Envelope-to: lojban-list-archive@lojban.org
Delivery-date: Sun, 24 Jan 2021 13:35:39 -0800
Sender: lojban@googlegroups.com
Received-SPF: pass (google.com: domain of scope845hlang343jbo@icebubble.org designates 2607:f2f8:a1d8::b19:0:f0b as permitted sender) client-ip=2607:f2f8:a1d8::b19:0:f0b;
From: scope845hlang343jbo@icebubble.org
To: lojban@googlegroups.com
Subject: [lojban] questions about camxes PEG grammar
Date: Sun, 24 Jan 2021 20:36:01 +0000
Message-ID: <86h7n66pji.fsf@cmarib.ramside>
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Reply-To: lojban@googlegroups.com
Precedence: list
Mailing-list: list lojban@googlegroups.com; contact lojban+owners@googlegroups.com
X-Spam_score: -2.8
X-Spam_score_int: -27
X-Spam_bar: --

doi la camxes .a la masatos.xagiuaras .a la gleki .a la ilmen .e zo'e

I'm looking at the PEG grammar for Lojban in the repository at
https://github.com/lojban/ilmentufa and have some questions...


le preti xi pa pi'e pa zo'u:

  {za'a} The build system is written in Javascript, and doesn't use a
  traditional build system, i.e. "make".  {.uanai} Why?

  {.e'u} I think the system would be much clearer if it used standard
  tools like make, sed, awk, and/or diff/patch.  Because PEG files are
  line-oriented, awk would probably be preferable to sed.  Most systems
  that have git (the format of the repository) will already have these
  tools.  Providing the alternative grammars as diffs would probably be
  ideal, because diffs are easily read and understood by human beings,
  and can be applied manually, without even needing any extra software
  at all.  (Any text editor would suffice.)

  If I actually did use the JS script to build the alternative grammars,
  the next thing I'd do would be to run a diff on them, anyway, to see
  what changes are made in each of the grammars.  So, using diff & patch
  would seem to be most sensible.


le preti xi pa pi'e re zo'u:

  {za'a} The README.md is rather confusing and written in broken
  English.  {ru'a} Without intending any offense, it appears as if
  English is not the author's native language.

  {.e'u} I would be happy to discuss the README, off-list, to figure out
  what the author means, and help re-work the README to make it clearer
  and more understandable.


le preti xi pa pi'e ci zo'u:

  {za'a} The .peg files contain a brief description of the syntax of
  parsing expressions:

    ...
    # 3)  Concatenation is expressed by juxtaposition with no operator symbol.
    # 4)  / represents *ORDERED* alternation (choice).  If the first
    #     option succeeds, the others will never be checked.
    # 5)  ? indicates that the element to the left is optional.
    # 6)  * represents optional repetition of the construct to the left.
    # 7)  + represents one_or_more repetition of the construct to the left.
    ...

  {.uanai} However, it does not specify the order of operations or
  relative precedence of each of the operators.  For example, does
  "a <- b c / d" mean "a <- (b c) / d" or "a <- b (c / d)"?  And
  does "x <- y / z+" mean  "x <- (y / z)+" or "x <- y / (z+)"?

  {ja'o} One can intuit the intended order of operations by reading the
  file, with a knowledge of Lojban grammar, but {.e'u} more specificity
  would be helpful.


le preti xi re pi'e pa zo'u:

  {za'a} I noticed that you handle elidible terminators differently in
  the PEG than in the EBNF.  For example, in the EBNF, one of the
  alternatives for sumti-6_97 is:

    LI # mex /LOhO#/

  But in the PEG, it's:

    li_clause <- LI_clause free* mex LOhO_elidible free*

  {ja'o} In the EBNF, the free* are only allowed if the terminator is
  present.  But, in the PEG, it appears that free* can follow an elided
  terminator.

  {.uanai} What's the rationale for doing this?  How do you determine to
  which construct the free modifiers apply when the terminator is
  omitted?


le preti xi re pi'e re zo'u:

  {za'a} There is at least one place in the PEG where the PEG grammar
  differs from the EBNF (and the YACC).  For example:

    #stag = simple_tense_modal ((jek / joik) simple_tense_modal)*
    stag <- simple_tense_modal ((jek / joik) simple_tense_modal)* / tense_modal (joik_jek tense_modal)*

  {ja'o} The PEG rule which would comply with the EBNF grammar is
  commented-out, and replaced with a rule which allows full "tag"s
  (i.e., full {fi'o} clauses) anywhere that only simple tags ("stag"s)
  used to be allowed.  This changes (by expanding it) the set of
  utterances that would represent grammatical Lojban text.

  {.uanai} What's the rationale for doing this?

  {.e'u} If the intent is to implement an extension to Lojban's grammar,
  shouldn't this be split-out into a separate "experimental" PEG, like
  the others?


le preti xi re pi'e ci zo'u:

  {za'a} The PEG grammar, in several places, uses constructs like:

    pehe_sa <- PEhE_clause (!PEhE_clause (sa_word / SA_clause !PEhE_clause))* SA_clause

    cehe_sa <- CEhE_clause (!CEhE_clause (sa_word / SA_clause !CEhE_clause))* SA_clause

  This is an idiom which appears repeatedly in the PEG, but there is no
  explanation for what this is doing or why.

  {.e'u} Some higher-level explanation of how erasure words are handled
  would be helpful.


le preti xi re pi'e vo zo'u:

  {za'a} The PEG contains many non-terminals of the form
  "<someword>_clause", "<someword>_pre", "<someword>_post",
  "<someword>_sa", "pre_clause", and "post_clause", but there is no
  explanation of what this is doing or why.

  {.e'u} Some higher-level explanation of the conventions used for
  naming the non-terminals, and how they interact, would be helpful.

.i ki'esai fa'o

-- 
You received this message because you are subscribed to the Google Groups "lojban" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lojban+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lojban/86h7n66pji.fsf%40cmarib.ramside.