[gclist] ref-counting performance cost
Thu, 12 Oct 2000 16:55:27 -0400
At 08:20 AM 10/11/00 +0000, Richard A. O'Keefe wrote:
> Also, it is perceived by them that Java is the only environment
> (possibly .NET as well) which makes garbage collection worthy of
> (because it is becoming so much more popular), and so I am also being
> swayed towards focusing on performance specifically in the context of
> a JVM.
>This is a very odd perspective. A limited, limiting, and flawed language
>becomes popular, so therefore it defines what is worthy of study?
I agree completely, but in practice, very much research follows
that route. The major flaw that I see in pursuing that course is
that actually handling all of Java, (or C++, or Fortran, or .NET)
turns out to be intractably large for any single gradual student,
so they subset, and then (as appears to certainly be the case with
Java papers I have seen) they subset themselves into irrelevance.
Worse yet, while you're doing research on your Java-like
toy, people working on the real thing are going all-out to
solve the same problems in the real language. Our knowledge
of the dark corners of the Java type system led what got
published by several years, and what got published was
couched in some incomprehensible notation that made it
not entirely clear to me whether it was correct or not,
and I didn't care enough about their result (since we had
our own correct result) to take the trouble to find out.
The number of people who (1) care about this problem
in Java (2) don't already know the answer and (3) find it
easier to understand the incomprehensible notation than
solve it themselves is approximately zero. Does that
sound like good research to you?
Given this, I think it makes substantially more sense to
pick a language that is a good match for the problems you
are studying, and solve them there as quickly as you can
so you can get that degree. Try not to write too much
code in the process -- it has an annoying habit of wasting