[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Wikineurotic] Wiki page BPFK Section: Formal Grammar changed by Ilmen



The page BPFK Section: Formal Grammar was changed by Ilmen at 21:48 CET
Comment: Trying to fix a formatting issue

You can view the page by following this link:
 
http://www.lojban.org/tiki/BPFK%20Section%3A%20Formal%20Grammar

You can view a diff back to the previous version by following this link: 
http://www.lojban.org/tiki/tiki-pagehistory.php?page=BPFK%20Section%3A%20Formal%20Grammar&compare=1&oldver=28&newver=29


***********************************************************
The changes in this version follow below, followed after by the current full page text.
***********************************************************


+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
@@ -Lines: 13-17 changed to +Lines: 13-17 @@
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

!!Issues
- * Trailing empty tags within a bridi-head forethought termset aren't absorbed into the following selbri as selbri tcita in Jbofi'e's output, while [http://ilmen.tk/lojban/camxes.html|Camxes-pegjs] seems to parse it correctly. Concrete example: ''nu'i ge do gi mi pu klama'' parses as ''([{nu'i <([ge do gi mi] [__pu KU__]) NU'U>} CU {klama VAU}])'' in Jbofi'e, whereas it parses as ''([{nu'i ge do NUhU gi mi NUhU} CU] [__pu__ klama] VAU)'' in Camxes-pegjs.
+ * Trailing empty tags within a bridi-head forethought termset aren't absorbed into the following selbri as selbri tcita in Jbofi'e's output, while [http://ilmen.tk/lojban/camxes.html|Camxes-pegjs] seems to parse it correctly. Concrete example: ''nu'i ge do gi mi pu klama'' parses as ''(![{nu'i <(![ge do gi mi] ![__pu KU__]) NU'U>} CU {klama VAU}])'' in Jbofi'e, whereas it parses as ''(![{nu'i ge do NUhU gi mi NUhU} CU] ![__pu__ klama] VAU)'' in Camxes-pegjs.
* CLL 9.9, example 9.8: ''mi bai ke ge klama le zarci gi cadzu le bisli''. Camxes' PEG sees this like ''mi bai ku ke ge klama le zarci gi cadzu le bisli''. There doesn't appear to be a way to fill the tag in the gek-sentence production. Jbofihe and the official parser both get this right. Also relevant is example 18.15 from CLL 14.18: ''mi pu ge klama le zarci gi tervecnu lo cidja''. Jbofihe and the official parser both fail on this. Camxes sees it as ''mi pu ku ge klama le zarci gi tervecnu lo cidja''. Is this what is intended?
** Good catch. I think what camxes does is the Right Thing™, and that the unreachable tag should just be eliminated from the grammar. mi'e ((xorxes))




***********************************************************
The new page content follows below.
***********************************************************

For non-BPFK discussion of the formal grammar, see ((Grammar)).



This section describes a proposed update to the Formal Grammar, defining the grammar in ((PEG)) rather than ((YACC)).  This section requires cleanup and a shepherd.



!!Proposed grammar changes



*((Internal grammar of tags))

*((Move NAI to CAI))

*((Allow free modifiers anywhere))

*((Make sei more permissive))

*Take away the official status of the ((YACC)) grammar, and instead make an official grammar in ((PEG))

*[http://groups.google.com/group/lojban/browse_thread/thread/f1ddd5ff0eff69ad|Connectives + PU issues]



!!Issues

* Trailing empty tags within a bridi-head forethought termset aren't absorbed into the following selbri as selbri tcita in Jbofi'e's output, while [http://ilmen.tk/lojban/camxes.html|Camxes-pegjs] seems to parse it correctly. Concrete example: ''nu'i ge do gi mi pu klama'' parses as ''(![{nu'i <(![ge do gi mi] ![__pu KU__]) NU'U>} CU {klama VAU}])'' in Jbofi'e, whereas it parses as ''(![{nu'i ge do NUhU gi mi NUhU} CU] ![__pu__ klama] VAU)'' in Camxes-pegjs.

