Watcom 1.9 is what supposedly everybody should use, so I called that "official". But for a seldomly used feature ("based pointers") it produces an incorrect binary (for code executing in Ring 0 -> device drivers) where Watcom 2.0 will work properly. But Watcom 2.0 is "Jiri's fork" where you never know what he will do next.
Is there a possibility at all to report a bug against Watcom 1.9 ?
Is this bug new to 1.9 or 1.8 or ...
There's still alive the #watcom chat group and Michal is hanging around there. I don't think the perforce server will ever get started again. There's also Stevens OW copy at github -
https://github.com/StevenLevine/openwatcom-v1. But it seems he don't allow issues there. Maybe we should email Frank? But I fear he lost interest on OW since a while.
Edit - Michal says this
https://github.com/iainnicol/open-watcom-1.x/tree/openwatcom 'should be a very accurate conversion of the P4 depot to git, but it's not meant as a working copy:'. Looking a bit in the commits it seems to me Stevens repository is much ahead of that. Although Michal says he will accept fixes I think an email to Steven would be the best chance to get something fixed in the near future.
Basically it's the same problem as always - if different people put different repositories online no one knows what's the best one to use. If there's no clear lead for a project it's useless in the long run even if there exist different source repos. Currently I don't know what I should think about the next fork of Martin here -
https://github.com/OS2World/DEV-TOOLS-IDE-openwatcom-v1. Of course I circumvent Jiris sites including his sourceforge.net one.