cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel Fagerstrom <dani...@nada.kth.se>
Subject Re: Public/Private classification
Date Thu, 13 Oct 2005 16:56:54 GMT
Vadim Gritsenko wrote:

> Reinhard Poetz wrote:
>
>> --- Daniel Fagerstrom <danielf@nada.kth.se> schrieb:
>>
>>> IMO we need to find two set of interfaces/classes:
>>> the API of Cocoon, and the set of classes (components) that an
>>> application programmer need JavaDoc for.
>>
> If it ain't public API you ain't need Javadoc for it, period.
>
>>> WDYT?
>>
>>
>> I agree completly with you!
>
>
> I don't. You both miss the point, which is: It is JFDI effort which 
> should bring  in results quickly, namely set of public classes which 
> are agreed by *whole community* to be public, which means *everybody* 
> agrees that these classes and interfaces will be supported and 
> carefully evolved.

Vadim, I don't think you got my point. Having an opinion about 673 
classes seemed to be a lot of work. And asking the *whole community* for 
having an opinion about all of them seem to be a rather time consuming 
way. To me it seem better to have some discussion about how to find an 
efficient way to choose.

Furthermore we cannot just have any ad hoc selection of 
classes/interfaces as public API, all classes/interfaces that is refered 
to from the public API must be part of it. Is there any convenient way 
to get all refered interfaces from a given set of intefaces?

> If you don't agree with public denominator on one of the classes, just 
> mark it D  (we are looking for consensus here!) and have a healthy 
> lengthy discussion about it,

Concensus about exactly what? To me it was unclear what the exact 
purpose of the activity was, and I wrote my mail because I wanted to get 
a better idea about that. And if you are prepared to have helthy lengthy 
discussions about the specific classes I can't understand why you don't 
even bother about commenting my opinions about how to classify.

> but please don't block this effort to produce *good-enough* results 
> quickly in the sake of ivory tower of perfect API separation and 
> bundles and what not...

That is BS, I tried to find a practical way of solving the problem of 
finding a public API with a reasonable amount of effort. I'm not talking 
about any perfectness here. And for the bundle and separation step 
aspect, I think I made it pretty clear that it was a separate and later 
step.

> So, are you on board?

Ity should be rather obvious from my previous post in this thread and 
from earlier posts that I want us to find out the public API for Cocoon. 
But to get any further in this discussion I propose that you re-read my 
last post and give some constructive critic instead of just dissmissing it.

/Daniel


Mime
View raw message