John Keiser

Another issue where I think we probably can push large file / input streaming past 1.0 ...

Is this a common enough case to block? There's a similar issue for "large integers": #167 . Moving to 2.0 for now; feel free to move back / disagree!

A note: by default, on demand assumes keys are raw utf8 (no escapes) when looking up fields, getting a significant speedup for the overwhelmingly common case.

(As such, I'm feeling like we don't absolutely need this for 1.0. Moving out; feel.free.to.move back / disagree :))

I don't think most json field lookup algorithms consider grapheme equivalence or normalization anyway :)

I think this is what's left to do for actual streaming parsing.

> Instead of walking dom trees (even with ondemand) we can choose to run a query. Sorry to be short--my time today is like that :)--but what I'd say is,...

So, I feel like we could make a 1.0 without true input streaming support; on demand's streaming frontend seems to give a lot of the performance benefit we're looking for...

Not feeling the urgency for 1.0.

Backwards and forwards iteration is a real thing, and seems worthwhile. Local 15-byte strings would be awesome! One thought: if we broke up the storage across two separate tapes--each with...