foobar wrote: > 1. common human markup such as: _foo_ (underline), *foo* (bold) etc,
Yeah, that's a pretty good idea. I agree with the others that it should keep the text symbols (especially since I've seen these algorithms wrongly flag things *a lot*) but a basic implementation is ok. > 2. parse BBCode. This probably isn't a good idea... unless it is a web input only filter. So posts pulled off the news server are treated as plain text - no BBCode parsing is attempted. But posts made through the website may be parsed, and converted to plain text before being forwarded to the news server. (Note that I use my beloved mutt mail client for reading the newsgroups myself, so anything that would break plain text email browsing is a no.) I already have pretty decent bbcode -> html and html -> text functions in my bag of toys, so regular participants never need to know what kind of input was used. It would let web users feel more at home without impacting everyone else. The only downside I see is if people think bbcode is accepted, someone might write it in their newsreader or email client, where it won't be parsed. I don't want the groups to get filled up with bizarre markup everywhere, but, the kind of users who use email clients and newsreaders probably won't make that mistake anyway. So yeah, let's give it a try for web posting and see if it works out.
