inline unicode()

Darin Adler khtml-devel@kde.org
Mon, 10 Mar 2003 23:11:45 -0800


On Monday, March 10, 2003, at 09:52  PM, Dirk Mueller wrote:

> On Mon, 10 M=E4r 2003, Darin Adler wrote:
>
>> Except that Safari programmers use development builds all day long=20
>> when
>> doing tasks other than optimization, so the speed does matter a =
little
>> bit to us.
>
> premature optimization is the root of all evil.

Agreed.

> Your inlining is nothing more than a hack, its not even an=20
> optimisation that is worth a penny.

I strongly disagree. I made this particular change based on noticing=20
something being much too slow. Making the change made things noticeably=20=

faster. And it was in a class where the abstraction wasn't working at=20
all. I don't recommend generalizing this change to other classes that=20
don't represent a single character.

> but in other cases you do
> want to have the method wrapper to be able to easily make it noninline=20=

> or
> otherwise more complex in case the implementation has to be changed.

I agree.

> Sorry, I hate to be a pain here but coding style is more important=20
> than the
> runtime performance of debugging builds.

I agree.

> I don't know about your coding skills,

Quite true.

> but I tend to waste a lot more time on thinking about the code and
> browsing files to find a bug than waiting for the compiler. And=20
> sacrifying
> coding style certainly won't make the brain-based jump-around time to
> become any smaller.

Yes.

     -- Darin