NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c++-part2/17774

Internet Message Format | 1996-08-05 | 1.9 KB

Path: svnews.ubinet.ubs.com!ubszh![e-mail]
From: [e-mail] (Ian Johnston (by ubsswop))
Newsgroups: comp.lang.c++
Subject: Re: Why don't you use garbage collection
Date: 17 Apr 1996 13:06:17 GMT
Organization: UBS
Distribution: world
Message-ID: <4l2qc9$[e-mail]>
References: <[e-mail]> <1996Apr16.110526.1846@ittpub> <[e-mail]>
NNTP-Posting-Host: nol2179.fh.zh.ubs.com

In article <[e-mail]>, [e-mail] (Lars Farm) writes:
|> In article <1996Apr16.110526.1846@ittpub>,
|> [e-mail] (Wil Evers) wrote:
|>
|> >OK, let's get serious about this:
|> >
|> >1. Don't we think C++ is complex enough as it is?
|>
|> See the Detlefs/Ellis GC proposal. Seems a fairly small change in the
|> language compared to what is done with templates and exceptions and rtti,
|> and... A couple of keywords: gc/nogc to be used optionally with class
|> declarations and optionally with operator new to select collected or non
|> collected heap. Later gc/nogc overrides previous. There is a way to
|> finalize, but the need for this is made much smaler. What is inside most
|> copy/op=/dtors? Code to copy heap data and release data held by pointers.
|> What would copy/op=/dtors look like when pointers can be shared and you
|> don't have to say delete? Many of them would become redundant! The
|> (current) compiler generated op= and copy would very often suffice. The
|> minority of objects with non empty dtors would be deleted manually just
|> like now. This makes things simpler, not harder.

Well, most of your classes may manage memory in ctors/dtors/op =, but
I have many many classes which manage other resources, not memory. In fact,
the classes I have which manage memory are in the minority.

So, how does the Detlefs/Ellis, or any other GC, finalise my non-memory
resources?

Ian