On 7/11/2022 5:02 AM, Liam Proven wrote:
There are at least half a dozen generations. Given the ones that have been adopted outside Wirth's institutions and used in many countries, there are things that we could call Pascal, Pascal 3 (Modula-2), Pascal 4 (Oberon), and several different successors to Pascal 4. There are also non-Wirth variants that had some adoption, including Modula-2+, Modula-3,Turbo Pascal, Object Pascal, Delphi, Kylix and FreePascal. All legit, all sold and were widely-used at some points in time.
Well, real life fact is that Wirth, not once, was involved in developing a single, commercially available compiler, for any of his language "inventions"/developments. All of Wirth's (or his immediate minions) implementations were just sample implementations of his ideas how to promote and teach structured programming. That is in contrast with K&R for example, who's C compiler in fact was used to develop real world applications (like that thing called Unix :-) )...
In the real world, things sometimes take their own ways, and that was certainly the case for any Pascal (or Wirth's ideas) related. The University of California at San Diego (UCSD) picked up Wirth's rather easily (simple pass) to implement Pascal compiler and combined that with the p-System, a virtual machine that had already been developed at UCSD to create what became know as UCSD Pascal. p-System, or p-Machine, stood originally for pseudo-System/Machine, as it implemented a virtual machine with a pseudo 16bit CPU that could be implemented easily cross-platform. Some people started to call it Pascal-System/Machine, kind of ignoring the fact that from early on, there also was a quite widespread FORTRAN IV compiler, and elusive BASIC compiler (beside the mentioning in some UCSD p-System documentation, I have seen it only once, on a computer fair in the early/mid'80s, and then never again) and apparently an early Ada implementation. UCSD Pascal was then licensed through UCSD to several OEMs, with the most common implementation being Apple Pascal (for the Apple II and Apple III) and HP Pascal, and then IBM, for their new IBM PC. At the same time, several other compiler companies picked up on the increasing popularity of Pascal and offered their own implementations, though probably due to the licensing terms of UCSD, which gave pretty much an exclusive license to SoftTech (later sold to Cabot, UK, then Pecan, UK) for cheap and SoftTech then wanted to charge an arm, half a leg and your first born to license the system then from them, were doing all native code generation. One of those was Anders Hejlsberg and his company PolyData, which produced Z80 native code, which then was bought by Borland to become Turbo Pascal 1.0. Digital Research did their Pascal MT+, also native Z80 and link compatible with their C compiler (but not CBASIC). DEC did their own thing for their VAX/VMS PAscal (late DEC Pascal, then HP Pascal, which had no relation to the earlier UCSD based HP Pascal), and their are persistent rumors that the PDP-11 and VAX CPU instruction sets were "pilfered" from the p-Code from UCSD, but I never used the assembler on any of those, so I don't know for sure ;-) Microsoft has early on their own Pascal, which produced code for their own p-Code implementation, the same one they used in their early COBOL and FORTRAN compilers, as well as in some applications like Multiplan, the predecessor to Excel. They bought a 3rd party Pascal compiler and sold it as QuickPascal for a (short) while in order to try and compete with Borland, but gave up on that one pretty quick, just like with their QuickC attempt, which all got cancelled along with their "full" Pascal and COBOL compiler, though Fortran lived a Fortran Powerstation for a little bit longer. And there were a few more compiler companies that tried (Prospero, Metaware, Metrowerks) but all of them rather failed as they either try to be "standards compliant" or were running behind the relative fast paced developments at Borland, beside not being able to compete on price.
GNU's GPC was just a crutch, a unwanted step-child, that didn't even get a fraction of the attention that it should have gotten early on. That's what resulted in Florian Klämpfl's early Open Source implementation of FPK (Florian's Pascal Kompiler), which started as a (pretty well made) i386 generating Turbo Pascal compatible compiler, before following the path set by Borland/Inprise/Codegear/Embarcadero with their Delphi implementation of Object Pascal. Unfortunately, there seems to be very little interest in doing a backport of FreePascal into a 16bit , Turbo Pascal compatible compiler for DOS. The 8086 target version of FPC is still a cross-compiler with demanding resources which prevent it from running on (Free)DOS itself. Which makes me wonder if it would be possible to do such a "back port" from the sources of one of the earlier versions of FPK,at least those that started to be self-compiling, before the more widespread adaptations of Delphi'isms.... :?
Ralf _______________________________________________ Freedos-user mailing list Freedos-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-user
