On 07/09/14 16:25, Thomas Wrobel wrote:
Of what benefit would contributing this to wave be?
Doesn't wave do a superset of this functionality already? (albeit with
messier code)
In a similar way to XML being a superset of JSON (and Wave includes bits
that *neither* does well or indeed at all).
Seems (possibly) more useful for Wave to contribute its (federation)
functionality to this or a fork of it. Then you would have a new
pseudo-wave that does much of the same stuff, but with a much neater
codebase (and mobile support) to build from.
Alternatively if elements of this could replace waves code to
simplify/neaten it that might be good...but at least from an outsiders
perspective that seems rather hard.
Regarding the point earlier about rich text in json - wouldn't it be easier
to use html encoding of styled text? To my knowledge html strings work in
json just fine as long as a few things are escaped. Or isnt this possible
with the OT method being used?
HTML strings fit in JSON--and technically it's not JSON we're talking
about but the data structures JSON serialises--the problem is that then
you're editing a textual serialisation, rather than the actual data
structure. Given that GR, Wave, and Etherpad all use something other
than a plain text string, it's probably reasonable to conclude that
using plain strings to represent rich text opens too many possibilities
for an OT or similar system to get into a state where the document is
unparseable.
Or for that matter, for creating pairs of operations which cannot be
resolved in any reasonable way by an algorithm that isn't aware of the
syntax involved.
