From: "Mike Champion" <>
> I'm hoping we can get enough information to get past the "binary XML is
> evil", "XML is bloated and slow", etc. generalizations and understand the
> alternatives and tradeoffs more thoroughly.
+1
What would also be interesting would be, for different compression or binarization
schemes, which document sequence patterns are optimal. I think there are three stages
- rewriting (to optimal forms, e.g. to get as much information implied by sequence
rather than bits)
- binarization
- compression
any of which could be a null transform. At the moment, we have no common
schema languages which tell us "sequence is important here: don't rearrange things."
We can say "elements can appear in any order here" but we cannot say "their
order is not significant". Until we get that, the first stage of rewriting into optimal-for-
binarization-and-compression sequence (a subset of marshalling, I suppose) has to
be done by hand by the schema writer, who has no idea.
One
