I admit I haven't been able to catch up with the flood of messages on the Hebrew list...
> On 15/08/2003 07:57, Paul Nelson (TYPOGRAPHY) wrote: > > >>This brings us back to the earlier quesion of whether it is > >>legitimate to use ZWJ or ZWNJ between combining marks > >> > >> > > > >It sure better be. This is done in Khmer for controlling register shift I noticed, somewhat to my surprise! > >combinations for exception words. I have seen nothing in Unicode that > >states that the ZWJ/ZWNJ can only be used with base type characters. > > > >Paul > > > > > > > > > Well, I understood that ZWJ/ZWNJ followed by a combining mark gives a > defective combining sequence, Yes, it does. I would suggest reclassifying ZWJ and ZWNJ as combining marks, i.e. Mn instead of Cf as "general category". They would have to keep their combining class of 0, in order not to change any normal forms. As I've mentioned, I think class 0, with its special treatment is problematic, and hence ZWJ and ZWNJ should not be "acknowledged" just anywhere in any combining sequence. There is no problem to have them at the end of a combining sequence, that is what has to all intents and purposes been the case all the time, even as Cf *just after* a combining sequence. "New" cases, that are not "acknowledged" with ZWJ/ZWNJ as Cf, but should be with ZWJ/ZWNJ as Mn: just after a base character (even when followed by more combining characters) [this takes care of the Khmer cases], between two combining marks in a combining sequence that would be in canonical order *if* the ZWJ/ZWNJs were removed from the sequence (this may take care of some Hebrew ligation cases, I'm not sure). Other placements of ZWJ/ZWNJ should not be "acknowledged", but should be deprecated. ZWJ/ZWNJ would remain "default ignorable". /kent k