home *** CD-ROM | disk | FTP | other *** search
- Path: wcap.centerline.com!vajra!chase
- From: chase@centerline.com (David Chase)
- Newsgroups: comp.lang.smalltalk,comp.object,comp.lang.c++,comp.lang.java
- Subject: Re: The Good, the Bad, the Ugly, and the W
- Date: 12 Apr 1996 19:54:56 GMT
- Organization: CenterLine Software
- Distribution: world
- Message-ID: <4kmceg$76t@wcap.centerline.com>
- References: <RMARTIN.96Apr10133335@rcm.oma.com>
- Reply-To: chase@centerline.com
- NNTP-Posting-Host: vajra
-
- In article 96Apr10133335@rcm.oma.com, rmartin@oma.com (Robert C. Martin) writes:
- > You missed my point. new/delete can *always be made* to be completely
- > deterministic. Consider:
-
- > [user-written memory allocation]
-
- > This is a typical strategy in real time systems.
-
- There's nothing stopping you from doing this in a garbage-collected
- system, either. I think that's why we missed your point, because it
- was trivial. There's no rule against pre-allocating a bunch of
- objects in a garbage-collected system, and writing your own
-
- thing = my_alloc()
-
- and
-
- my_free(thing)
-
- subroutines to move them on and off your private freelist as
- necessary. You're getting no assistance from your garbage collector
- *for that particular piece of code* but it isn't interfering with you,
- either, and if you aren't allocating any memory, then the collector is
- not running.
-
- There are also approximations to real-time garbage collection
- available, though if I understand correctly, microsecond-level
- deadlines are still difficult to meet.
-
- speaking for myself,
-
- David Chase
-
-