Re: Porting autogsdoc to OSX

[Gregory Martin Pfeil]

On Tuesday, February 26, 2002, at 05:18  AM, Marcus Müller wrote:
> Well, that's something I don't understand. In my understanding #import 
> statements should be better by far, because they take the burden of 
> "safeguarding" your headers from you and they're also faster, as they 
> prevent headers from being opened more than once - something which the 
> #include thing cannot do. So why is it you consider #include superior 
> over #import statements (asking out of curiosity)?

I think the problem with #import is that it prevents multiple inclusion 
at all. You may have a header file that you do want to include multiple 
times (say, if you want something like a mix-in in ruby. You can have a 
header that contains a group of methods, and include it into the body of 
whichever classes need it (there is probably a better example, but I 
can't think of it)).

With #include, the header designer decides whether or not the header 
should be included multiple times, with #import, the compiler decides to 
_never_ include the header more than once.

I'm pretty sure this is the reason #import has been deprecated.