Andreas: Actually, it occurs to me that I can work up an example from Marpa's internal trace mechanisms, showing the internal rules. I'm away from my development environment now, but I'll do that in a bit.
On Fri, Mar 6, 2015 at 12:30 PM, Andreas Kupries <[email protected]> wrote: > On Fri, Mar 6, 2015 at 12:08 PM, Jeffrey Kegler > <[email protected]> wrote: >> No. Sorry if I was not clear. > > No problem. > >> >> The rewrite for precedenced rules removes the ambiguity of the pure >> BNF in the SLIF DSL, and in this way is very much like the one you see >> described elsewhere, particularly for dealing with expressions using >> top-down parsers. You add a new symbol for the expression at each >> level of precedence, and the rewritten BNF will be unambiguous. > > Ok. Guess I have to google then for bnf rewriting. > Thanks for your time. > >> >> On Fri, Mar 6, 2015 at 11:50 AM, Andreas Kupries >> <[email protected]> wrote: >>> On Fri, Mar 6, 2015 at 11:33 AM, Jeffrey Kegler >>> <[email protected]> wrote: >>>> For the precedenced rules, you create new symbols for each precedence, >>>> proceeding with associativity in the obvious way. Much of it is done >>>> in this routine: >>>> https://github.com/jeffreykegler/Marpa--R2/blob/master/cpan/lib/Marpa/R2/MetaAST.pm#L568 >>> >>> Thank you. >>> It reminds me that I have not read perl code since Perl4. >>> I believe it will take me a while to decipher that. >>> >>>> One thing to note is that it is *not* operator precedence -- the rule >>>> have precedence, and the operators play no special role. >>> >>> So, in essence the rewrite generates an ambiguous grammar which the >>> engine has no problems with, parsing them all in parallel, right ? >>> >>> And at the end the rule priorities are then used to collapse >>> ambiguity, by keeping only the highest priority at each node ?! >>> >>> And I further guess that the only ambiguities which survive to the >>> parse forest are those which have the same highest priority at a node. >>> >>>> It's not dissimilar to rewrites you'll see elsewhere, but I might be >>>> its inventor, if only because nobody previously had a parser that >>>> could be expected to parse the rewritten rules. >>>> >>>> On Fri, Mar 6, 2015 at 11:22 AM, Andreas Kupries >>>> <[email protected]> wrote: >>>>> >>>>> >>>>> On Thu, Mar 5, 2015 at 4:51 PM, Jeffrey Kegler >>>>> <[email protected]> wrote: >>>>>> >>>>>> Quantified rules are semantic sugar for BNF, and you can write BNF rules >>>>>> and have access to all the symbols in them, including those you think of >>>>>> as >>>>>> separators. I say "think of", because if separators are given a >>>>>> semantics, >>>>>> they are (conceptually) more than just separators. Sequences can be >>>>>> written >>>>>> as recursions -- internally, that is what Marpa::R2 actually does. >>>>> >>>>> >>>>> Is there any documentation around on how Marpa::R2 rewrites the sugar into >>>>> basic BNF ? >>>>> >>>>> Ok, I know how it can be done for the quantified rules, via left recursion >>>>> and additional non-terminals. However for the priority rules I am still >>>>> not >>>>> fully clear how these would be done. >>>>> >>>>> If there is no documentation, then is the rewrite engine in some >>>>> central/singular place in the sources, or is its activity distributed >>>>> through it, and where would I have to look ? >>> >>> >>> >>> -- >>> Andreas Kupries >>> Senior Tcl Developer >>> Code to Cloud: Smarter, Safer, Faster™ >>> F: 778.786.1133 >>> [email protected], http://www.activestate.com >>> Learn about Stackato for Private PaaS: http://www.activestate.com/stackato >>> >>> -- >>> You received this message because you are subscribed to the Google Groups >>> "marpa parser" group. >>> To unsubscribe from this group and stop receiving emails from it, send an >>> email to [email protected]. >>> For more options, visit https://groups.google.com/d/optout. >> >> -- >> You received this message because you are subscribed to the Google Groups >> "marpa parser" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> For more options, visit https://groups.google.com/d/optout. > > > > -- > Andreas Kupries > Senior Tcl Developer > Code to Cloud: Smarter, Safer, Faster™ > F: 778.786.1133 > [email protected], http://www.activestate.com > Learn about Stackato for Private PaaS: http://www.activestate.com/stackato > > -- > You received this message because you are subscribed to the Google Groups > "marpa parser" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "marpa parser" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
