lkml.org 
[lkml]   [2005]   [Nov]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: typedefs and structs
On Wed, 9 Nov 2005, J.A. Magallon wrote:

> On Tue, 8 Nov 2005 20:51:25 -0500, Kyle Moffett <mrmacman_g4@mac.com> wrote:
>
> >
> > Pass by value in C:
> > do_some_stuff(arg1, arg2);
> >
> > Pass by reference in C:
> > do_some_stuff(&arg1, &arg2);
> >
> > This is very obvious what it does. The compiler does type-checks to
> > make sure you don't get it wrong. There are tools to check stack
> > usage of functions too. This is inherently obvious what the code
> > does without looking at a completely different file where the
> > function is defined.
> >
> >
> > Pass by value in C++:
> > do_some_stuff(arg1, arg2);
> >
> > Pass by reference in C++:
> > do_some_stuff(arg1, arg2);
> >
> > This is C++ being clever and hiding stuff from the programmer, which
> > is Not Good(TM) for a kernel. C++ may be an excellent language for
> > userspace programmers (I say "may" here because some disagree,
> > including myself), however, many of the features are extremely
> > problematic for a kernel.
> >
>
> Why is it not good for kernel ?
> You want to pass an struct to a function in the best way you can.
> Reference just pases a pointer instead of copying, but you don't
> realize.
> If you want the funcion to be able to modify the struct, code it as
>
> void do_some_stuff(T& arg1,T& arg2)
>
> If you DO NOT want the funcion to be able to modify the struct, code it as
>
> void do_some_stuff(const T& arg1,const T& arg2)

A diligent C programmer would write this as follows:
void do_some_stuff (struct T * a, struct T * b);
versus
void do_more_stuff (const struct T * a, const struct T * b);
So I don't see C++ winning at all here.

> This is far better than in C,. because you get the benefits from
> reference pass without the problems of accidental modification of
> pointer contents. And get rid of arrows -> ;).
>
> If the function modifies the struct it should be obvious from its name,
> not depending if you put an & in the call or not.
> And you stop worrying about argument pass methods.

I think I'll call this my rule #1:
The moment you stop worrying about something is the moment it bites you
in the butt. :-) Much firsthand experience.

> The person who programs the function decides and can even change it without
> you user even noticing.

And if the caller is passing in something that's not meant to be
modified, then the modification causes much badness. Happens with both
languages, too.

> And gcc does nice optimizations when you mix const& and inlining...

As far as I know, nothing stops GCC from doing the exact same
optimizations in the function prototypes given above.

>
> --
> J.A. Magallon <jamagallon()able!es> \ Software is like sex:
> werewolf!able!es \ It's better when it's free
> Mandriva Linux release 2006.1 (Cooker) for i586
> Linux 2.6.14-jam1 (gcc 4.0.2 (4.0.2-1mdk for Mandriva Linux release 2006.1))
>

-Vadim Lobanov
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-11-09 17:24    [from the cache]
©2003-2011 Jasper Spaans