[Xml-bin] Binary XML - summary of discussion to date

[Al B. Snell]

Ok. I'm creating a mailing list for this issue to get it off of xml-dev,
since I'm sure not everyone wants to hear us ranting :-)

To date:

1) There is still debate over the gains to be had from a binary
format. There are widely varying claims of massive increases in efficiency
or, in some cases, decreases. However, several domains that require binary
storage (when there are embedded binaries, or random access, or in
minimal-hardware systems) have certainly been identified.

2) There are existing binary-XML efforts, but many of them are domain
specific; it would be advantageous to try to create a standard that people
agree with to aid interoperability.

3) Some people are worried that standardising a binary format will
encourage vendor-specific extensions to that format, but this has not been
a problem for zip/png/jpeg and all the other successful open binary
formats out there; in general, it is not "binaryness" that is at fault for
this, it is vendor control of the format. Similar examples can be made for
text files.

4) Regarding human-readability; if the format is ubiquitious like JPEG or
ZIP or gzip, then there will be tools to view them, just like there are
tools to examine zip files and so on. The human readability argument
really only tells us that non-public proprietary standards are bad; it,
again, is not really about text vs. binary. The fact that text viewers are
very widely available already is a plus point for textual encoding, but
it's probably of similar magnitude to the speed gain of binary encoding in
many applications :-)

5) There are two quite well developed binary-XML projects lurking in the
wings - the ITU XML / ASN.1 project - which requires feedback, an open
reference implementation, and implementing in standard XML toolkits, and
SDW's random access indexable format, which will be ideal for XML
databases and similar large-dataset applications; by the sound of it, it's
at a similar stage to the ITU one. It is my belief that we would probably
be best helping one (or both) of these projects acheive
ubiquitousness; perhaps the ASN.1 one (if we choose a good encoding, eg
PER) for pipeline streaming applications and SDW's one for data storage.

6) http://lists.warhead.org.uk/mailman/listinfo/xml-bin - join, discuss,
don't flame :-)

7) In a few months I may be available to write an RFC. That'd be nice.

ABS

-- 
                               Alaric B. Snell
 http://www.alaric-snell.com/  http://RFC.net/  http://www.warhead.org.uk/
   Any sufficiently advanced technology can be emulated in software