Java vs Python

[Harry George]

I've often ported Modula-3 NT<-->UNIX by just recompiling.  The
key seems to be abstraction of the os and file system interfaces.
E.g., if you use hardcoded "/" vs "\", you are in trouble.  If you go
to the trouble of using abstracted path joins, it is a piece of cake.
Python allows (but does not require) use of those abstractions.

In contrast, I've lost portability in Java apps when one version of a
needed library was not available on a given platform (e.g., swing).
And you can of course still screw up by hardcoding those "/" and "\"
paths.

Robb Shecter <shecter at darmstadt.gmd.de> writes:

> > >Marketability?  Hm.  The thing that makes Java marketable is
> > >the tremendous amount of money that Sun has poured into
> > >getting it into every trade rag that they could possibly
> > >find. ... is there any way we can effectively market Python as a
> > >competitor to a language / platform with that kind of
> > >agressive force behind it?...
> 
> I don't this does justice to the Java idea.  Java has the design goal
> of hiding -all- system dependencies.  And, features that are dependent
> on certain operating systems are clearly seperate from the standard
> Java packages.
> 
> Python seems to go about half-way in this direction.  Some libraries
> seem to hide OS differences via an adapter sort of design, but there
> are many standard libraries that are purely only for certain types of
> systems.   Not that this is bad, it's just different.
> 
> I use Python for 90% of my new projects.  (I work only on Linux.)
> But, if I know that the software has to run on, say Linux and NT, then
> I'll use Java.  I've seen, first hand, successful cases of "porting" a
> Java program that was written on Linux to NT by just copying the class
> files.
> 
> - Robb
> 

-- 
Harry George
hgg9140 at seanet.com