RE: sgml-parse and GC

Subject: RE: sgml-parse and GC
From: "Didier PH Martin" <martind@xxxxxxxxxxxxx>
Date: Mon, 19 Jul 1999 20:07:07 -0400
HI Avi,

You are right, both memory mapped files and virtual memory have the same
performance level (because the former uses the latter). The problem we have
with memory mapped files for permanency is that it is hard to grow a memory
mapped file. I cannot speak for Linux so I'll let the other Linux
knowledgeable person answer for this. But on win 32 it is not until the next
release (i.e. Win 2000) that we'll be able to have growable memory mapped
file. On actual winxx or WinNT it is not feasible out of the box except if
you re-write the file and make it bigger.

regards
Didier PH Martin
mailto:martind@xxxxxxxxxxxxx
http://www.netfolder.com

-----Original Message-----
From: owner-dssslist@xxxxxxxxxxxxxxxx
[mailto:owner-dssslist@xxxxxxxxxxxxxxxx]On Behalf Of Avi Kivity
Sent: Monday, July 19, 1999 6:01 PM
To: 'dssslist@xxxxxxxxxxxxxxxx'
Subject: RE: sgml-parse and GC


On Tuesday, July 20, 1999 00:23, Peter Nilsson [SMTP:pnidv96@xxxxxxxxxxxxxx]
wrote:
> I think the best solution would be to replace the current in memory grove
> implementation with an implementation on disk with mmap'd files. Then all
> groves would be cached and memory management would be passed to the
> OS kernel. This, I think, was proposed several times before.
>

I believe virtual memory and memory mapped files are the same thing in terms
of performance and OS management level. Allocating memory is similar to
mapping some of the swap space into your virtual memory space, except the OS
is free to choose where to place the data (possibly striping over several
disks).

What you would get is grove persistence over invocations of the style
engine, which may or may not be a win.

---
"The only words which have meaning are the last ones spoken"



 DSSSList info and archive:  http://www.mulberrytech.com/dsssl/dssslist


 DSSSList info and archive:  http://www.mulberrytech.com/dsssl/dssslist


Current Thread