[TYPO3-hci] Slightly OT: Is the current 4.5 Roadmap wrong?
Phillip
t3hci at projekte.richdale.de
Fri Jun 16 17:02:46 CEST 2006
Hi.
Yesterday night I took the time to install the T3 Database localy and
reverse engineer it into Eclipse/Clay.
32 tables. Zero DB level relations enforced with foreign keys.
Within seconds I found at least 2 large things that would need
normalization imediately, before attempting anything else. After
browsing the model for a few minutes you could literaly see where
Kasper or some other Coredev went into a coding spree and added some
new feature. Recognizable either by field-naming or entity-clusters
with simular naming. Don't get me wrong: This is no critique. T3 is
historically grown and that allways shows in software of this size.
Coding is done by humans and with a specific result in mind at a
given time. And one can also overdo it with abstraction and design.
Besides, I've looked at the guts other PHP/MySQL CMSes and compared
to some of them T3 is the hallmark of clean design. :-)
On top of that, we all know that back then and up to a year ago MySQL
wasn't a Database but more some kind of Flatfile with an SQL
Frontend. And that it really didn't matter wether you had a datamodel
or not. You'd just write the stuff whereever you felt it would fit at
the time.
But PHP and MySQL finally have arived in th big league. Given that
the system requirements for T3 are at the top end of PHP CMSes anyway
I'd say one needn't hesitate and require a minimum of MySQL 5 and PHP
5.1 for the next major version of T3.
But the most important thing to BE design is: I honestly am starting
to think we should consider the old T3 finished. The post-classic
frame-based backend got a new skin this year. It looks a little
better *and* still is what the year long crowd of T3 backend users is
used to. Shouldn't we just do a little maintainance, content
abstraction and naming cleanup for 4.5 and leave it at that? Instead
focus on tearing T3 apart and rebuilding it from scratch while
accepting the proven usage concepts as a given, but building an
entirely new persistance layer, offloading object-relational logic to
lower parts of the system. We could totally abstract the BE and make
a larger attempt to streamline the backend GUI. The possiblities are
endless. I'd rather stick with the current BE knowing that we'll have
a new Architecture in place faster and then will be able to do lots
of neat BE stuff.
What I'm saying here is that not only the state of the backend and
the fact that it's sort of arived in a fully developed 'dead-end'
situation but also the state of T3 as a whole actually asks for
moving a total GUI redo to a place where we have a post-5.0 situation.
I could imagine that this topic has been discussed already or is
being discussed among coredevs right now. Are there any conclusions
to this allready?
Phillip
More information about the TYPO3-team-hci
mailing list