Re: How do you create a grammar for a multi-language language?

On Sunday, March 6, 2022 at 3:22:57 PM UTC-8, gah4 wrote:


(snip)


> I don't believe I ever tried preprocessor statements inside C string
> constants, but as far as I know, it works.
  (snip)


> [PHP effectively treats material between ?> and <?php as an instruction
> to print the material as if it were a quoted string.


I suppose it works that way.


> C says that the input is tokenized before it does the preprocessor
> phase, so it does not look inside quoted strings. The # and ##
> preprocessor operators allow some preprocessor time creation of quoted
> strings. -John]


It seems that the traditional C compiler, sometimes used with other
languages, such as Fortran, has different parsing and tokenizing rules.


gcc -E --traditional quote.c


will process files with a preprocessor statement inside quotes.


On the other hand, the result is likely not what was wanted,
at least not for C. Among others, it puts out lines like:


# 6 "quote.c" 2


which then end up inside the string.


On the other hand, if you language uses ' for other than strings,
or for two uses, then it should work.


One of the stranger things that I have known for about 50 years,
is direct access I/O in IBM Fortran IV. There are statements like:


            WRITE(1'N) X, Y, Z


where N is the record in the direct access file. The compiler also
accepts string (Hollerith) constants with apostrophes.
(But not in direct access I/O statements.)


In any case, the --traditional C preprocessor is commonly used with Fortran,
where I suspect C tokenizing could cause problems.


Some C preprocessors, but not the current ISO C version, and it
seems not gcc -E --traditional, will substitute preprocessor symbols
inside quoted strings.


As the OP notes, mixed parsing can have surprising effects!

Return to the comp.compilers page.
Search the comp.compilers archives again.