From: Tom Sawyer <transami@tr...> - 2002-09-10 10:03:33
one last thing and i'll give it a rest.
by the way, has anyone looked at this: http://www.waterken.com/dev/Web/Object/
well, their serve just went down. guess their technology ain't so good
oren. i have to agree with you to a large extent. you've made a strong
argument. and i'll bite on half of it. but i think order and duplication
are just too common not support inherently. they are used all over the
place! and certainly fall within the 80% use case range: log files,
config files, sgml, xml, html, prolog, php,... (just to name a very
common handful) now whether we offer a distinct syntax for these or
relegate them to types is one thing. i'll settle either way, though i
think it would be nice to have some easy syntax for these. but that's
not as important as the more pressing issue: that core yaml needs to
support logical and ordered maps inherently. in other words, we at least
have to have a logical model to work with. if we do not, the generic
implimentaion will be a cludge. faking logical models through
translated, multiple node functional models. inefficient at the very
least. does that make sense?
so okay, if we don't add the syntax, fine. its just a shorthand
after-all. and perhaps it's something that will be seen useful enough
down the road and be added. even so, can we please have the model?
have you taken a look at my wiki page on this?
you will see that a new model is presented. i spilt the serial model
between serial and aliased. this is important too, so that comments,
directives and the like can be accessed via native models (without
having to parse the raw text) and/or via higher(lower?)-order ypath
queries, for example.
we will get a lot of milage out of these, syntax or no.
tom sawyer, aka transami