Hi,   
      
   On Monday, August 21, 2017 at 9:48:00 AM UTC-5, Robert Prins wrote:   
   > On 2017-08-20 20:40, rugxulo@nospicedham.gmail.com wrote:   
   > >   
   > > On Saturday, August 19, 2017 at 12:30:24 PM UTC-5, Robert Prins wrote:   
   > >>   
   > >> I don't think anyone here would be too pleased with 400+ lines   
   > >> of Pascal   
   > >   
   > > Maybe not, but you can post it to a Pascal newsgroup (or PasteBin or   
   > > whatever).   
   >   
   > - comp.lang.pascal.borland is dead for all intents and purposes   
   > - comp.lang.pascal.iso-ansi is dead for all intents and purposes   
   > - comp.lang.pascal.misc has turned into the personal blog of one   
   >   Amine Moulay Ramdane   
      
   Yes, but focusing on the Pascal version probably doesn't belong here   
   in CLAX. So we should migrate to news://comp.lang.pascal.borland   
   instead.   
      
   > The full source can be found on my Google Drive in lift32bit.rar   
      
   I only see liftdat.rar (and other bagatela), thus can't find *.PAS .   
      
   > >> PS: What I would love to see is what a really optimizing C   
   > >> compiler would make of the code when translated from Pascal to C!   
   > >   
   > > Depends on the Pascal dialect. Clearly you mean GCC or Clang,   
   > > but I guess FPC and GPC aren't working for you? Have you tried   
   > > them? Those are probably your best bet.   
   >   
   > The source files contains both Pure Pascal and assemblerised sections,   
   > and, with one small tweak, FPC actually compiles the Pure Pascal version.   
      
   Again, I can't see the *.PAS source at all. There is no obvious download   
   link on your homepage either.   
      
   > However, FPC has decided to throw a few conventions that stood   
   > since TP1 overboard, which means that the code no longer runs.   
      
   I find that unlikely. FPC is highly compatible with TP.   
      
   > Also, even when compiled -O3, the resulting assembler output   
   > shows code that is hardly better than the code produced by VP   
   > (or even BP).   
      
   That's impossible. Just avoiding ENTER/LEAVE (which VP favors)   
   helps a lot. You need -Cp and -Op (and maybe -Si and -Mtp and ...).   
      
   > And the thing that really has killed using FPC for me is the   
   > miserable IDE, the developers seem to be hell-bent on adding   
   > more features, units, and targets, but little time on the IDE   
   > and the code generation.   
      
   The IDE is older code, but overall the codegen is very good.   
      
   > GPC is dead, as can be seen on the website, which hasn't   
   > changed since 2005.   
      
   That's misleading. The website isn't the same as the actual code.   
   The frontend was last officially updated (AFAIK) in late 2007.   
   The various GCC backends changed too much for them to worry   
   with keeping up, but there were working versions for at least   
   GCC 4.2 and 4.3 (circa 2011).   
      
   But here I'm actually referring to old static builds (DOS/DJGPP)   
   that still work fine. And by "old" I mean GCC 3.4.6 atop DJGPP   
   2.05 with latest COFF BinUtils. It's not miniscule output size,   
   but even older GCC is still good with codegen.   
      
   > > Or try this (ISO 7185) with modern GCC:   
   > >   
   > > https://sourceforge.net/projects/pascal-p5c/files/pascal-p5c   
   code-35.zip/download   
   >   
   > Borland killed off iso-pascal with Turbo Pascal. Reverting to a theoretical   
   > standard is not really an option.   
      
   Here I'm talking more about actual working code translation for   
   "modern" GCC, with its multitude of optimizations, than anything   
   else. I wasn't really trying to force the classic dialect on you   
   (although you should be a little sympathetic, IMHO, due to its   
   provenance).   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)