RE: [xml-dev] JavaScript (was Re: [xml-dev] Whither XML ?)

["Alexander Philippou" <alex@noemax.com>]

> > Just use Fast Infoset. There are several FI implementations around. Combined with GZIP they should give you excellent results in terms of both compactness and performance.
> 
> But doesn't that require you to first know the schema of every thing you'll be sending and receiving?  That's not always possible.

No, Fast Infoset is not schema dependent (maybe you were thinking of EXI?). There is never any requirement to know the schema in order to encode/decode an FI document. To use FI you just replace the XmlTextReader/Writer with the XmlFastInfosetReader/Writer. The FI writer supports three representations (literal/binary/restricted alphabet) that can be used to encode numeric/datetime/boolean values (binary values are always encoded as binary). If a numeric/datetime/boolean value is written using the methods provided by the FI API (e.g. WriteValue(Double)) then the writer will select the representation it considers as most efficient for this particular value, otherwise this value will be written as literal. When writing a value the representation used to encode it is also written within the FI document. This means that the same document can be encoded in several different ways (for example, with emphasis on compactness or on processing performance) but the FI reader will always be able to decode all values irrespective of how the FI writer encoded them. So no schema is needed either when encoding or when decoding FI.

Alexander