Is the first problem still relevant with the new JSON parser?

[arthsmn]

After Emacs 30 introduced a new JSON parser I was wondering if the first problem is still relevant, and how faster the parsing might be.

[Ergus]

The new parser is actually very fast and it does not rely in external libraries. So, a new benchmark set could be interesting now in Emacs 30.

BTW: I don't totally understand the current benchmarks numbers I see in the Actions. Some of them seems actually confusing. How can I see the speedup difference with and without booster?

[jdtsmith]

The benchmarks provided are for the artificial case of reading from JSON vs. reading from the equivalent (translated) bytecode. That's related to but not identical to the actual speedup you might get.

You can actually run the booster without doing JSON conversion; see --disable-bytecode. If you use eglot-booster it has the option --eglot-booster-io-only to enable this. If you do so maybe report your findings.