🚀 go-pugleaf

I've spent several days trying to build libg++-2.2 on a Convex using
gcc-2.2.2.  I've had a number of compile problems, mostly to do with
clashes between system header files and gnu header files, and fixed
things by hand in an ad hoc way until it compiles.  But now it flunks
the tFile-regular test by very wrong I/O of floating point types.
dtoa.C was one of the files I had trouble with in the compile.

If it matters I'll make a proper bug report; but first I wonder if
I've been barking up a well-known wrong tree.  For example, should
I be using glibc?  Is the Convex hopeless?  Is the Convex terra
incongnita for libg++ folks?

To answer a few anticipated questions, yes gcc passed the stage2
vs stage3 compare, yes I ran fixincludes, yes I am using the latest
gnu-make, no I'm not using glibc, gnu-ld, or gas.

For the next month (till I get back home to scripps.edu) I can't read
UseNet, so please reply by email to bashford@mpi-dortmund.mpg.de and
I'll summarize to this list.

Thanks,
Don Bashford