Yea, code that does allInstances just does not get along with GemStone
... That code has been around for awhile and I'm not quite sure if it
actually needs to run on every load....
Of course the problem is showing up because I repackaged some of the
basic monticello methods:
As many of you know, I have spent the fair part of the last few months
porting Monticello to the base 3.0 and that is what triggered these
changes in packaging ... methods need to be packaged differently when
you are bootstrapping with Monticello files instead of a filein ...
The plan is to get to the point where monticello loads/migrations are
"atomic" ... I think we can get pretty close ...
I made the decision to repackage in 2.x even though the real need was in
3.0 so I can see at this point that my decision to repackage in 2.x was
a completely bad idea ... so I'll undo the repackaging for 2.x
altogether ...
Dale
On 11/14/2010 04:52 PM, Nick Ager wrote:
> Hi,
>
> When I tried loading the latest Pier, I got a walkback triggered by:
>
> PRMigrations>>#migrateChildren
>
> GRGemStonePlatform>>#magritteAllSubInstancesOf: aClass do: aBlock
>
>
> which calls the unimplemented:
>
> aClass allSubInstancesDo: aBlock
>
>
> I tried fixing the problem with the following implementation:
>
> ---
>
> Class>>#allSubInstancesDo: aBlock
>
> "Evaluate the argument, aBlock, for each of the current instances of the
>
> receiver and all its subclasses."
>
> self allInstancesDo: aBlock.
>
> self allSubclassesDo: [:sub | sub allInstancesDo: aBlock]
>
>
> Class>>#allInstancesDo: aBlock
>
> self allInstances do: [ :each | aBlock value: each ]
>
>
> ---
>
> The problem is that #allInstances causes an abort which signals an error
> while loading the code.
>
> Note: that in Pharo both methods above are implemented on Behavor not
> Class - I choose Class to match Class>>#allInstances
>
> Nick
Locked
