On 26 juin 2013, at 13:34, Esteban Lorenzano wrote: > > On Jun 26, 2013, at 1:22 PM, Camille Teruel <[email protected]> wrote: > >> >> On 26 juin 2013, at 12:54, Esteban Lorenzano wrote: >> >>> Hi >>> >>> I finished the integration of the new class organizer, and I need people >>> who can test it before integration. >>> Please, merge the SLICE 6757 >> >> Tested and approved señor! > > cool :) > >> Only one question why the AllProtocol is a subclass of Protocol? > > no idea.... I just integrated it into the system (a lot of work by itself)... > but there is always room for improvements in future steps... I would like to > integrate it soon because I need to keep it loading and that means constant > refactors... :) > >> As I understand it, AllProtocol shouldn't respond to addMethod: >> removeMethod: ... and it shouldn't even have a methods instance variable. >> Why don't make an AbstractProtocol with two subclass Protocol and >> VirtualProtocol and make AllProtocol a subclass of VirtualProtocol. >> Like this, in the future we could for example add a new SmartProtocol >> (subclass of VirtualProtocol) that automagically categorize methods given >> some predicate. > > the idea is to add at least an "ExtensionProtocol" who points directly to the > owner rpackage. Also we could star to have "composite protocols" in the > future. > but yeah, as I said... there is a lot of room for improvement, now that we > will have "first class protocols", not just plain strings.
So appart this small design point, since everything works fine, I vote for integration. > >> >>> >>> https://pharo.fogbugz.com/f/cases/6757/ >>> >>> and after that execute this script: >>> >>> Smalltalk allClassesAndTraitsDo: [:c || org proOrg | >>> org := c theNonMetaClass organization. >>> proOrg := ClassOrganization importFrom: org. >>> c theNonMetaClass organization: proOrg. >>> >>> org := c theMetaClass organization. >>> proOrg := ClassOrganization importFrom: org. >>> c theMetaClass organization: proOrg ]. >>> >>> This is an important step not just to have a better class organizer >>> (current is pre-existance of dictionaries... so... :P) >> >> Ah ok this is why it doesn't use Dictionary! I was wondering about the >> rationale behind the old organizer ;). >> >>> It is important also for the next step in RPackage stuff. >>> >>> Esteban >>> >>> >> >> > >
