At 11:34 PM 3/7/00 -0500, "Todd L. Miller" <[EMAIL PROTECTED]> wrote:
>You keep giving a 4 MB figure for the distribution size -- does
>that include the (required) class library (uncompressed zipfile)?
The binary editions created by Corrado Santoro
<[EMAIL PROTECTED]> are 4MB, more or less. The josbin-1e.tar.gz
contains jjos-nbi. And, jjos-nbi definitely includes all those classes.
Based on size alone, I'd guess that josgrub-1e.tar.gz contains jjos-grub
and jjos-grub includes all those classes. But, I don't know for sure.
Somewhere in the build process, classes.zip is "linked" to jjos-nbi and
jjos-grub. I think an executable is concatinated with classes.zip to create
them.
Let's look ahead. Should we perform the "concatination" step before or
after the files are distributed? If this step is performed before, everyone
must test the same classes at the same time. That might not be a bad idea
for now.
On the other hand, concatination is eay. The cat command is widely
available. This step could be performed *after* the files are distributed.
Anyone could replace classes.zip with their own experimental classes. We
could jump to parallel development. Different people can test different
subsystem at the same time.
Maybe the development of JOS-specific and decaf-specific classes should
move to a CLASSES module in JOS CVS. By creating a binary distribution of
jJOS/decaf, we can open up the development of classes to more Java
programmers. Many hands may lighten the load. But we can't do it
prematurely either.
_______________________________________________
Kernel maillist - [EMAIL PROTECTED]
http://jos.org/mailman/listinfo/kernel
