Re: Tiny optimizations of bit testing operations

[Rein Klazes]

On Wed, 20 Oct 2004 19:48:05 +0900, you wrote:
> Hello,
>
> Changelog:
>     Dmitry Timoshkov 
>     Tiny optimizations of bit testing operations.

Testing with gcc 3.3.5:
> -
>        BOOL min_or_max_box = (wndPtr->dwStyle & WS_MAXIMIZEBOX) ||
> -
>                              (wndPtr->dwStyle & WS_MINIMIZEBOX);
> -
>        BOOL min_or_max_box = wndPtr->dwStyle & (WS_MAXIMIZEBOX | WS_MINIMIZEBOX);

At moderate optimization levels (-O1) gcc compiles this to identical
code.
> -
>    if((wndPtr->dwExStyle & WS_EX_LEFTSCROLLBAR) != 0)
> -
>    if (wndPtr->dwExStyle & WS_EX_LEFTSCROLLBAR)

At any optimization level, gcc compiles this to identical code.
>          if ((wndPtr->dwStyle & WS_VSCROLL) &&
> -
>            ((((wndPtr->dwExStyle & WS_EX_LEFTSCROLLBAR) != 0) && (ptClient.x <= rcClient.left + GetSystemMetrics(SM_CXVSCROLL))) ||
> -
>            (((wndPtr->dwExStyle & WS_EX_LEFTSCROLLBAR) == 0) && (ptClient.x >= rcClient.right - GetSystemMetrics(SM_CXVSCROLL)))))
> -
>            (((wndPtr->dwExStyle & WS_EX_LEFTSCROLLBAR) && (ptClient.x <= rcClient.left + GetSystemMetrics(SM_CXVSCROLL))) ||
> -
>            (!(wndPtr->dwExStyle & WS_EX_LEFTSCROLLBAR) && (ptClient.x >= rcClient.right - GetSystemMetrics(SM_CXVSCROLL)))))

Not different then the previous example.
> From the C point of view these bit logics are identical. The compiler

finds that out easy.

Rein.

Tiny optimizations of bit testing operations

[Dmitry Timoshkov]

"Rein Klazes"  wrote:
> From the C point of view these bit logics are identical. The compiler
> finds that out easy.

Yes, I know. That's why the optimizations are "tiny". I believe that
the code itself becomes more readable with my changes and makes it
not depend on the optimizations a compiler is capable to perform.

Tiny optimizations of bit testing operations

[Rein Klazes]

On Wed, 20 Oct 2004 22:29:35 +0900, you wrote:
> > From the C point of view these bit logics are identical. The compiler
> > finds that out easy.
>
> Yes, I know. That's why the optimizations are "tiny". I believe that
> the code itself becomes more readable with my changes and makes it
> not depend on the optimizations a compiler is capable to perform.

I think even "tiny" is an exaggeration. Gcc has been capable of doing
this for years and years. I'd be surprised if you could find a compiler
that could not, and still compile wine satisfactory.

Rein.

Tiny optimizations of bit testing operations

[Marcus Meissner]

On Wed, Oct 20, 2004 at 04:13:07PM +0200, Rein Klazes wrote:
> On Wed, 20 Oct 2004 22:29:35 +0900, you wrote:
> > > From the C point of view these bit logics are identical. The compiler
> > > finds that out easy.
> >
> > Yes, I know. That's why the optimizations are "tiny". I believe that
> > the code itself becomes more readable with my changes and makes it
> > not depend on the optimizations a compiler is capable to perform.
>
> I think even "tiny" is an exaggeration. Gcc has been capable of doing
> this for years and years. I'd be surprised if you could find a compiler
> that could not, and still compile wine satisfactory.

I guess such changes are acceptable if it makes the sourcecode more readable.

Ciao, Marcus

Tiny optimizations of bit testing operations

[Alexandre Julliard]

Marcus Meissner  writes:
> > I think even "tiny" is an exaggeration. Gcc has been capable of doing
> > this for years and years. I'd be surprised if you could find a compiler
> > that could not, and still compile wine satisfactory.
>
> I guess such changes are acceptable if it makes the sourcecode more readable.

Here it's not clear at all that one is more readable than the other;
and in such a case the policy is that whoever writes the code in the
first place gets to decide how it's done.