From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: *
X-Spam-Status: No, score=1.2 required=5.0 tests=BAYES_00,FROM_WORDY,
	INVALID_MSGID autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,184737148aef02ac
X-Google-Attributes: gid103376,public
From: "Nick Roberts" <Nick.Roberts@dial.pipex.com>
Subject: Re: GC+HC for GNAT/GCC (was: Building a compiler)
Date: 1999/02/05
Message-ID: <79enud$lvc$2@plug.news.pipex.net>#1/1
X-Deja-AN: 440966562
References: <78sojm$crk$1@plug.news.pipex.net>
 <F6BzAA.KKI.0.-s@inmet.camb.inmet.com> <7982p7$nll$1@plug.news.pipex.net>
 <87aeyv4kbg.fsf@mihalis.ix.netcom.com> <79asc3$cq3$1@nnrp1.dejanews.com>
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Organization: UUNET WorldCom server (post doesn't reflect views of UUNET
 WorldCom)
Newsgroups: comp.lang.ada
Date: 1999-02-05T00:00:00+00:00
List-Id: <comp.lang.ada>

robert_dewar@my-dejanews.com wrote in message
<79asc3$cq3$1@nnrp1.dejanews.com>...
|A very nice project would be to try to add full garbage
|collection to GNAT ....


Now _there's_ a challenge!

Actually, I'd be really surprised if someone hasn't already done this.
Still, I'm quite happy to throw in a contribution if it would be genuinely
useful.

First, let's be clear about what we are talking about. By "full garbage
collection", I assume Robert means both 'garbage collection' (i.e.
recognising which allocated objects are no longer accessible), and 'heap
compaction' (i.e. moving allocated objects so as to remove unused fragments
from their pool).

There are several possible combinations of schemes, all with their own
peculiar merits and problems, so I would suggest that a wide selection be
provided (as different pool types).

It should be understood that all of these schemes would have various
ramifications for both the GNAT front-end intermediate code emitter, as well
as for the GCC back-end object code emitter. Locking variations would
require a special machine instruction (different for each target
architecture).

It would probably be sensible for someone familiar with GNAT and GCC to help
me to 'splice' my code into GNAT and GCC. Any volunteers?

Finally, a quote:

"Machines with so-called 'virtual storage' may give the compiler writer a
large number of logically separate segments of storage, all of which can be
of effectively boundless size. The catch is that if storage is used as if it
were boundless, the time taken to run programs will be equally boundless.
Thus, irrespective of whether virtual storage is available, you do well to
be miserly with your storage." -- P.J.Brown, "Writing Interactive Compilers
and Interpreters", Wiley 1979 [written in the days of MULTICS: for "number
of logically separate segments" substitute "linear address space"]

And another:

"Garbage collection is a feature of the run-time support for some imperative
languages (such as Ada) ..."
-- David A. Watt "Programming Language Processors", Prentice Hall 1993

In practice, full GC isn't a feature of Ada - but I reckon it should be (at
least as an option). So, like Captain Picard, let's "make it so"!

Nick Roberts