search

moznion / Perl-Lint

Yet Another Perl Source Code Linter
https://metacpan.org/pod/Perl::Lint
Other
95 stars 12 forks source link

Just A Suggestion #23

Open thaljef opened 10 years ago

thaljef commented 10 years ago

I'm really excited to see you take this project on. A faster Perl-Critic would be very useful to the community. However, you might want to consider a different approach...

The Perl::Critic::Policy modules have evolved over the years to handle many edge cases that appear in real-world Perl code. Some of them have become quite complex, and reproducing all that code by writing equivalent modules for Perl-Lint is a very big task. The Perl-Critic engine also has several optimizations that are not trivial to implement.

So instead, you could build a layer over Compiler::Lexer and Compiler::Parser that produces a DOM identical to PPI. That way, Perl::Critic (and everything else that uses PPI) can get the speed benefits without having to be re-written for an entirely new DOM. And the unit tests for PPI would be a perfect test suite for your implementation.

I'm not really familiar with Compiler::Lexer and Compiler::Parser, and they may have completely different goals from PPI. But I know Adam Kennedy put a lot of thought into designing the DOM for PPI and it has proven to be good enough for many purposes.

If it is feasible at all, I think generating a faster PPI-compatible DOM would be easier and more productive than rewriting all of Perl-Critic for a new DOM.

moznion commented 10 years ago

Thank you for your suggestion.

That sounds nice idea, but I have the problem realistically; Compiler::Lexer is almost stable, but Compiler::Parser is not.

It is a thing that I know you have been carrying out this project, Compiler::Lexer can analyze the perl code practically (but it has some problems, so I sent issues). But Compiler::Parser can not parse most of the perl code and it occurs SEGV frequently.

So I think it is difficult to construct the PDOM (or compatible structure with it) by Compiler::Lexer and Compiler::Parser.

However, I think if this idea can realize, it is so nice. Therefore I think it is a good effort to implement the compatible constructor. But it is big job in any case, so I think that it is better to do it in another project.

How do you feel?