Re: Eaten by the Tiger? Open Source WO is coming...

Am 30.08.2005 um 09:28 schrieb Georg Tuparev:

> Nat!
>
> What was your motivation of working on a more or less strict clone  
> of EOF - your believe it is still the best technology, or because  
> the gold owners insisted?

The motivation is because I judged it to be the easiest route to  
achieve the goal, which was getting the application independent from  
WO frameworks.

The advantages that immediately come to my head are:

o if you code strictly to the WO 4.5.1 API, you already have the  
documentation
o if you stay binary compatible, you can exchange and mix original  
and custom frameworks as you go along which makes testing and  
migration much easier. We don't have to change the original  
applications at all and can maintain and enhance them alongside the  
framework development
o we eventually have a possibly marketable solution for other hapless  
WO users

For our purposes (multi-user desktop applications), there is no other  
middleware I know of, that would be preferable. Apple EOF is slow,  
but I think it need not to be. At first I thought it needed a lot of  
internal reconstruction, but my current belief is, that it just needs  
better coding.

>
> If you had the chance to improve the current EOF, in what direction  
> would you go?

As thats outside of the scope of my current project, that would have  
to be pretty much demand driven I guess.

Of the top of my head, here's just a few things that one might want  
to do:

I'd be good to provide a few more "set" based operations to make  
better use of SQL. It would be probably good to have another faulting  
technology available. I think the protocol to save changes in  
EOCooperatingObjectStore is fairly weak. I'd like to see a two-phase  
commit possibility. The qualifiers are lacking some expressiveness.  
There could be some sort of communication between different  
EODatabase (or some other class) instances across machines.  
Serialization of entities using the database would be nice.  
EOGlobalID could be based on a different superclass. And so on...


>
> (We also started EOF-like project, but with the time we ended up  
> with something completely different. That's why I am asking these  
> questions)
>

Well what did it turn out to be ? I suspect your applications are  
more web centric ?

Ciao
     Nat!

      Jedenfalls sind zehn Fehlstarts hintereinander    [E. Fuchs]
           ein sehr interessanter Beweis
               fuer unsere Theorie
  von der natuerlichen Ueberlegenheit des Dezimalsystems