Lists Home |
Date Index |
Stephen D. Williams wrote:
> I have a constraint that bars this solution: data must be
> self-describing and not require specific pre-defined schema due to the
> versioning/schema migration issues that this requires. There are many
> circumstances where requiring a pre-arranged schema and derived codec
> engine is not useful.
That's far from impossible, the way we do it the compiled schema can optionally
be stored with the encoded data. I've been recommending that for storage one
should use that method (since data outlives applications) but for ephemereal
transmissions it shouldn't be needed.
I also have ideas about how to store it once for multiple files.
Robin Berjon <email@example.com>
Research Engineer, Expway http://expway.fr/
7FC0 6F5F D864 EFB8 08CE 8E74 58E6 D5DB 4889 2488