Em 29/12/2011 16:48, Benoit St-Jean < [email protected] > escreveu:

Benoit,

You rise some important points, this is my attempt to contribute:

> What if  a class  was deleted  in Pharo? 

We should get a thorough analysis of the case and understand if the deleted 
class does indeed makes its lacking or if the package in SS has to be 
re-engineered to be compatible with the Pharo view.

Then two outcomes are possible: a recipe for the similar cases where the lack 
of the deleted class is created or we understand the deletion as a mistake and 
reintroduce or replace it with a new implementation of it.

> What  if some  methods are missing?  

Same general methodology: were they suppressed but replaced by others? Or were 
they removed due being only specific to some package and did not belong to the 
places they were in the classes hierarchy? Perhaps were they just 'syntactic 
sugar' and thus have to be replaced by another code snippet?

> What  if it's  referring to a  Preference while  we dumped
> this in favor  of settings ?  What if it's  using some Morphic stuff
> that  was  cleaned up.   Your  comment  assumes  that everything  on
> SqueakSource is outdated an unusable. 
>
>  FYI,  lots of  packages there  (that are  5 years  old)  ***do load
> cleanly*** in Squeak 4.3...

Makes sense, after of all there is called SqueakSource, isn't it?

>  Besides, there are  LOTS of Squeak users that  maintain their stuff
> only for Squeak without considering Pharo.  

To which I think the whole Stef's post applies! If _we_ Pharo comunity long or 
want that package in our changed Smalltalk, we should adapt it.

Our wage is that in time the package originators may become convinced of the 
advantages of: or switching to Pharo, or maintaining in both 'dialects'.

> Should we maintain those
> packages by  adding some  "PharoCompatibility" classes or  we should
> start our own repository ?

I'm a firm believer that '"PharoCompatibility" classes' are no-no. We rather 
have a repo which has packages adapted to Pharo as Stef described.

>  Finally, it's  not because a package  is 5 years old  that it's not
> useful...  If  it's there and  freely available, why would  we throw
> all this code  to the trash ?   To reinvent the wheel ?   I think we
> should seriously  consider porting  a lot of  this stuff to  our own
> repository  before the distance  between Squeak  and Pharo  gets too
> big.

I think each package needs to be considered in isolation, no general rule can 
be created for all cases without incurring in the risk of going irrational...

My 0.0199999....

--
Cesar Rabak

Reply via email to