Joseph Langley

If you encode your Chinese characters in UTF-8 most features will just work. Character classes (i.e. [aAcC-_] will not work, though. They will treat each byte of the code point...

Flex regex doesn't support Unicode escape sequences. You can construct rules or patterns for code points a byte at a time using hex escape sequences. The range character class constructor...

You have defined "variable length" rules when you use constructions like `[[:space:]]+` or `[[:space:]]{1,4096}`. Variable length rules use some of the same machinery as REJECT. When we hit this corner...

You're right, I left out a path component: examples/manual/string2.lex.

Answering your broader question about safe lengths. You're safe using variable length repetitions if you can reasonably expect them to match text that's about the length of your other tokens....

I'd think of the problem differently. Let Flex scan the words `END` and `PROGRAM` as tokens. Ignore the whitespace between them for a moment. Use your parser to match the...

Flex supports systems that run everywhere, including on tiny hardware. Minimal resource requirements is a design goal. You can edit your generated scanner however you like, of course. On Wed,...

You can enlarge your buffers. Look at the FAQ section in the manual for the question, "Is backing up a big deal?" There's information about the changes you'll need to...

That sounds familiar but it will take a minute to trace it and be sure. ASAN evaluates crash logs, right? So this is a real crash rather than a linter...

This wasn't what I thought. It was more interesting. I think everything is fine. Here's what I see happening and why I think ASAN is mad. `yy_current_state` is a pointer...