On Wed, May 2, 2012 at 11:44 AM, Mariano Martinez Peck < [email protected]> wrote:
> > > On Wed, May 2, 2012 at 8:21 PM, Eliot Miranda <[email protected]>wrote: > >> Hi Mariano, >> >> On Wed, May 2, 2012 at 1:27 AM, Mariano Martinez Peck < >> [email protected]> wrote: >> >>> >>> >>> On Tue, May 1, 2012 at 10:17 PM, Mariano Martinez Peck < >>> [email protected]> wrote: >>> >>>> >>>> >>>> On Tue, May 1, 2012 at 10:05 PM, Igor Stasenko <[email protected]>wrote: >>>> >>>>> On 1 May 2012 21:45, Mariano Martinez Peck <[email protected]> >>>>> wrote: >>>>> > >>>>> > >>>>> > On Tue, May 1, 2012 at 9:22 PM, Francisco Garau < >>>>> [email protected]> >>>>> > wrote: >>>>> >> >>>>> >> Your code snippet is comparing the associations. The below one >>>>> evaluates >>>>> >> to true: >>>>> >> >>>>> >> (Smalltalk globals associationAt: #ScriptLoader) value >>>>> >> == >>>>> >> ((SmalltalkImage >> #shrinkToCore) literalAt: 4) value >>>>> >> >>>>> > >>>>> > Yes, but the literals of CompiledMethod that refer to classes should >>>>> have >>>>> > the SAME association as Smalltalk globals. In fact, that's the >>>>> reason why we >>>>> > have: >>>>> > >>>>> indeed. >>>>> The question is what leads to creation of multiple associations >>>>> pointing to same global? >>>>> >>>> >>>> >>>> Exactly. I can easily fix it with a Compiler recompileAll. But the main >>>> question is if we have a bug or something that could let us in that state. >>>> >>>> >>> >>> Ok, I wrote an ugly test and I noticed that the ONLY method with this >>> problem in the image is >>> >>> >>> | behaviorAssociations broken | >>> behaviorAssociations := Smalltalk globals associations select: [:each | >>> each value isBehavior or: [each value isTrait]]. >>> broken := OrderedCollection new. >>> (CompiledMethod allInstances select: [:each | each isInstalled]) >>> do: [:aMethod | aMethod literals >>> do: [:aLiteral | aLiteral isVariableBinding >>> ifTrue: [ >>> ((behaviorAssociations includes: aLiteral) >>> and: >>> [(behaviorAssociations identityIncludes: aLiteral) not]) >>> ifTrue: [broken >>> add: (aMethod -> aLiteral)] >>> ] >>> ] >>> ]. >>> broken inspect >>> >> >> This doesn't look for unbound class variables. >> > > Indeed. > > >> I just wrote something that handles this in Squeak trunk: >> >> methodsWithUnboundGlobals >> "Get all methods that use undeclared global objects that are not listed >> in Undeclared. For a clean image the result should be empty." >> >> "SystemNavigation new methodsWithUnboundGlobals" >> >> ^self allSelect: >> [:m| >> m literals anySatisfy: >> [:l| >> l isVariableBinding >> and: [l key isSymbol "avoid class-side methodClass literals" >> and: [(m methodClass bindingOf: l key) isNil >> and: [(Undeclared associationAt: l key ifAbsent: []) ~~ l]]]]] >> >> so the test would be >> self assert: SystemNavigation new methodsWithUnboundGlobals isEmpty >> >> >>> > Ok. If I understand this correctly, this is yet ANOTHER thing to test, > right? I mean, it does not cover my original problem but another one > possible ;) > right? > I was assuming that my test is a superset of yours, but now I see it is not strict enough. It also needs to compare bindings to see that they're identical. I've attached a stricter version. This says that for any global, it should match either the class's notion of what bindingOf: the key is, or bindingOf: should be nil and the binding should be in Undeclared. If the class answers a different binding through bindingOf: or answers no binding and the binding is not in Undeclared then the variable in the method is wrong. methodsWithUnboundGlobals "Get all methods that use undeclared global objects that are not listed in Undeclared. For a clean image the result should be empty." "SystemNavigation new methodsWithUnboundGlobals" ^self allSelect: [:m| m literals anySatisfy: [:l| l isVariableBinding and: [l key isSymbol "avoid class-side methodClass literals" and: [(m methodClass bindingOf: l key) ifNil: [(Undeclared associationAt: l key ifAbsent: []) ~~ l] ifNotNil: [:b| b ~~ l]]]]] So I think with this stricter definition self assert: SystemNavigation new methodsWithUnboundGlobals isEmpty is an adequate test. > Thanks Eliot. > > > > >> And I have 6 methods from the classes SmalltalkImage and ScriptLoader >>> and all the 6 literals are pointing to ScriptLoader. So something weird >>> happened with this class. >>> Any idea? >>> >>> For the moment I will update the test and do a Compiler recompileAll. >>> >>> >>> >>> >>>> >>>>> > Association >> #literalEqual: otherLiteral >>>>> > "Answer true if the receiver and otherLiteral represent the same >>>>> > literal. >>>>> > Variable bindings are literally equals only if identical. >>>>> > This is how variable sharing works, by preserving identity and >>>>> changing >>>>> > only the value." >>>>> > ^self == otherLiteral >>>>> > >>>>> > instead of the Object implementation. >>>>> > >>>>> > >>>>> >> >>>>> >> On 1 May 2012 17:52, Mariano Martinez Peck <[email protected]> >>>>> wrote: >>>>> >>> >>>>> >>> (Smalltalk globals associationAt: #ScriptLoader) == >>>>> ((SmalltalkImage >> >>>>> >>> #shrinkToCore) literalAt: 4) >>>>> >>> gives false when it should be true. If I do a Compiler >>>>> recompileAll it >>>>> >>> gets fixed. So, my question is, is that normal? how could that >>>>> happen? >>>>> >>> is there any real problem behind? >>>>> >>> >>>>> >>> anyway, can we do a recompileAll for the moment? >>>>> >>> >>>>> >>> thanks! >>>>> >>> >>>>> >>> -- >>>>> >>> Mariano >>>>> >>> http://marianopeck.wordpress.com >>>>> >>> >>>>> >> >>>>> > >>>>> > >>>>> > >>>>> > -- >>>>> > Mariano >>>>> > http://marianopeck.wordpress.com >>>>> > >>>>> >>>>> >>>>> >>>>> -- >>>>> Best regards, >>>>> Igor Stasenko. >>>>> >>>>> >>>> >>>> >>>> -- >>>> Mariano >>>> http://marianopeck.wordpress.com >>>> >>>> >>> >>> >>> -- >>> Mariano >>> http://marianopeck.wordpress.com >>> >>> >> >> >> -- >> best, >> Eliot >> >> > > > -- > Mariano > http://marianopeck.wordpress.com > > -- best, Eliot
SystemNavigation-methodsWithUnboundGlobals.st
Description: Binary data
