Hallo Gábor, > ... That code must be simple to be translated easily for the > target CPU with the maximal possible performance ... Quite some time ago I started to write my own Pascal compiler. The idea was that it should at least be capable of compiling its own sources. The output would be an assembler file but one only filled with macros. The assembler must turn these macros into a binary. My self programmed assembler is capable of handling various processors including the 6502 plus various successors, Z80, 6800 and 6809. A directive in the Pascal source tells the assembler for what machine it was meant. The assembler simply replaced the macros with the assembler code found in the macro file made for this machine. And yes, machine, not processor. The idea was to use the onboard ROM routines from that machine for certain tasks. If I have to output a character to the screen, why not using the routine that is already present in the ROM? It surely is the easiest way, if it is the fastest, that's another question. If you think the ROM routine is too slow, just replace the according lines in the macro file by new and better ones and recompile the ASM file. So far I am able to compile my own sources and I'm quite sure I covered all Pascal commands. Better: the ones of Turbo Pascal. Next I started to create the macros. And so far I haven't finished the one for the C64, it is a lot of work :( An interesting detail: Pascal pushes a lot of things on the Stack. The one of the 6502 is much to small. So I created my own Stack. IIRC two locations of the zero page points to this stack. If I have to push something on the Stack, I use "sta (zp),Y" (where Y=0) and then decrease the pointer. And all this is done inside the macro file. But the Z80 and 65816 can use all 64 KB as Stack. So the above trick isn't needed for these two processors and normal Stack operations can be used. I hope this was clear enough. If not, you know where to find me :) -- ___ / __|__ / / |_/ Groetjes, Ruud Baltissen \ \__|_\ \___| http://Ruud.C64.org Message was sent through the cbm-hackers mailing listReceived on 2011-10-26 19:00:22
Archive generated by hypermail 2.2.0.