[2026] in java-interest
Setting Boundaries (was Re: JDK Question)
daemon@ATHENA.MIT.EDU (James C Deikun)
Fri Sep 22 03:40:47 1995
Date: Thu, 21 Sep 1995 20:35:05 -0400 (EDT)
From: James C Deikun <jcdst10+@pitt.edu>
To: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Cc: Philip Brown <phil@bolthole.com>, java-interest@java.Eng.Sun.COM
In-Reply-To: <Roam.1.2.811468729.5987.avh@acorn>
On Mon, 18 Sep 1995, Arthur van Hoff wrote:
> The mandatory networking classes are in java.net. We've got some
> additional classess that we've put in sun.net.* to avoid name
> clashes with other software vendors.
>
> We are in control of what is part of the java.* classes that
> are available in every Java environment. As time goes on we
> will add more functionality in a controlled and upward compatible
> manner.
Speaking of that, I've wondered this since the JDK came out:
Should all this stuff really be required for every Java implementation,
or just every HotJava implementation?
Yeah, some of this stuff is awfully nice, but it sure puts a lot of
weight on the head of the implementor. (Not that this is any surprise
considering the involvement of large corporations here.) I really can't
see why for some purposes a simple implementation containing only files
and text i/o wouldn't suffice. For example, if one were producing a
super-optimized Java runtime for use on dedicated compute servers, I don't
see what the use of being burdened with writing awt classes would be.
Indeed, in contexts like that it would be good for user classes to
*expect* a more restricted environment. After all, on a machine in
Maine, opening up X windows wouldn't be particularly useful to me here in
Pittsburgh, even if they were redirected properly to my X server. (Hey, can
you say "slow"?)
--
James "e unix pluribus" Deikun
-
Note to Sun employees: this is an EXTERNAL mailing list!
Info: send 'help' to java-interest-request@java.sun.com
