Yes, that really sucked but that wasn't until Solaris.
Hey Larry/Rob is that true? I thought the unbundling happened @ Sun when they created a compiler group and finally wrote their own production quality compilers like Masscomp, Apollo, and DEC had at the time. They really needed their own by the time Sun switched from 68K to SPARC. The original SunOS 68K C compiler was based on the MIT RTS compiler that I think Jack Test did the original back end for the Johnson compiler (and Tom Teixeira wrote the assembler). I know it was a couple of years before Sun invested in their own compiler technology. But Sun, (like Masscomp and Apollo), had a lot of ex-DEC folks in Marketing/Sales. DEC had always looked at the compiler has a revenue source [which I have always said is why C beat out BLISS -- C came with UNIX, BLISS cost $5K/cpu for VMS].
I do agree with Lyndon Nerenberg's comment, that Sun start to charge for the C compiler was the biggest help/legitimization that gcc ever got.
At Masscomp we had our had own compiler by the second year (which was the primary reason why we won all the performance tests. Our marking weenies wanted to charge for it also. Tom and I were the primary ones that fought it and said, with UNIX you get a C compiler (plus we needed to compile conf.c and a few other things at the customer site for a custom kernel). What sales wanted to s was supply the RTS compiler for free and then charge for the better compiler. We won that skirmish, although I actually think it might have been that the Roger Gourd realized that compiler folk would have had to continue to support the old C compiler, too.
So in the end, Fortran and Pascal were unbundled, although I think most (>90%) of the customer base did buy the Fortran system and probably about ⅓ bought Pascal too.
Clem