I have two apologies to make for statements I've made previously in
TWOG.
First, to the Twenex Fortran compiler. My previous report of the lack
of support for character string manipulation is due not to a short-
coming of the compiler, but to my reliance on a manual outdated even
by Twenex standards. The CHARACTER variable type was apparently not
supported when the 1977 edition of the Fortran Language Manual was
published, which manual I was reading instead of the recently
located 1987 edition, wherein CHARACTER is documented.
A quick compile proved that it is also supported by the Fortran
compiler installed on Twenex. This puts Fortran back at the top of my
list of candidate languages for my Twenex quick-and-dirty programming
projects. Fortran has a bit more of a learning curve compared with C,
but Fortran gives me tiny executable files and more retro cache'. It
may also be useful on other old systems where C isn't available (like
the Living Computer Museum's DEC-10 and TOAD-1).
My second apology is to the KCC compiler. Previously I stated that the
Twenex C compiler produces extremely large executable files compared
to those generated from other language compilers. My comparison was
based on the size of .EXE files produced for Hello World programs
written in various languages. I realized that technically the .EXE
files are produced by the linker program and not the compiler, but I
mistakely assumed that the link step didn't make much difference in
the size of the executable produced.
My assumption was very wrong. Comparing the size of .REL files output
by the various compilers for equivalent Hello World programs, KCC
actually generates a file with respectably small size between that
generated by MACRO assembler and those generated by the Fortran and
Cobol compilers. The executable bloat in C programs appears to be
the fault of the wrongly-assumed-innocuous linker.
