Oh I had some really good times with ohm editor, when I was evaluating ohm, although I could not use ohm for the project but I do miss such editor/visualization for any parser toolkit.

What's the future of parsers now that we have LLMs?

Don't get me wrong, certainly parser generators produce fast parsers. And we know that their behaviour is deterministic. LLMs have neither of those; however they can provide intelligent feedback.

I suppose that questions like mine are lazy, because the range of possible futures is enormous.

As long as compilers exist, we will probably have parsers. And at present we need compilers in order to build the software that is used to run LLMs.

Here's how to visualize packrat parsing: imagine a picture of the OOM-killer crashing your machine.

Then learn how to use a real parser.

Can Ohm grammars be translated to Tree Sitter grammars? If so, this would be an extremely useful tool for developing the latter.

I suspect not, or at least not easily, since the underlying algorithm is very different (PEG vs GLL).