At 2:50 AM +0100 on 11/20/99, M. Uli Kusterer wrote:
>>Yeh, well, take a few whacks at Bison. It's been -- um -- emotionally
>>disturbed by this news. Apparently, it likes being dumped as much as people
>>do ;-)
>
>Hmmm...
>
> sounds kinda scary to have emotionally disturbed a Bison ... !
Fortunately, it's depression. The thing is suacidal. I've already called a
psychologist for it.
>
>>Possible, but that code would be even scarrier. The idea is to write a tool
>>so I never, ever have to even open the generated C file.
>
> What's so scary about interpreter code? If you have some discipline
>coding, I'm sure one could write pretty understandable code.
Be definition, a finite state machine is not understandable :( I think mine
might be a _little_ better.
The algorithm is _very_ hard to follow. Regardless of how nice the code is,
it will still be hard to follow.
>>How about any reasonable compiler, because an ANSI-compliant C++ compiler
>>on every platform is a fool's paradice (at least until 2050).
>
> Would be nice, but since everyone's heading towards ANSI these days, I
>hope we can live with creating modified versions for non-ANSI compilers
>that just differ in 2 or three #defines or so.
Not likely. Consider the template mess in XBF. I'm not sure if there is a
single ISO C++ compiler anywhere...
