[TYPO3-core] RFC: core speedup

[Kasper Skårhøj]

Hi Dmitry.


> Many slow pieces slow down the whole system. Sometimes significantly.

Many nit-picking issues slows down our overall progress as  
developers. Sometimes significantly.

> That code is far from optimal.

I agree but I just don't care unless you can show practical  
significance and not just theoretical.

Care for optimization while you write new code if anything.

>
> I understand that we have even more time consuming places in the code,
> but generally this is not a reason to keep such obvious performance  
> killers.

_Yes_ it is. We are always short of time. We will never be able to  
fix everything. And as Ingmar points out the code is at a very  
significant place in the system which by todays standards would  
require us to produce unit tests that can verify that what we changed  
didn't change behavior. It's not a simple thing. Its risky. (I  
wouldn't care if this was in the direct mail module)

And it is not just about if pow(2,x) or >> is faster. It's also about  
PHP: I always use functionality in PHP which I have used before and  
know how works. While in theory things should work the same they  
don't always in reality. I'm being conservative in order to reduce  
the risks of bugs. Once I couldn't imagine that "kasper"==0 is TRUE!

So must prioritize. The pow(2,x) issue comes out pretty week then.  
Optimize the database indexes or something else instead. That  
couldn't even lead to security bugs.


>
> Kasper, any chances that your decision can be changed?

For 4.0: Absolutely not.
For 4+: Only if there are practical reasons to. So you must prove  
that performance is gained. Otherwise I will insist that we leave it  
be so we can all be reminded that we are here to solve problems of a  
higher order and not get obsessed with details that only programmers  
enjoy.

Finally I would like to notify that I know very well the urge to fix  
things that "hurts my eyes". Sometimes I also allow myself to do it -  
just to satisfy that inner need for beauty. But I also realized that  
it can easily distract me from what is important. I think the same  
may apply in this case (unless above criteria for practicality can be  
proved).


Happy 4.0-release-party!

- kasper

"A contribution a day keeps the fork away"
-------------------------------
kasper2006 at typo3.com | +45 20 999 115 | skype: kasperskaarhoej |  
gizmo: kasper_typo3