[TYPO3-hci] New to T3 internals - Is the backend abstracted?

[t3hci at projekte.richdale.de]

Am 14.06.2006 um 00:46 schrieb Kasper Skårhøj:

> Hi Philip,
>
> It sounds fantastic if you could help with a cool alternative backend
> interface.

You struck the right cord there ... Nearly a year ago I considered  
booting a project to build a full-blown industry-strength asyncronous  
rich-client backend for T3. No other CMS project would benefit so  
notably from that than T3. This is, of course, some seriously crazy  
thing to attempt. Even just kicking it of and maintaining it up to  
usable releases is a bucketload of work. Since I'm struggling in more  
serious areas right now and don't exactly have heaps of resources to  
spare I'll have to ponder that project again before making a decision.

Yet generally I think it is possible to push the usual backend stuff  
OSS CMSes have to a true next level. I mean beyond a reborn Multi- 
Browser T3 Classic, now also known as Ajax. Especially with good  
asynchronous caching mechanisims and a good VM to implement them.  
Xical (www.xical.org) is a Flash based Rich Media Framework and E- 
Learning/Presentation Player I've developed over the past 2,5 years  
and it has some sophisticated mechanisims to ensure optimal bandwidth  
awareness. Some abstraction of the gadgets I built into that and we  
have ourselves the underpinnings of a serious T3 Flash Client.

But there's also room for partial aproaches. For instance, one idea  
I've been having since maintaining www.derma.de (a large T3 site  
under heavy load, 80+ Kilo Hits/Month) is the File-Funnel(TM!). You  
have a little square window app running on you desktop (Java, RunRev,  
XUL, Ajax, Whatever) - only like 150 pixels across. Whenever you want  
to upload a file to a T3 Server you just drag-and-drop it on that  
File-Funnel which then connects to the preset T3 Server Account and  
advertises that files position. Next time around of the asynchronous  
refresh of the regular T3 backend you get an update in your "file- 
pipe" listing that file and can tell T3 to upload it whenever you  
fancy. Cut's the Clickfest of multiple uploads even more and offers a  
little tighter intergration between admin client and server without  
to much of a hassle.

> If our results around usability points out that we need to do many   
> changes that warrants an investment in underlying abstraction,  
> that  is what we will do. We will also make sure our efforts are  
> somehow  modern enough to go into 5.0 development in order not to  
> waste time
> on developing lots of soon-obsolete stuff.
>
> If our results show the opposite we will make a big hack. Reason:  
> The  lifetime of the 4.x branch is probably going to be 3-4 years  
> with a  parallel overlap with 5.0 of say 2 years (just my ideas at  
> this very  point in time, not necessarily what the R&D committee  
> decides).

I hereby request that all aditional backend mechanisims be built are  
built with the new architecture underpinnings in place. Did I observe  
correct in that somewhere between 3.6 and 4.0 T3s weight in code got  
reduced from roughly 40 to about 20 MB? Doesn't that give us some new  
room to just include Symfony (www.symfony-project.com/) and start  
refactoring bit by bit? We could build the new stuff and just 'turn  
on' new architecture things that have been proven to be in place with  
working old-to-new migration mechanisims. T3 is a stunt to install  
anyway. We could have the new architecture underpinnings in place and  
make an PHP 5 OOP compliant install optional for anyone who wants to  
use newer components. I'd personally rather have the installation  
backend and the configuration pages look even scarier than they are  
right now for another 2 years than reprogramm all the neat stuff once  
again when it's 5.0 time.

Sidenote: According to their marketing mill the Joomla! people did a  
transition to the new PHP 5 stuff within the last few months already,  
which, if true, means they will be up to speed for enterprise  
architecture whilst T3 is still trying to catch up with their  
pleasant looking (and extremly apealing) backend design.
Don't get me wrong, T3 is deeply entrenched in Germany, and others  
such as Joomla! don't stand a chance - even the magazine  
(www.joomlamagazin.de) went broke after 2 issues :-). But Rhuks  
awesome default templates (sorry for being blunt folks, but the T3  
default templates look like shit) and Joomlas momentum (http:// 
www.joomla.org/content/view/1231/74/) should have us on our toes.  
Just some sidethoughts on competion and T3 position in the market  
right now.

I know this is not exactly the right place for R&D strategy talk, but  
I presume that you have a hot wire to the R&D commitee. Consider this  
some lobbying for a certain approach at the 5.0 issue. :-) ...
Even just having an extra R&D commitee clearly shows that we've got  
the ('us') germans all over this. ;;-))

Phillip