Re: language design implications for variant records in a pascal-like language

George Neuner schrieb:


> Why do so many languages offer (at least) two forms of conditional
> loop: one with the test at the beginning and another with the test at
> the end? Why not just offer an infinite loop and a way to break out
> that can be tied to any conditional?


With a look at grammars themselves, I found EBNF much easier to read and
write than BNF, except for (some) repetitions. The >=0 repetition {...}
    does not allow to express >0 repetition, so that one has to write
      list ::= some lengthy construct { "," some lengthty construct }.
That's why constructs like
      list ::= ( some lengthy construct )/",".
have been added to e.g. Borland EBNF.


[The typo in the first list left in intentionally...]


Obviously the required loop constructs vary with the domain of a
language. That's why IMO general purpose languages should allow for
several loop constructs, which should be easy to distinguish by
humans.


DoDi

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