Jerome,

I think this is only an issue for developers trying to use the DOM. A parser
used for SAX should essentially work along the data stream, and not need as
much saved context as when using a DOM instance. My own 2-year old XMLite
library (one of the many bearing that name), using 32-bit file IO, is
limited to 2GB but appears to work fine up to that size, at least on XP, so
I don't see why a serious parser like Xerces or MSXML should have any
trouble with larger files.

The issue you raise is interesting in another way, though: I hadn't
considered how I was going to read gbak files above 2GB using the Win32 or
Posix API, and I'll have to examine this. For writing, I think the solution
will be along the lines of Gsplit: spread data over chunks less than 2GB...
interesting issues with element nesting vs data size in sight :-)



Date: Fri, 4 Jun 2004 16:25:18 +0200
From: "Jerome Bouvattier" <JBouvattier@...>
Subject: Re: RE: Gbak work ?

Frederic,

> - having backups available in an easily accessible and
> documented XML format guarantees a "way out" with
> all enterprise data in case any of these unforeseen
> problems should happen, within a narrow time window

Out of curiousity, what XML parser will use to read a multi-giga xml file ?
Or maybe it's just the gbk's structure that you want to see in the xml file
?

Regards.

--
Jerome