IRC channel logs

<ekaitz>hi, i have a PEG grammar that uses underscores in the nonterminals

<ekaitz>that's not compatible with guile's PEG parser

<ekaitz>is there any reason not to add _ in the nonterminals?

<ekaitz>also peg-string-compile is documented but not exported :)

<mwette>ekaitz: at some point `_' became syntax in Guile. It's not in 2.2. I don't know where it comes from.

<ekaitz>hmm

<dpk>that shouldn’t preclude rebinding it (though you might have to disable importing it)

<ekaitz>dpk: i don't understand what you mean

<dpk>‘_’ is just a normal identifier like any other

<dpk>it might be an error if you try to define it at the top-level of a library or program after having imported it from another library (such as (guile) or (rnrs (6)) or whatever)

<dpk>but you should be able to exclude it individually from the import set

<ekaitz>oh yes

<ekaitz>that's why i don't understand why the PEG parser does not allow _ in identifiers

<ekaitz>it doesn't even allow this_identifier

<ekaitz>for terminals

<dpk>oh, it’s a problem with the string syntax for PEGs in (ice-9 peg) ?

<ekaitz>yes

<dpk>ahhh

<ekaitz>it only allows [A-Za-z-]

<ekaitz>or whatever that is

<dpk>i thought you had a nonterminal called _ defined in the string syntax, and it was erroring because it was trying to redefine _

<ekaitz>i just added the _ to the regexp and it just works

<ekaitz>but i had to make a new module on my own

<ekaitz>does it make any sense to allow the _ in the peg parser strings directly? should I make a patch with this?

<ekaitz>i guess they should have a similar syntax to scheme symbols, as they are going to be defined in the current scope

<ekaitz>and symbols support _

<lilyp>I'm not sure why you'd want "_" to be allowed in the PEG parser – what's the rationale?

<lilyp>sym

<ekaitz>lilyp: just compatibility with some other parser tools

<lilyp>symbols also allow *%&

<lilyp>even space if you know how to type them ;)

<ekaitz>yes, i know

<ekaitz>we could add that too

<lilyp>my point was rather that PEG ids could be a subset of Scheme symbols

<ekaitz>yeah, of course

<ekaitz>what I meant is adding _ would not be a superset of the scheme symbol syntax

<ekaitz>but yes, they can stay simpler

<ekaitz>i asked if there was any reason for it

<ekaitz>does the peg string support escape sequences?

<ekaitz>i don't find anything like that

<lilyp>I don't think so – maybe the Scheme API has some

<ekaitz>ugh this is going to be a little bit harder than I expected

<lilyp>for the underline: I think funny characters shouldn't show up in the peg due to possible ambiguities

<lilyp>like a* vs a*

<ekaitz>and a_token <- ...

<ekaitz>?

<weary-traveler>peg as in parsing expression grammar?

IRC channel logs

2024-09-05.log