Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-06-03

| Channels | #marpa index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:59 jeffreykegler joined #marpa
01:52 idiosyncrat joined #marpa
04:35 idiosyncrat joined #marpa
04:36 idiosyncrat rns (and everybody else): You'll recall the SLIF allows you to "hide" symbols from the semantics using parentheses
04:37 idiosyncrat I am thinking of dropping "hiding" in Kollos
04:37 idiosyncrat The reason it was helpful in the SLIF, is that using a Perl callback to do semantics was costly -- "hiding" allowed lots of tricks
04:38 idiosyncrat For example, by combining "hiding" and the "first" built-in, you could, in effect, select any RHS symbol as the value of the rule --
04:38 idiosyncrat just hide all the symbols before it.
04:39 idiosyncrat With Kollos, however, callbacks will become cheap, so you can just write Lua code to do that.
04:40 idiosyncrat If I eliminate "hiding" it frees up parens for other uses -- this was a non-standard way to use parens anyway.
04:41 idiosyncrat Other ruminations -- Kollos will allow you to have, not just dedicated sequence rules like
04:41 idiosyncrat seq ::= item*
04:41 idiosyncrat but sequences as part of a more complex RHS, like
04:42 idiosyncrat complex ::= a b* (c d)? e
04:42 idiosyncrat In effect, "nesting" rules.
04:42 idiosyncrat I plan, however, to forbid the nesting of precedence rules -- that is, rules like
04:43 idiosyncrat exp :: exp '*' exp || exp '+' exp || number
04:43 idiosyncrat must be "top level"
04:43 idiosyncrat oops, typo, make that
04:44 idiosyncrat exp ::= exp '*' exp || exp '+' exp || number
04:44 idiosyncrat In other words, you cannot do
04:45 idiosyncrat over-complex ::= a b (exp '*' exp || exp '+' exp || number) c* d
06:05 ronsavage joined #marpa
06:06 ronsavage I don't mind () being re-purposed.
07:22 lwa joined #marpa
07:23 rns joined #marpa
07:34 rns idiosyncrat: re hiding -- in the current LUIF draft, square brackets are for hiding and round brackets are for grouping, so dropping hiding (which I'm ok with as it simplifies syntax) boils down to dropping squares and leaving rounds as they are.
07:44 rns Per the current LUIF draft, grouping cannot cross | or || boundary -- so thing like over-complex ... are not allowed in current syntax.
15:53 jeffreykegler joined #marpa
16:36 jeffreykegler left #marpa
19:15 ernimril joined #marpa
22:41 ronsavage joined #marpa
22:46 jeffreykegler joined #marpa

| Channels | #marpa index | Today | | Search | Google Search | Plain-Text | summary