Lists Home |
Date Index |
- From: "Anders W. Tell" <firstname.lastname@example.org>
- To: Rick Jelliffe <email@example.com>
- Date: Sat, 27 Mar 1999 13:14:39 +0100
Rick Jelliffe wrote:
> I have done a few tests on how much compacter forms of XML (e.g.
> shortrefs) impact arrival characteristics of document packet-groups
> under TCP/IP compared to compression. If your packet size is small, and
> you really need to get at data in the first packet (so that you can
> piggy back request for auto-linked resources in with the ACK for the
> first packet group), then more compact forms of markup may make a
> difference. But in general, compression is more effective. (It also
> depends on where the bottlenecks are in your data path.)
It seems that there are more use-cases which should benefit from having a
compressed or a binary format.
I made some tests using following XML data.
<xi4 value="0" name="VALUE"/>
<xi4 value="32768" name="VALUE"/>
The resulting sizes was:
XML 602830 (Standard XML text)
FML 131143 (Fast ML, a binary ml that Im working on)
XML.gz 75528 (gzip'ed XML text using -9 as compression rate)
FML.gz 20886 (gzip'ed Fast ML using -9 as compression rate)
The facinating result here is the dramatic reduction in size obtained by first
converting to FML and the GZIP the markup stream.
> And select your element and attribute
> names so that their length is inverse to their frequency, as much as
> possible: so use "a:s" not "abracadabra:shazamarama" (you may even make
> two versions of your DTD: an authoring one and a transmission one.) One
> pof the main bottleneck on many SOHO systems is the modem speed:
> reducing the end-to-end character count means fewer packets, and more
> data arrives earlier, so more auto-links are followed earlier.
On the other hand there is a big drawback using "manual tag compression"
which is Readability.
/ Financial Toolsmiths AB /
/ Anders W. Tell /
xml-dev: A list for W3C XML Developers. To post, mailto:firstname.lastname@example.org
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/ and on CD-ROM/ISBN 981-02-3594-1
To (un)subscribe, mailto:email@example.com the following message;
To subscribe to the digests, mailto:firstname.lastname@example.org the following message;
List coordinator, Henry Rzepa (mailto:email@example.com)