[TYPO3-core] Re: Re: Memory system woes

[Mathias Schreiber]

Quote: pgampe wrote on Wed, 13 February 2013 13:06
----------------------------------------------------
> It was removed, because it was not used by TER, thus TER extensions never 
> had this flag. At this time, nobody had time to fix the TER - this has 
> change fortunately.

It was documented.
Do we now remove everything that is documented just because someone feels it's not used?
This gotta be a joke, right?

> The refactored code from Christian will provide a better ground to evaluate 
> such a feature in the future again.

Jup. And until then we are to simply tell our clients to stop using TYPO3 and wait for a refactored version?
C'mon...

Again, in case I didn't make myself clear:
The website does not work without this setting because we are at the limits of hardware (aside from a IBM z900).
So basically whoever removed this flag fucked up a 400k project.
Please understand that I am not really happy about this.

> Maybe it can be reimplemented without causing unwanted side effects or 
> complicating the API.

Again: The problem is that it was removed in the first place.

> If you want to have such a feature again, then
> a) Make sure that the TER will be adopted
> b) Add core patch
> c) Add EM support
> d) Patch code usages (sysext)

I want GOOD things to stay where they are and in case someone wants to remove other peoples work to at least THINK about:
- WHO put it there
- WHY it was put there (and FULLY understand the implecations of removing it)
- WHAT WILL CHANGE if I remove this.

> Also thing about other possible side effects, e.g. should this flag change 
> TS include (for extbase - I know that is ugly anyway)?
> Or just extTables?

Improvements are already welcome, of course, but please don't remove stuff that you have no idea about why it was there.
Imagine I would remove the brakes on your car because I have some great ideas how to refactor your ride...

> There are lots of ideas to improve the bootstrap that are currently 
> evaluated (and tested). If you have some ideas, then please join the 
> discussions* - but be open minded.

What about "benchmark first, then refactor"?
I don't really see what's so hard about it.
Could you explain?