Never understood this backwards way of thinking. You first write/immagine the grammar that you want to write, whatever is most intuitive for you.
Then you figure out what kind of parser your grammar needs.
Why would one contort a grammar? I mean, even "contorting it to not be context-dependent" is already a limitation enough...
Left recursion in things like expressions is completely trivial to eliminate. If you want to e.g. embed a DSL in a language, something that's really easy to do with recursive descent, it's worth spending 30 seconds or so to remove left recursion from your expression rules. Etc.
> You first write/immagine the grammar that you want to write, whatever is most intuitive for you.
IMO this is the road to hard to parse languages; if you genuinely advocate this path, you'll end up needing a GLR parser before you know it. I think it's much better to ensure your grammar is parseable with LL(1); not only is the tooling simpler, but the language will be less ambiguous for human users too. Follow the example of Pascal, not C++. This is what I'd advocate even for people doing hand-written recursive descent on a new language: validate your grammar with an LL(k) tool even if you don't use the generated parser.
(You've ended up sounding a bit like Ira Baxter, who's often seen around the net selling his DMS toolkit. He has a strong self-interest in people writing complicated grammars, so that he can sell them his services...)
Haha, I was thinking about him too. He does have a point, though, that on 2018 machines we might be able to afford parsing technology a bit more expensive than the 1970s approaches commonly used today.
I'd bet against this. The human brain works very differently, we always keeps something like "trees of contexts" in our head. And why would you thing left recursions is less intuitive to a human user than other kinds of recursion.
> ended up sounding a bit like Ira Baxter
Now I gotta take time and google who this guy is :P
For this, you want (as much as possible) for local edits to result in local modifications to the syntax tree, and ensuring that can require thinking carefully both about lexical syntax and abstract syntax. At the lexical level, if the literal syntax forbids newlines in string literals, then the changes to the token stream on a single-character edit are limited to changes to the tokens on the same line. Likewise, in the abstract syntax, if top-level declarations have a keyword, then the effect of a single-token change (eg, adding a mismatched paren) are limited to the declaration it occurs in.
There are a lot of tricks like this (eg, setting things up so you can find and parse the type declarations without having to parse a whole expression), but it's reasonable not to want to learn them. In that case, a really good heuristic is to make your language LL(1) parseable. It's not perfect, but it will rule out the vast majority of things that make incrementalization hard.