> > > I wrote it. I used XML in the end because it can map a complex data > structure into a single document. In that case what you lose on the > swings (XML) you gain on the roundabouts (TCP/IP), particularly if you > use HTTP for transport.
Just a thought. The TechFell protocol as used in uSQLite would be quite usable (almost) as is over HTTP. It would ignore any lines in the header section that do not start with ':' i.e. all the <html> header info. Then passing to the record section each line would be a field as normal. It would be necessary to include an additional ':R' to toggle the recordset mode off at the end so that trailing html ( </body> etc. ) does not get shoved into fields. ETX mode would not be available of course (AFAIK), but I could perhaps escape control chars in fields C style ( \r etc). Two interesting side effects: 1) It would be possible to use HUMAN mode to get a readable HTML page of a query (no interest to people with Apache/PHP servers or similar, but some minimalistic HTTP servers on emebedded systems eg. invoke a user callback for each request). 2) By closing a recordset with a :R it would perhaps be possible to retrieve multiple recordsets in a single query, which is great news for use over the internet when you need a lot of little result sets. Yes I know this is balsphemy, embedded data in HTTP is supposed to be <XML>, but any compelling reason why this simpler approach may not be adopted? it is, apparently, OK to embed scripts into HTML using whatever sysntax you want, why must data be compelled to be bloated?
