NetNews Usenet Archive 1992 #18/NN_1992_18.iso/spool/comp/lang/cplus/12341

Text File | 1992-08-14 | 1.6 KB | 32 lines

Newsgroups: comp.lang.c++
Path: sparky!uunet!mcsun!Germany.EU.net!news.netmbx.de!zrz.tu-berlin.de!Sirius.dfn.de!news.uni-stuttgart.de!rz.uni-karlsruhe.de!stepsun.uni-kl.de!uklirb!uklirb!hf
From: [e-mail] (Harald Fuchs)
Subject: Re: Garbage Collection for C++
In-Reply-To: [e-mail]'s message of 13 Aug 92 09:56:29 GMT
Message-ID: <[e-mail]>
Sender: [e-mail] (Unix-News-System)
Nntp-Posting-Host: demisec.informatik.uni-kl.de
Organization: Kaiserslautern University, Germany
References: <[e-mail]> <[e-mail]>
Date: Fri, 14 Aug 1992 11:32:52 GMT
Lines: 18

> For reference, here's KCL's solution (roughly):  Memory
> is divided into pages.  All objects stored on a given page will
> have the same size.  There is a "page table" which contains, for
> each page of memory, the size of the objects stored in that page
> (or zero for pages which do not contain GC-able things).  Given
> any 32-bit quantity which "might" be a pointer, first take the
> high bits to find the page, check if this is a valid GC-able page,
> and if so, round the page offset down to a multiple of the page's
> object size to get a pointer to an object on the page.

This works fine for Lisp systems where you have only a limited number
of basic object types and thus a limited number of possible sizes. I
don't think that it would work for C++ where the application
programmer can create lots of objects with different sizes.
But you can use this scheme if you want to GC only some of your objects.
--

Harald Fuchs <[e-mail]>