[V3dot10] Report testing MinimalKernel - MNU when Opening
Monticello. Later error loading SquekmapPackageLoader
Milan Zimmermann
milan.zimmermann at sympatico.ca
Fri Mar 16 06:36:56 UTC 2007
Hi Pavel,
Thanks for your comments - sounds like it makes sense to wait with Morphic
testing till you change the version to 3.10. Let me know if you can use help
testing or debuging something, I am not sure how much time there will be on
my end, but will try to help.
Later Milan
On 2007 March 14 14:21, Pavel Krivanek wrote:
> Hi Milan,
>
> On 3/14/07, Milan Zimmermann <milan.zimmermann at sympatico.ca> wrote:
> > Hi Pavel,
> >
> > First sorry for the late reply here, i have been away for a week, and
> > spent the rest (barely) catching up ...
> >
> > Thanks for your comments, a few notes inline.
> >
> > On 2007 March 4 01:23, Pavel Krivanek wrote:
> > > Hi Milan,
> > >
> > > On 3/4/07, Milan Zimmermann <milan.zimmermann at sympatico.ca> wrote:
> > > > Hi,
> > > >
> > > > I have tried the following steps:
> > > >
> > > > 1) Downloaded KernelImageMC.20070226.zip from
> > > > http://comtalk.net/public/pub/KernelImage/
> > > >
> > > > 2) Started the above image in 3.9 VM, and typed in:
> > > > ((MCHttpRepository location:
> > > > 'http://squeaksource.com/KernelImage'user: '' password: '')
> > > > loadVersionFromFileNamed:'MinimalMorphicLoader-pk.24.mcz') load. (one
> > > > line)
> > > > MinimalMorphicLoader load.
> > > >
> > > > This got me into the Squeak world
> > > >
> > > > 3) Saved and Quit
> > > >
> > > > 4) Restarted
> > > >
> > > > 5) clicked Open->Monticello Browser
> > > >
> > > > ##Got a MessageNotUnderstood:
> > > > DisplayText>>privForegroundColor
> > > >
> > > > Is this expected?
> > > >
> > > > If I abandon, Monticello does open. It is also interesting, when I
> > > > quit after that, and open monticello again, I do not get the MNU
> > > > again....but after a while I did again. Abandoning seems to have
> > > > worked ok.
> > >
> > > This kind of initial errors is quite common but of course not welcome.
> > > I have just solved some of them. So if you will see that, you may try
> > > to find a solution. One iffy problem is with the LRUCache.
> >
> > OK, I will try again with latest version. If I see a problem like this, I
> > will at least try to understand what causes it, or try to fix.
> >
> > > > 6) Then I added this repository:
> > > >
> > > > MCHttpRepository
> > > > location: 'http://squeaksource.com/KernelImage'
> > > > user: ''
> > > > password: ''
> > > >
> > > > 7) Then I loaded EToysLoader-pk.10.mcz
> > > >
> > > > Loaded ok
> > > >
> > > > 8) Then, trying to Open->SqueakMapPackageLoader, I get a MNU:
> > > >
> > > > UndefinedObject>>select
> > >
> > > I looked at it and "SMSqueakMap bootStrap" solves it but we have to
> > > find a better solution.
> >
> > Does this need to run before opening the SqueakMapPackageLoader..
>
> I think that SM should do this automatically if it has no data. We
> will have to check what's wrong in the SM classes initialization.
>
> > > > 9) I loaded and tried my (simple) eToysTests here:
> > > >
> > > > http://www.squeaksource.com/eToysTests.html
> > > >
> > > > and they worked on the build image. They are simple, but still, this
> > > > is great that the built image passes them.
> > >
> > > Great, what about MinimalMoprhic system?
> >
> > After having loaded both Minimal Morphic and eToys, I ran Morphic tests
> > that are available (they are in categories such as EToys-MorphicTests-AAA
> > where AAA=Basic, Kernel, etc). When running them, I get a question:
> >
> > Which sound service would you prefer? Answering "Basic" the test
> > continues.
>
> This should be part of the image initialization.
