Re: Is Lua used as a data representation language?

[steve donovan <steve.j.donovan@...>]

On Sun, Jan 6, 2013 at 5:35 PM, Rapin Patrick <rapin.patrick@gmail.com> wrote:
> Why aren't boolean values true and false accepted in LTIN?
> In JSON they exist, and can be very useful.
> Also, nil should be an acceptable value, although not completely necessary.

That _is_ a strange oversight.  As for nil, it's a bit like
introducing a foreign word in Lua; we can have NIL, NAN, INF, etc.  We
retain 'strict Lua subset' property (i.e. LTIN must be parse-able with
Lua) by saying that the loading environment must provide standard
values for these constants.

Might be useful to think about what a schema for LTIN might look like.
I know JSON was a reaction against the big XML machinery movement, but
LTIN is a good format for expressing what another piece of LTIN must
contain.

E.g. have a look at how Hisham does table verification in Luarocks:

https://github.com/keplerproject/luarocks/blob/master/src/luarocks/type_check.lua

Finally, these things live and die by the quality of their C
implementations. In particular, it should be as simple as possible for
a C program to read LTIN, but no simpler.

steve d.