Internals & comparison to PEG

[dumblob]

I'm curious about the internals of nevod app. Could you shed some light on the internal workings (any bytecode? how the extraction virtual machine works? etc.)? Maybe in some blog post or here or on another public place?

After I read the language reference for Nevod, I got the impression it's quite close to PEGs (or at least it seems Nevod could be compiled to PEGs). But I'm probably wrong as it seems you have a hand-crafted parser of Nevod and I suppose the backend will also be your work.

Could you comment on any (potential) relation to PEGs and maybe make some performance comparison (perhaps with LPegs)? Maybe just rewrite the examples from your web page in LPeg and run it on the same machine with default settings to see how it performs. Or maybe some more rigorous comparison...

[dmitry-surkov]

While Nevod may seem close to PEG, it is targeted at natural language texts or combination of formal language with natural language (e.g. HTML). The performance requirement dictates usage of the programming languages to create parsers for the formal languages. C#, Java, TypeScript, JSON, XML, HTML, etc. - all have specially programmed and optimized parsers. However, we are considering usage of Nevod as a tool for syntax highlighting, where performance is not so important, but this might be a work on top of Nevod.

[dumblob]

Thanks Dmitry. Any benchmark results yet (as I said ideally compared to PEGs - those from Lua are very popular and provide a simple Lua interface so it shouldn't be hard to use them as a fair competitor to Nevod).