https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7645
--- Comment #13 from Bill Cole <[email protected]> --- (In reply to Robert R. Richter from comment #11) > I am no expert, so is it safe to just ignore these "Wide character in print > at..." warnings/errors? Or are there any other sideeffects so that I should > remove this ruleset? "Safe" is an imprecise concept, but I think ignoring those messages is safe for my understanding of safety. My understanding is that all of the rules are still being converted into compilable C and that only the specific rules that contain utf8 characters are being mangled in the process, making them generally non-matchable. See Henrik's comments above (comment #6 and comment #12) > FYI: I still have one 3.4.1 installation left and there are no such warnings > using this ruleset on 3.4.1. Seems to be an issue only on 3.4.2. That's probably because 3.4.1 was liberally sprinkled with "use bytes;" pragmas, which effectively removed handling of "wide" characters as characters rather than as a sequence of unrelated bytes. That wasn't a maintainable strategy given the modern reality of how Perl handles Unicode. If you want to understand the details, "perldoc bytes" is a place to start and it references additional documentation that may be helpful. Because this could be seen as a problem with a 3rd-party rule distribution that is distributing rules in a bad format, I am tempted to just close this as "INVALID" (i.e. not OUR problem,) but I do think we need to nail down the code truth in documentation and probably rework sa-compile for 4.0 to create re2c input files in a more tightly specified way. -- You are receiving this mail because: You are the assignee for the bug.
