Mail Archives: djgpp/1998/01/22/14:02:32
S. M. Halloran wrote:
>
> <rhetorical question>
> When will GNU put out a version of gcc which has an
> AutoCorrect feature (of the kind seen in programs like MS Word) so
> that actions clearly intended by the programmer are carried out to
> the final conclusion?
> </rhetorical question>
I knew a compiler which did! Seriously. It's always annoyed me
when compilers say things like "missing semicolon" - if they know
it's missing why don'tthey just insert it?
One of the early Ada compilers did just that. If it could work
out what was wrong it corrected it and tried to compile the rest
of the program. If it succeeded then it output a message warning
you that the executable might not be correct but still generated
it, allowing you to make the decision.
I have to say that I never caught it out in an incorrect 'correction'.
It almost certainly could have been done, and perhaps Ada is a little
bit easier to do that with than C (more 'redundancy' in te linguistic
sense), but it certainly saved time.
Of course, that was in the days when a 200 line program took 20
minutes to compile; far more time was wasted in editing the source
and recompiling than in just trying to run the program and letting
the debugger catch the errors. And it wasn't a production compiler,
in fact it said in the manual that for production code that option
must be switched off.
What I want is the telepathic interface. "Look, I know I /wrote/
i++ but it's obvious I /meant/ j++!"...
> <humor>
> Indeed Microsoft probably uses such an autocorrecting compiler,
> since it clearly seems to be the compiler that produced Windows 95.
> </humor>
I don't regard that as humour. I regard it as the most probable
scenario. Oh, you said 'humor' not 'humour' - that's the
difference, I was thinking of British humour <g>...
Chris C
- Raw text -