[Home] [Groups] - Message: [Prev in Group] [Next in Group]

nu.kanga.list.mud-dev

8491: [MUD-Dev] Re: Bruce Sterling on Virtual Community goals

[Full Header] [Plain Text]
From: Chris Gray <cg@ami-cg.GraySage.Edmonton.AB.CA>
Newsgroups: nu.kanga.list.mud-dev
Date: Wed, 21 Oct 1998 23:19:30 -0600
Organization: Kanga.Nu
[Jon A. Lambert:]

 >You gotta have an assembler for those who want to touch the virtual 
 >metal. ;)  An assembler is probably only useful for quick testing 
 >of the bytecode.  This is something so simple to produce, if and 
 >when needed, it's not worth giving it a second thought.  In fact, 
 >labels are probably an artifact of the assembler and belong there.    

An assembler is easy yes. But, if you have one, then you really do
have to verify your byte-code before you try to execute it. That's
extra code and extra work. If you only generate byte-code via a
compiler, then you have the luxury of removing all of the checks once
you have it working. Free faster execution!

If the byte-code gets sufficiently complicated that you can't easily
write MUD-language code that generates the codes you want to test,
then perhaps its too complicated. If you can't generate a given
sequence that way, do you care? If you, the designer of the system
can't convince the compiler to generate it, then you can feel some
justification that no-one else will either, so actually testing the
sequence can be skipped. A bug waiting to happen? Possibly, but if
the byte-code is simple enough, visual inspection by a couple of
people ought to find the problems. (I'm a *big* fan of code reviews,
at least the informal ones we do at work.)

--
Chris Gray     cg@ami-cg.GraySage.Edmonton.AB.CA