On Mon, 11 Dec 2006, Jules Richardson wrote: TurboC 2.0x DOES explicitly support inline assembler. http://bdn.borland.com/article/images/20841/tc20ad.jpg (notice the list of features of TC2.0 in the box at lower left) If you are not using symbolic labels, remember that C passes arguments to a function on the stack (pushed from right to left), and returns the function return in AX. Restore the values of all segment registers, and BP.

Jerome Fine replies: I am having difficulty getting the following code assembled using the watcom assembler. I understand that TASM will do much better, but I have not been able to find a copy. Do you know if TASM might be able to assemble the following code and where I might obtain a copy? ======================================================================= .radix 8 ;????? .386 ;(allow constants over 16 bits) csr1= 17777100q ;base CSR address in PDP-11 space;????? csr2= 4194304d-64d ;17777100q ;base CSR address in PDP-11 space .186 ;*not* USE32 segments code segment use16 'code' assume cs:code .386 start: retf shtdwn: retf wep: public wep mov ax,seg dgroup retf data: mov ebx,ds:addr ;get pointer into memory ;????? add dword ptr ds:addr,2 ;increment address ;????? ret alowi: mov ax,word ptr ds:addr ;fetch value ;????? ret alowo: mov word ptr ds:addr,ax ;save value ;????? ret ahighi: mov ax,word ptr ds:addr+2 ;fetch value ;????? ret ahigho: mov word ptr ds:addr+2,ax ;save value ;????? ret .186 code ends dgroup group data ;????? data segment use16 'data' .386 db 16d dup(0) ;task header (LocalAlloc ptrs etc.) addr dd 0 ;extended address (forced even) ;????? .186 data ends end start ============================================================================ Sincerely yours, Jerome Fine

On Mon, 11 Dec 2006, Jerome H. Fine wrote: AMONG OTHER THINGS, TASM would probably not be happy about naming both a code offset AND a segment "data" Strip it down. Remove everything that does assemble, until you isolate specific items that won't assemble. You can put them back in afterwards, but right now, they get in the way of finding out what is going wrong.

--- Frank McConnell <fmc at reanimators.org> wrote: Is it a *recent* development of compilers that as an intermediate step the source code will first be reduced to assembler mnemonics, before being reduced to object code? __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

On 12/16/06, Dave Dunfield <dave06a at dunfield.com> wrote: For an old embedded product (c. 1986), we had to maintain a Perkin-Elmer workstation to compile the product C source to .ASM files that were compatible with our home-grown relocating linker. We could have replaced everything with more modern software, but it was just easier to keep the ancient workstation running for the odd code change. At the time we were still using it, it was more common for C compilers to not generate intermediate ASM files than it was years earlier, making the transition to a newer environment less than trivial. -ethan

Dave McGuire wrote: I can verify that many compilers on VSE have done that for a long time too. I don't know MVS and VM/CMS well enough to comment on either of those. Peace... Sridhar

Many, many years ago when I first started at IBM (1964) there was a cute cartoon in Datamation Magazine. It showed a big machine labeled decompiler. A worker was feeding cans of apple sauce into one end of the mation and another worker at the opposite end removing apples and putting them in a box