[EMAIL PROTECTED] writes:
> Who's quoting man pages? I didn't quote man pages -- I'm describing what
> the C++ standard says.
Actually David, I believe he was referring to my man page quote.
> And of course it makes a difference -- I didn't
> saying that it doesn't. What I said was:
>
> 1. The behavior of <> vs. "" is implementation-defined.
> 2. The use of <> to indicate "system" include files is imposing some
> semantic meaning that is not defined by the C++ standard.
>
> The only explicit difference between the two is that a sequence of
> characters within <> does not necessarily need to be the name of a file.
> That means that a compiler is free to implement:
>
> #include <vector>
>
> as it sees fit. Most implementations will have a file named "vector"
> somewhere in a system-defined place, but it's not required to. Using ""
> requires a source file of that name be included. How that file is found is
> implementation-dependent.
>
> The behavior you describe regarding "searching the current directory" or
> not is a commonly-implemented difference between <> and "", but it's not
> guaranteed by the standard.
>
> Probably the best thing to do is replace all uses of <> with "", but it's
> probably not a high priority for most people.
The issue is how the underlying implementations deal with it. Many of
them are using the standard C pre-processor to deal with things. On
Unix and Windows there are specific meanings to the use of <> vs. "".
I didn't realize the C++ spec said something different about
it. However, given that both Windows and Unix seem to treat it
similarily, I'd say someone got their wires crossed.
I will admit that it's low priority.
jas.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
