From: Christian Kaiser <bchk@...>
Date: 2004-10-13T22:09:36+09:00
Subject: Oniguruma bugs?

Well I did replace the regex engine with Oniguruma, and found 2 bugs:

a) RegEx strings > 16 KB result in an "undefined bytecode (bug)" error. This
happen when I create a regex with

    NAMES = <<TEXT
        ... 3000 names ...
    TEXT
    s1 = (NAMES.to_a.map { |sWord| sWord.strip }).join('|')
    @reTitlePatterns = /(?i)\b(#{s1})\b/

    The standard RegEx engine has no problems.

b) the following code does not work as I expect it to work.

    VALIDNAMECHAR = '[A-Za-z\xa2\xc0-\xff]'
    puts /^(\w{2,}).* (\w{2,}?)$/.match("Hallo Welt") --> "Hallo"
    puts /^(\w{2,}).* (#{VALIDNAMECHAR}{2,}?)$/.match("Hallo Welt") --> nil

    Again, the standard RegEx engine has no problems. Which one is wrong?
Just replacing "\w" with [A-Za-z\xa2\xc0-\xff] should not cause any
problems...

Ruby 1.8.1 with regex engine replaced by Oniguruma. Operating system is
Win32.

Christian