>
> > There are 3 failures:
> > PasteUpMorphtest
> > SimpleSwitchMorphtest
> > TextMorphTest
> > I will try in the latest version and look into the details of why. This
> > is on the version about 10 days old, so the failures could have changed.
>
> I will look at it.
> I plan to change the Morphic package to correspond to the latest 3.10
> version soon. Over there are all platform independent tests green.
>
> > > > Apart from finishing and adding more eToys tests that I plan, time
> > > > permiting I'd like to help some with 3.10, but not sure where to
> > > > start. I guess first step is for me to ask whether the MNU in Step 5
> > > > and 8 are expected. Also, I assume one thing would be to get rid of
> > > > any overrides that might be in MorpicLoader or eToysLoader, are there
> > > > any?
> > >
> > > There's very simple test how to see the overrides. Firstly you have to
> > >
> > > evaluate this code:
> > > | sources |
> > >
> > > sources := Dictionary new.
> > > Smalltalk allClassesAndTraits do: [:c |
> > >
> > > | methods |
> > >
> > > methods := sources at: c name put: Dictionary new.
> > > c selectors do: [:sel |
> > > methods at: sel put: {(c sourceCodeAt: sel) asString. c
> > > organization categoryOfElement: sel} ].
> > > c isTrait ifFalse: [
> > > methods := sources at: c class name put: Dictionary new.
> > > c class selectors do: [:sel |
> > > methods at: sel put: {(c class sourceCodeAt:
> > > sel) asString. c class organization categoryOfElement: sel} ]]].
> > >
> > > Smalltalk at: #OldSources put: sources.
> > >
> > >
> > > then load the package, open transcript and evaluate this one:
> >
> > Pavel, do you mean "load the mcz which I want to test for overrides?"
>
> Right.
>
> > > OldSources keysAndValuesDo: [:class :methods |
> > > methods keysAndValuesDo: [:sel :source |
> > >
> > > | newSource newCategory |
> > >
> > > newSource := ((Compiler evaluate: class) sourceCodeAt:
> > > sel) asString. newCategory := (Compiler evaluate: class) organization
> > > categoryOfElement: sel. ((newSource = source first) and: [newCategory =
> > > source second]) ifFalse: [ Transcript show: class; show: ' >> ', sel;
> > > cr ]]].
> >
> > Let me understand what this means: It sounds like in the Monticello
> > world, an "override" is a method that exists in the MCZ that we loaded
> > (between evaluating the first and second code), and the method modified
> > it's methodCategory after being loaded from the MCZ (sort of "stole" the
> > methodCategory from what it was in the old code) - is that a correct
> > understanding? Thanks.
>
> Right, if you will want to publish the original package then it will
> be not complete because some methods changed its classification.
>
> > I will run this all over with your latest versions,
>
> There are no changes in MinimalMorphic/EToys because I focused on
> headless usage of the KernelImage as the web/database server in last
> days.
>
> Cheers,
> -- Pavel
>
> > Milan
> >
> > > this is list of overrides of MinimalMorphic:
> > > http://www.comtalk.net/Squeak/102
> > >
> > > Cheers,
> > > -- Pavel
> > >
> > > > Milan
> > > >
> > > > _______________________________________________
> > > > V3dot10 mailing list
> > > > V3dot10 at lists.squeakfoundation.org
> > > > http://lists.squeakfoundation.org/mailman/listinfo/v3dot10
> > >
> > > _______________________________________________
> > > V3dot10 mailing list
> > > V3dot10 at lists.squeakfoundation.org
> > > http://lists.squeakfoundation.org/mailman/listinfo/v3dot10
> >
> > _______________________________________________
> > V3dot10 mailing list
> > V3dot10 at lists.squeakfoundation.org
> > http://lists.squeakfoundation.org/mailman/listinfo/v3dot10
>
> _______________________________________________
> V3dot10 mailing list
> V3dot10 at lists.squeakfoundation.org
> http://lists.squeakfoundation.org/mailman/listinfo/v3dot10
More information about the V3dot10
mailing list