[Typo3-dev] The Right Crew

[Sven Wilhelm]

Hi Kasper,


with changes all three things are in.

I was one of the guys that did a lot of critism and I wrote a lot of 
Emails with Dan in the past. So I will put out my ideas now.

As I spoke with other guys on the last snowboard tour the biggest 
problem I saw was not to look enough on other projects how the work. 
Open Source means also learning from each other. I use many open source 
projects for private and for business an like to give something back. As 
I use this projects I also learn from each project.

Short example:

 From Eclipse I learned many about architecture. The best guys of 
software engineering are involved in this project. Next you can learn 
from this project is how extreme programming can work in real life. 
Eclipse releases every 6 Month with subreleases every month. That is 
your called mixture between visions and pragmatism.

But if you like to move over to this way, many things have to change. 
The developers have to be fit for the extreme programming style.
And there are other rules that must be setupped: strict unit testing, 
automatic code style check.

 From ZOPE I learned how a system can live without a backend. TYPO3 is 
nice but has a to overleaded backend even if you reduce it to the 
minimum. There are two many ways to do something. And for optimizing it 
the core needs to be a splittet by libraries and human interfaces.
Also what I learned from ZOPE is that the guys try to redeveloped whole 
parts if they don't like they own original implementation. This is at 
least a part of extreme programming, if a things doesn't fit, rewrite it 
  (but with unittests).
And another thing that is very nice are the SPRINTS, short hacker 
meetings in real life where they do a big amount of developing at one 
place together (perfectly for a community).

Things you can learn from your own Linux System.
You use a Linux Kernel that is developed by real cracks. This guys have 
deep understanding of software engineering of big projects (see the 
count of lines in the kernel project). In the kernel project one or two 
guys are MAINTAINERS of a named subsystem like network, ide,...
It's totally usable for TYPO3. Don't split you knowledge into multiple 
parts but concentrate on one part. And this guys are responsible mainly 
for design questions.
Last thing to learn from the Linux Kernel: Sign patches by two people 
before integrating :)

Next thing to learn from your Linux System:
There is an authentication, authorization, accounting layer (the famous 
triple A - AAA), called PAM -Pluggable Authentication Modules.
It's complete adoptable in a light way to TYPO3. There is no need to 
find a new way for a good implementation. It's extreme flexible and usable.

The things I have written until this point fit totally in your three 
categories. The only things that stops all this ideas is the backward 
compatibility, sorry. I am one the guys in the community that shouts for 
a core redevelopment as I'm extreme shure that we will have a better 
starting point after. You do not loose anything. The ideas are saved, 
the motivation should stay an the used technic is better after.

TYPO3 will not loose it's soule with a redevelopment.

At least some short topics that I have summarized:
-> A Virtual File System as a core of data storage with
    "backends" to different kind of data sources as a generic
    way to store data unique to sources like sql(mysql,...), filesystem,
    an xml document(it's nothing more than a filesystem :),
    webdav, ftp, ... (the list is neverending)
-> PAM as I wrote before
-> Registry xml based (much inspired by Eclipse)
-> scaling like ZOPE ZEO
-> Caching to filesystem not to database
-> PHPTAL from ZOPE as PHPTAL implementation available

I started for myself to write down drafts how this parts could work.
If someone is interested in I share this ideas soon, but their target 
where a redeveloping of the core.

Kind regards

Sven

Ps: It's not my motivation to hit anyone. This was just my mind.