reimplemented litmoon

[nonchip]

there you go. using lpeg this time, because what the hell did I hack there before :P

tried to break as little as possible, and went with a factory approach where I always default to litmoon=false and generate the default functions in case someone was depending on them somewhere else.

general idea:

--declaration:
- some_func=(...) -> blah
+ some_func_factory(litmoon=false) -> (...) -> blah
+ some_func = some_func_factory!

--call:
- some_func ...           -- this would still work elsewhere
+ some_func(litmoon) ...

also the mechanism for filtering is actually pretty simple: extend moonscript.parser.literals.EmptyLine on demand (using above factory design) and redefine Line as (CheckIndent * Statement + Space * #Stop) + EmptyLine

[nonchip]

whoops, thanks travis, missed something there, i'll fix

[nonchip]

done. Line must have EmptyLine at the end, not at the start.

[nonchip]

@leafo ping?

[leafo]

Haven't had a chance to review yet, this is a much bigger change.

[nonchip]

will this be reviewed anytime soon? most of the changes are autocompiled lua anyway, busted passes (ofc i added a test for litmoon), and there's at least 2 projects already using it, so imho it'd be worth the time…

[nonchip]

seriously, i don't see how this "bigger change" (consisting of some few adjustments in about 5 files and running make) can stall for half a year, given the fact you tweeted me about "looking into it real soon" last year…

[leafo]

I don't think adding another argument specific to litmoon is a good change to this interface. (If we add more formats are we supposed to keep adding new arguments?) Any reason you didn't re-purpose options?

[nonchip]

you're right, i'll try to reimplement this using the options tables instead.

did kinda mistake them for being strictly structured stuff i'd better not touch…