Re: Java virtual machine as target language for C/C++

khays@sequent.com (Kirk Hays) wrote:
>Remember, there is no problem in CS that can't be solved by another
>level of indirection.


True enough. Pascal for the CDC 170 did this too; the 170 has NO
stack and no native recursion (fine for Fortran, bad for Pascal). I
suspected that Pascal implementation (I can't remember whose it was,
but I don't think it was CDC's) built a jump stack.


If I remember my COMPASS (CDC assembly), there wasn't indirect
addressing so that programs that needed it actually built a jump
instruction with the desired address and branched to that instruction.
The 170 was an old archaic processor (no stack, but floating point!)
but I loved it! (Within its limits, the orthogonality was quite good
and code generation was relatively easy.)


Background: I'd used the Cyber in college and wrote a Pascal
compiler--unheard-of for undergrads there--for my senior project. I'd
planned on using the scheme I mentioned but I only got 70% finished by
the end of the term. :(


[That was in 1987, and I wish I'd had Usenet or even email at the
time...]


Dave
| David Moisan, N1KGH dmoisan@shore.net |
| 86 Essex #204 n1kgh@amsat.org |
| Salem. MA 01970-5225 http://www.shore.net/~dmoisan |
--

Return to the comp.compilers page.
Search the comp.compilers archives again.