At 02:45 PM 9/3/01 +0200, Erik Vanherck wrote:
>>This design is proving to be no good when trying to implement more of the
>>spec. So we are attempting to come up with a new design to layout the
>>objects. We are in desperate need of developers who can contribute to the
>Hmmz, that does not sound good for the overal development. Are you gonna
>start over or actually refactoring the current version ? The design as it
>is now appears inside the code seem to be a bit of spagethi like but if you
>are redesigning that would explain a lot. I guess if you are the code will
>be up to date inside the CVS but the readme files and such are badly in
>need of updating.

The original design, going back to James Tauber, is pretty clean, and well 
thought out. Problem is, the design really anticipated only the simple 
situations; as soon as we started hitting complex decisions (keeps, 
footnotes, floats, etc) then the infrastructure wasn't really there. 
Unfortunately, and this happens in OS, momentum of coding carried us well 
past the point where we simply ought not to have added any more code.

The elegance of code that has been added in the last 12 months in particular 
is inversely proportional, usually, to how closely it is related to core 
layout. It _is_ spaghetti, but it's more the implementation that is 
spaghetti, the top-layer, not the design.

It's not a start-over situation, but a fair amount of refactoring and 
rethinking does need to happen. And we have been thinking about it. The mail 
archives for the list are worth searching, and you should read Keiron's 
recent doc at

Your help is definitely appreciated.


Fairly Senior Software Type
e-plicity (
Wireless * B2B * J2EE * XML --- Halifax, Nova Scotia

To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, email: [EMAIL PROTECTED]

Reply via email to