🚀 go-pugleaf

Jack...
I can agree with you up to a point.  I believe it is a fine line.
And I believe that one "crosses the line" when the decisions in the
called routine have to infer a lot of semantic content based on the
parameters, and/or a combination of parameters.
John

jackf@sisna.com wrote:
> 
> On Sun, 18 Feb 1996 17:56:34 +1000, John Osborne
> <josborne@magna.com.au> wrote:
> 
> >Pat, et. al....
> >The compiled .vs. interpreted debate is partially bogus.  Even in
> >compiled languages, on a PC or a mainframe, as soon as ones code enters
> >a runtime library routine, you are -probably- off into
> >interpretation-land.
> >I say -probably- because if your runtime routine has any conditional
> >statements that cause the routine to act differently depending on
> >parameter content (i.e., different logic paths) it is
> >"mini-interpreting".
> >
> >For example, you call the printf routine in -C-, or a formatted -write-
> >statement in Fortran, VB's Format$ routine, or just about any
> >non-trivial Delphi component, and you are executing an interpreter of
> >that language feature.  Sure, the printf function in the -C- runtime
> >library is compiled, as is the ChartFX component in Delphi, or the
> >read/write routines in a Tandem Cobol program, but the runtime system
> >has to -interpret- your parameters via lots of conditional statements to
> >make sense of your parameters/data and perform the requested action.
> >When using VB or other -statement based- interpreted languages, where
> >most people suffer a performance hit is doing things like writing their
> >own INSTR function -within- VB.  Each p-code statement has to be
> >interpreted via VBRUNxxx.DLL which is far worse than just using the
> >library routine.
> >
> >Regards, John
> >
> >viewer wrote:
> >>
> >> Pat White <pat@pwhite.demon.co.uk> wrote:
> >>
> >> >In nearly 40 years in computing I've never come across anything in any
> >> >language definition that prevents compiling. That includes BASIC which
> >> >is where this discussion started and p-code. Likewise, they COULD all be
> >> >interpreted, only most companies realise compiling is better. Microsoft
> >> >is the exception - they tried and failed badly, anyone remember their
> >> >Fortran compiler for the PC. When it compiled without falling over you
> >> >were amazed. When you ran the results you were sorry!
> >>
> >> OTOH, their DOS Basic compilers were fantastic.
> 
> I believe John is incorrect if he is saying as it appears that any
> code which has a decission tree is interpretive.  Interpretive
> languages vs compiled has nothing to do with the logic of the code.
> The point is that compiled (true compiled not pcode) languages are
> already in binary form and so executable directly while each line of
> an interpretive language is converted to binary, each time it is
> executed.  Four calls to the BASIC print function results in for
> "compiles" at run time.  Four calls to printf in C results in one
> compile at compile time and no compiles at run time.  The difference
> is not trivial in terms of speed of execution.  Note that the printf
> function is a bad example as it carries more overhead than most C
> functions.
> Jack