RE2 VS pomsky

Looking at the benchmark where C++ is worst compared to other languages, it's depending on the library used. I would guess if they used Google's re2 Regex library instead of Boost's, the result would be different.

https://github.com/google/re2

https://github.com/greensoftwarelab/Energy-Languages/blob/ma...

That page says it just includes "some of the most common special characters", and following the link to the Examples page in turn includes a link to the full list.

Python uses backtracking, so this probably isn't O(n), especially with the ability to choose the dictionary.

But with there are non-backtracking matchers which would make this O(n). Here's re2 from https://github.com/google/re2 :

  >>> import re2

RE2 is a Google regular expression library

google's regex library: https://github.com/google/re2

I am currently working with Regex, specifically Re2, and was wondering if there is a real difference between the above expressions for repeated sub-regex.

I ported Google RE2, a regular expression library, to SPVM as Resource::Re2, and created SPVM::Regex, a wrapper for it.

I searched for I found that there is a Perl compatible regular expression called Google RE2. It is written in C++, and with Google RE2, I can use Perl-compatible regular expressions as a library.

Yes, but there is an interesting clarification here. RE2 has used the "caching" approach documented in the Ruby bug ticket linked for quite some time (since its birth?): https://github.com/google/re2/blob/954656f47fe8fb505d4818da1...

It is mentioned only briefly in Cox's article on regex matching in the wild. Look for the word "bitstate": https://swtch.com/~rsc/regexp/regexp3.html

I didn't know Perl had implemented this trick too.

The paper[1] cited in the Ruby bug ticket was published very recently. When I first read the Ruby bug ticket, I immediately wondered how they sidestepped the memory use problem. The paper's abstract seems to suggest there is some technique for doing so, as it rebuffs the idea of doing "full" memoization. Alas, I do not have access the paper. (Which is fucking ridiculous.)

[1]: https://ieeexplore.ieee.org/abstract/document/9519427

I started with JS, and my first step was to write a simple script that checks if a regex is valid. I can call this script from Rust, but there's a problem: Starting a nodejs process takes about 100ms, which is not acceptable, especially for fuzzing.

I noticed this when a test case of a parser I wrote failed in CI on Windows. I then learned that the default stack size on Windows is only 1 MiB whereas its 8 MiB on Linux if I remember correctly. The parser has a hard-coded recursion limit to prevent stack overflows, which is currently set to 128. However, this limit is lower than necessary on Linux, and might still be too high for some platforms (embedded?)

I was informed that Rulex is a registered trademark and I'm not allowed to use the name for my project. A lawyer contacted me and gave me a week to rename the project, so I have to come up with a different name :(

In the other discussion, there's also a link to Rulex, which has similar goals but is more concise. Also claims to compile to multiple regex dialects.

Something similar happened to me a week ago. Someone emailed me that they found panics in rulex, and then created a PR to fix them. They even explained to me how to create a security advisory on GitHub because the panics could be used to DoS someone. It was very helpful.

Sorry for not being more specific. Since this is a parser, the span points into the text file that is being parsed, so it's not that relevant for most libraries. Here is the parser, it uses nom parser combinators. Unfortunately, and adding the spans to the errors involves some boilerplate.

I welcome contributions for rulex. It's a medium-sized project that should be fairly easy to understand, and has some "good first issues" :)

Rulex – A new, portable, regular expression language\ (102 comments)