Basically the VAX is screwed in this regard. One could easily imagine 
forking a GCC (heck just start from old EGCS release) and making a "VAX 
only" cc. That would be pretty trivially maintainable by a single 
individual once it was complete because the VAX isn't changing, (nor is it 
likely to change :-).

The problem is that gcc isn't "C", it is base ANSI-C with a bunch of 
enhancements which you _could_ tell gcc to ignore and it will (for most of 
them) but then it can't compile NetBSD. Further, the NetBSD community is 
unwilling to sacrifice some of the non-standard features of gcc in order to 
make an ANSI-C OS that could be compiled on other platforms with different 
tools (after all gcc is available on everything, even PICs :-)

The intermediate solution that is emerging is the use of GCC on a PeeCee or 
other machine with gobs of CPU and Memory in a cross compilation 
environment and basically ignoring the problem.

If this emergent solution is adopted "for ever" (ie the core team decides 
from point X on all compiles of NetBSD/Vax will occur on some other 
platform) then we open the door to put a nice tight, and fast, ANSI c 
compiler in the release that *users* can use to develop with but won't ever 
be able to recompile the system. With a bit of discipline that same 
compiler might support recompiling the kernel which would be essential for 
system evolution without a cross platform.

Not a lot of people care about the VAX outside of this relatively small 
community, and while that may be distasteful it seems most certain. The 
other alternative might be something like NetVMS but given that HP is 
pressing ahead with that as an Itanic OS I can't see them releasing source 
access any time soon.

--Chuck