[TYPO3-core] Re: Announcing TYPO3 CMS 6.1.0 Alpha1
mathias.schreiber at wmdb.de
Sat Mar 16 21:22:24 CET 2013
Quote: Christian Kuhn wrote on Thu, 14 March 2013 23:17
> Executing a PHP driven enterprise CMS for every single request with such
> a workload is a waste of resources. Use a proxy like varnish. And yes,
> our clients have such workloads, too.
While in general I agree about caches I think it is not optimal to simply point to caches while the whole system bogs down.
It is slower. Nothing to argue about.
This sucks. Nothing to argue about either.
I never had a client complain about his website loading too fast, did you? :)
We need to determine TOGETHER how much performance impact we are willing to take.
I do see the problems to balance "nice code(tm)" and website speed but right now it feels a bit like (of course overstated) "yeah, the thing is getting slower but we don't care, just cache your stuff away".
I also don't look at the "enterprise vs. non enterprise" argument anymore, but more closely take a look at the competition.
If I have two rather comparable systems next to each other and one forces me to set up caches and stuff and the other one doesn't, why should I bother with the one that make me do more work?
Think about it this way:
There are a lot of hired TYPO3 developers.
If the market switches to something because it is faster and it does solve their problems just like T3 does, you put a lot of people out of work.
To the client, the beauty of the code does not count... at all.
Clients want their problems solved - how the dev does it? They don't care.
So I think it is also (!) important to take a look about the market around us and check out the demands of other systems as well.
Apart from that website performance gets more important outside of the "enterprise" level.
Requesting a (singluar) page on the current master takes significantly longer that on 4.5.latest.
So the argument with filling up RAM does not really count here since even a single request takes longer.
Apart from that I don't know about others, but we host at 5 different datacenters and they reduce our power by a huge amount - so the times of setting up huge servers is over as well.
Resources in general are very precious and should be treated alike.
Again: I do get the idea about getting "nice code (tm)" although up to now noone was really able to tell me whats wrong with browsing a lot of files in one folder over browsing a lot of folders with one file in it.
But still I hope we all can agree on the fact that it is important to keep things in a balance so we don't become something like this: https://github.com/Herzult/SimplePHPEasyPlus
This framework is meant as a joke btw. in case someone does not get it (Good laugh btw :))
In general we have two factions atm which is the "by the book" faction on one side and the "simple, efficient, fast" faction on the other side.
I think we can all agree that these two factions are working in two different directions right now and each side feels offended by the other.
I also think we can agree that this situation is rather frustrating for each side as well.
Like in any relationship it is about giving in a bit and taking a bit - think about how you guys handle your girl/boyfriends of wifes.
What I would like to achieve is that both sides sit down at the same table (DevDays maybe?) and try to understand the others and work together.
For my side (the fast one) this means that once someone implements something new or refactors something there is proper performance testing with differently sized datasets and that there is a common understanding that speeding things up is good, slowing things down is bad (in general).
On the other hand the "nice code" department has to have their way (within certain boundaries) since they change things for the better with having tested code, where major refactoring can be handled with ease (correct me if I'm wrong, but this is the overall goal isn't it?).
So as far as I understand it the whole namespace thing together with moving things around was a huge step towards by-the-book coding.
It would be a good time to focus a bit on performance now, wouldn't it?
How can we make the "other side" more aware about the needs (see? not demands) of the other side?
Right now it is just a huge cat-fight while practically I thought we all were partners in crime and fighting for the same course.
I am pretty amazed (in a bad way) about how hardened the fronts are right now.
I can just plead for having a productive session which the maximum of core-devs and try to come up with a solution for that.
Basically this is an open invitation to everyone participating in TYPO3 (which is more that core-work) to get back on the same track and I would love to get this going.
More information about the TYPO3-team-core