The future of SqueakCore (Re: [squeak-dev] [ANN] New Squeak and SqueakCore)

Levente Uzonyi leves at elte.hu
Fri Nov 11 15:48:44 UTC 2011


On Fri, 11 Nov 2011, Bert Freudenberg wrote:

>
> On 11.11.2011, at 11:08, Edgar J. De Cleene wrote:
>
>> Folks here you have the links
>>
>> http://ftp.squeak.org/4.3alpha/Squeak4.3alpha-11777.zip
>> http://ftp.squeak.org/4.3alpha/SqueakCore4.3alpha-11722.zip
>>
>> SqueakCore was build from Squeak4.3alpha-11777.
>
> Nice!
>
>> In the procedure I have DNU because Stacks I choose solve commenting ONE LINE and saving Morphic-edc.588.mcz
>
> A better way would not be commenting that line, because it still is needed when someone actually uses a stack. A real refactoring would be better, but as an interim solution this should work:
>
> addStackMenuItems: menu hand: aHandMorph
> 	"Add appropriate stack-related items to the given menu"
>
> 	(self respondsTo: #isStackBackground) ifFalse: [^self].
> 	self isStackBackground
> 		ifTrue:
> 			[menu add: 'card & stack...' target: self action: #presentCardAndStackMenu]
>
> That's because #isStackBackground is added by the Etoys package, so if that is not present, you get the DNU.
>
> Better yet, addStackMenuItems:hand: and its related methods like presentCardAndStackMenu should be moved to the Etoys package so they can be cleanly unloaded. Some mechanism needs to exist to hook them in (using #respondsTo: is simplest but less elegant).
>
>> Maybe it?s time to discuss how SqueakCore should be in the future.
>
> Yes.

It's pretty straightforward: SqueakCore is the same as the Squeak, but 
all unloadable packages are unloaded. I see no reason to change this in 
the future. If we want to improve SqueakCore, then we should split the 
packages (first of all the Tests package to be able to run the tests of 
all packages which are not unloaded) to define, simplify and minimize the 
dependencies and allow clean package unloading.


Levente

>
>> The other guys have a bootable ISO and we still in nap mode....
>>
>> Edgar
>
> Not sure what you're referring to.
>
> - Bert -
>
>
>
>



More information about the Squeak-dev mailing list