Hi Akim,
Thanks for the response, and for all your work on Bison!
Sorry, I discovered this recently too, see
https://lists.gnu.org/archive/html/bison-patches/2018-11/msg00042.html. I
don’t know whether that means I should release a 3.2.2. WDYT?
Looks like you've just got a 3.2.2 out, so I guess the fix is there?
Does that mean that you are a user of glr.cc? I don’t much feedback about it,
unfortunately. Also, I’m looking for the opinion about a proposal for GLR
parsers: see
https://lists.gnu.org/archive/html/bison-patches/2013-02/msg00105.html and
https://lists.gnu.org/archive/html/bison-patches/2018-11/msg00008.html and
following.
Yes, indeed, we use it for the parser of the Birch probabilistic
programming language (birch-lang.org). We started with the basic LALR(1)
parser, but as the language has developed, a GLR parser has become
necessary for our chosen syntax.
I'll have a better look at these proposals later, but yes, in general,
it would be nice to be able to specify precisely where the conflicts are
expected, rather than just the total number expected. When using %expect
declarations, I find myself writing explanations as to where these
conflicts are (see e.g. lines 84-91 here:
https://github.com/lawmurray/Birch/blob/model/bi/parser.ypp), and it
would be preferable for these to be formalised.
Cheers,
Lawrence
