiar with C, most of them will not be familiar with writing ARM/linux apps in C. There are also a lot of people (like myself) who can read/understand C at a moderate level, but can't write anything bigger than 'hello world' and some simpler apps in it. I can however write (and have written) a lot of larger apps in python. I think there are probably lots of people out there with the same type of skill set (maybe in a different languages like Perl, Java, Tcl/Tk, etc.).

I agree with you that no one will want to use a device if there are no apps. But we either need to build them or adapt existing ones. I think the best way to build these apps is to do it in an interpreted language with a native toolkit binding (i.e. pygtk, perlgtk, etc.)

> -----Original Message-----
> From: Dirk Grunwald [mailto:grunwald_at_cs.colorado.edu]
> Sent: Wednesday, September 27, 2000 10:44 AM
> To: Keith Packard; Russ Nelson
> Cc: handhelds_at_handhelds.org
> Subject: RE: [Handhelds] interpreted languages
>
>
> People are combining two issues. The first is "should the programs be
> machine and system independent" and the second is "should
> they be interpreted"?
>
> Of the latter, probably not, for the reasons mentioned. Of the former,
> definitly. My experience with the Itsy project when I was at
> WRL was that
> we had more people start app development once a simple
> machine-independent
> development environment was available. This allowed people to
> test on one
> machine and run on another. Also, the development environment (java)
> made software development easier. By comparison, the existing graphics
> environment was one that had been made incompatible with the
> mainstream
> version because e.g., the modular structure using indirect
> jumps had been
> removed to promote efficiency.
>
> One important thing to remember is that there's no point in
> having a pocket
> computer with only one or two very efficient applications.
> Allowing people
> to explore the range of opportunities for pocket computers is
> the most thing
> right now.
>
> We published an analysis of some aspects of the Java system
> in a sigmetrics
> paper. More recently, we published energy aspects of memory
> system allocation in the IMMS symposium. It's not always the case that
> e.g., garbage collection and JIT's are that expensive - in
> fact, running
> a JIT wasn't that much of an overhead on the itsy.
>
> A better approach all around would be to use a mechanism like
> "Omniware"
> or a machine independent format that allows more optimization and more
> rapid translation than Java.
>
> -----Original Message-----
> From: handhelds-admin_at_handhelds.org
> [mailto:handhelds-admin_at_handhelds.org]On Behalf Of Keith Packard
> Sent: Tuesday, September 26, 2000 6:18 PM
> To: Russ Nelson
> Cc: handhelds_at_handhelds.org
> Subject: Re: [Handhelds] interpreted languages
>
>
>
> > Question: Should most applications for the iPAQ be written in an
> > interpreted language?
>
> I think no. Here's my rational:
> 1) handheld computers have limited battery power.
> 2) each instruction/memory reference takes some joules.
> 3) interpreted programs execute more instructions and reference
> more memory than compiled programs.
> 4) interpreted programs will reduce battery life.
> 5) ref: newton battery life with NewtonScript
>
> Even if the interpeted program is "fast enough" for the user,
> interpreted
> programs will waste precious joules of battery life.
>
> keithp_at_keithp.com XFree86 Core Team SuSE, Inc.
>
>
> _______________________________________________
> Handhelds mailing list
> Handhelds_at_handhelds.org
> http://handhelds.org/mailman/listinfo/handhelds
>
>
> _______________________________________________
> Handhelds mailing list
> Handhelds_at_handhelds.org
> http://handhelds.org/mailman/listinfo/handhelds
>
Received on Wed Sep 27 2000 - 08:22:48 EDT