I have a lot of admiration for cinap, he's "deep". But he is also the best qualified person to estimate whether improvements in 9front are portable back to legacy and I'm sure that is, sadly, not high on his agenda.
Conservatively, I'd like legacy to be the entry system to Plan 9 and categorise mutually incompatible enhancements (there are a few I know about, but can't think of any off-hand) to be well documented so "forks" remain as close to each other as possible. Of course, there is implicitly no incompatibility where bug fixes occur or new developments are introduced, in my "perfect world". Your comments, hiro, are very helpful and reassuring. My focus at present is to continue my Go developments (not contributions, I have some long-term work I would not undertake in any other language - Go is hardly perfect, but it is wonderfully productive: I'm no longer surprised when modules work first-time after a successful compilation), but I'll be glad to contribute to any efforts to categorise Plan 9 updates since the demise is Bell-Labs into compatibility classes. Taxonomy, rather than archeology, I guess. I think it would be worthwhile, even at a hobby level. I'll tick off your questions as I get an opportunity to test them, report back here, or personally, as seems appropriate. Thank you for taking the trouble to encourage me along. Lucio.