* CLL 9.9, example 9.8: ''mi bai ke ge klama le zarci gi cadzu le bisli''. Camxes' PEG sees this like ''mi bai ku ke ge klama le zarci gi cadzu le bisli''. There doesn't appear to be a way to fill the tag in the gek-sentence production. Jbofihe and the official parser both get this right. Also relevant is example 18.15 from CLL 14.18: ''mi pu ge klama le zarci gi tervecnu lo cidja''. Jbofihe and the official parser both fail on this. Camxes sees it as ''mi pu ku ge klama le zarci gi tervecnu lo cidja''. Is this what is intended?

** Good catch. I think what camxes does is the Right Thing™, and that the unreachable tag should just be eliminated from the grammar. mi'e ((xorxes))

*[https://groups.google.com/group/lojban/tree/browse_frm/month/2007-01/185f36ddbf029663?hl=en&rnum=31&_done=%2Fgroup%2Flojban%2Fbrowse_frm%2Fmonth%2F2007-01%3Fhl%3Den%26#doc_6828412041702b6a|Regular suffixing rules and regular rules for creating lujvo.]

* [https://groups.google.com/group/lojban/tree/browse_frm/month/2008-11/9fe8b6e82b3aa8b5?rnum=151&_done=%2Fgroup%2Flojban%2Fbrowse_frm%2Fmonth%2F2008-11%3F#doc_59b0c460473b4130|more formal grammar stuffs]

* [https://groups.google.com/group/lojban/tree/browse_frm/month/2008-11/9b23e1824698f78e?rnum=171&_done=%2Fgroup%2Flojban%2Fbrowse_frm%2Fmonth%2F2008-11%3F#doc_c6b6df439debb5ca|more formal issues (re: tosmabru check)]

* [http://groups.google.com/group/lojban/tree/browse_frm/month/2008-05/782cb9b82e7524f2?rnum=21&_done=%2Fgroup%2Flojban%2Fbrowse_frm%2Fmonth%2F2008-05%3F#doc_3a097aaca47211a1|Clarity of assertion is necessary.]

* [http://groups.google.com/group/lojban/tree/browse_frm/month/2008-06/14b4b356082d11a3?rnum=11&_done=%2Fgroup%2Flojban%2Fbrowse_frm%2Fmonth%2F2008-06%3F#doc_723ca0b890e3d04f|Interaction of SE, NA, and JA.]

* [http://groups.google.com/group/lojban/browse_thread/thread/ad25c54c578c03fa|a proposal about cmevla] and [http://groups.google.com/group/lojban/browse_thread/thread/bc05cffacd38fc45|selma'o LA].

* [http://groups.google.com/group/lojban/browse_thread/thread/04b2157a4a2f4ca2|A very old discussion about] [http://groups.google.com/group/lojban/browse_thread/thread/bf1e9c176c296206|default][https://groups.google.com/group/lojban/tree/browse_frm/month/2005-01/dcfc0e664ca0518b?rnum=141&_done=%2Fgroup%2Flojban%2Fbrowse_frm%2Fmonth%2F2005-01%3F#doc_dcfc0e664ca0518b|quantifiers][https://groups.google.com/group/lojban/tree/browse_frm/month/2005-04/221ca085617b5096?rnum=31&_done=%2Fgroup%2Flojban%2Fbrowse_frm%2Fmonth%2F2005-04%3F#doc_37f5e7fe861f3f5e|for gadri].

* [http://groups.google.com/group/lojban/browse_thread/thread/880bad057758b750|Blah about sumti raising (you can probably ignore this).]

* [http://groups.google.com/group/lojban/browse_thread/thread/4e2858a625897f45|A cu-like cmavo for terminators in general.]

* [http://groups.google.com/group/lojban/browse_thread/thread/a1a7eb7236084fd4|Why x1 to the left?]

* [http://groups.google.com/group/lojban/browse_frm/thread/8465f7b219e442fb|can we drop lerfu+namcu strings]?  At least, can we do that if we drop lots of mekso crap too?  See that thread and the relevant bits of ((zasni gerna cenba vreji)).

* The grammar has a difference between “operand” and “number” that probably isn't intentional; because of this, constructs like “mo'e zo'e” cannot be used where other numbers are allowed.


_______________________________________________
Wikineurotic mailing list
Wikineurotic@lojban.org
http://mail.lojban.org/mailman/listinfo/wikineurotic