Modestas Valauskas

_for future reference_: steveroush pointed out in https://gitlab.com/graphviz/graphviz/-/issues/1656#note_1496402295 that a group attribute can be used instead of ranks to put nodes into a single row: ``` digraph { rankdir="LR" {node[group=Aa]...

(context: https://github.com/mdaines/grammophone/issues/33#issuecomment-1659161152) POC for adding custom edges to an exported SVG: https://github.com/mdaines/grammophone/issues/33#issuecomment-1660600675 is definitely doable.

I've been trying a bunch of things, and I have come to the conclusion that showing the DFA + all the constraints of the DeRemer & Pennello algorithm for constructing...

I'm working on a new algorithm for upgrading LALR(1) automata to LR(1) and the DeRemer `read` and `include` relations appear to be pretty fundamental to the act of building LR(1)...

I no longer agree with the statement that pursuing this would be worthwhile. While this would be nice to have, it would be much more useful to actually show ambiguities...

The proposed fix is incorrect.

I ended up using PNGs instead.

There's one huge issue that prevents a traditional EBNF grammar from being straightforwardly-useful with grammophone. The conversion between CFG and EBNF is not "lossless". There's some amount of ambiguity there...

> However, when considering this before, I wasn't sure how to present calculations with these additional rules. > > This would add S_ to the nonterminals table, parsing tables, etc....

My observation is that adding optional trailing commas almost always improves code readability, because it forces the formatter to put everything on a separate line, which implicitly enforces a "one...