Re: [Ogdl-core] Binary OGDL spec

[Rolf V. <rol...@er...>]

Patrick Doane wrote:
> Two comments on this:
> 
>  - It might be nice to support "random access" of the file.  This could 
> be implemented by including more size information.  For example, each 
> node could contain an offset to the the next sibling and the next child.

The binary version is intended as a file format as well as a message
protocol, and this would add complexity at the emitter side and
also add quite an amount of extra bytes. I'm thinking for example in
using it to exchange information between a small device (a measuring
instrument, an inductrial device, etc) and a host, through USB, CAN or
whatever.

>  - What is the reason for distinguishing between text-node and 
> binary-node?  Is it to save space by not including length data?

We need to identify text nodes without extra (external) type or
schema information because they are used to name other nodes.
We would not be able to do much with only untypified binary nodes.
Normal OGDL is pure text, its seems natural if binary OGDL is a
superset of that.

--
Rolf.