Problem upgrading seaside from an existing image (Pharo #14438 and Seaside 3.0)

I've some problems to upgrade Seaside. If I start from a standard Pharo image 1.4 summer (#14457) and evaluate:

Gofer new
    squeaksource: 'MetacelloRepository';
    package: 'ConfigurationOfSeaside30';
    load.
(Smalltalk at: #ConfigurationOfSeaside30) load.

everything's fine, but if I upgrade from my image I receive a SubscriptOutOfBounds :-(

My Version of Pharo and Seaside:
- Pharo #14438
- Seaside (3.0.7, I guess), from monticello browser I see my seaside packages:
-- Seaside-Canvas (pmm.117)
-- Seaside-Core (NickAger.768)
-- Seaside-Session (pmm.157)
If you need to know other packages, I can show them


Log extract follows:

THERE_BE_DRAGONS_HERE
SubscriptOutOfBounds: 0
5 September 2013 9:26:27.448 am

VM: Mac OS - intel - 1084 - CoInterpreter VMMaker-oscog-EstebanLorenzano.155 uuid: c53d1f34-009e-425c-9f88-398f8d39c5e8 Apr 15 2012, StackToRegisterMappingCogit VMMaker-oscog-EstebanLorenzano.155 uuid: c53d1f34-009e-425c-9f88-398f8d39c5e8 Apr 15 2012, https://git.gitorious.org/cogvm/blessed.git Commit: 043d256c0d39b78e04db8608c0ebfbf096223a3f Date: Sun Apr 15 14:18:12 2012 +0200 By: Guille <guillermopolito@gmail.com>
Image: Pharo1.4 [Latest update: #14438]

ByteString(Object)>>errorSubscriptBounds:
        Receiver: ''
        Arguments and temporary variables:
                index: 0
        Receiver's instance variables:
''

ByteString(Object)>>at:
        Receiver: ''
        Arguments and temporary variables:
                index: 0
        Receiver's instance variables:
''
.............


--- The full stack ---
ByteString(Object)>>errorSubscriptBounds:
ByteString(Object)>>at:
ByteString>>at:
ByteString(SequenceableCollection)>>last
GoferResolvedReference(GoferVersionReference)>>parseName:
GoferResolvedReference(GoferVersionReference)>>initializeName:
GoferResolvedReference>>initializeName:repository:
GoferResolvedReference class>>name:repository:
[:each | GoferResolvedReference name: each repository: self] in MCHttpRepository(MCFileBasedRepository)>>goferReferences
Array(SequenceableCollection)>>collect:
MCHttpRepository(MCFileBasedRepository)>>goferReferences
[aRepository goferReferences asSortedCollection asArray] in MetacelloGofer(Gofer)>>basicReferencesIn:
BlockClosure>>on:do:
........

Can you help me?
TIA

Dave

Hi Dave,

We have moved the Seaside configuration to ConfigurationOfSeaside3, and can be found on smalltalkhub. Can you try it with this configuration if an upgrade is possible?

Gopher new
        url:'http://smalltalkhub.com/mc/Seaside/MetacelloConfigurations/main';
        package: 'ConfigurationOfSeaside3';
        load.
(Smalltalk at: #ConfiguratoinOfSeaside3) project load: #stable (this loads version 3.0.8)

If the problem is here as well, please let us know. Then we have some work to do.

Cheers,
Diego

On Sep 5, 2013, at 9:57 AM, Dave wrote:

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

DiegoLont wrote

Hi Dave,

We have moved the Seaside configuration to ConfigurationOfSeaside3, and can be found on smalltalkhub. Can you try it with this configuration if an upgrade is possible?

Gopher new
        url:'http://smalltalkhub.com/mc/Seaside/MetacelloConfigurations/main';
        package: 'ConfigurationOfSeaside3';
        load.
(Smalltalk at: #ConfiguratoinOfSeaside3) project load: #stable (this loads version 3.0.8)

If the problem is here as well, please let us know. Then we have some work to do.

Cheers,
Diego

Hi Diego,
Thanks for your help. There are a couple of typos in your instructions, I rewrite them below for the readers of the mailing list:

Gofer new
        url:'http://smalltalkhub.com/mc/Seaside/MetacelloConfigurations/main';
        package: 'ConfigurationOfSeaside3';
        load.
(Smalltalk at: #ConfigurationOfSeaside3) project load: #stable


Anyway, unfortunately nothing changed, the log is the same:

--- The full stack ---
ByteString(Object)>>errorSubscriptBounds:
ByteString(Object)>>at:
ByteString>>at:
ByteString(SequenceableCollection)>>last
GoferResolvedReference(GoferVersionReference)>>parseName:
GoferResolvedReference(GoferVersionReference)>>initializeName:
GoferResolvedReference>>initializeName:repository:
GoferResolvedReference class>>name:repository:
[:each | GoferResolvedReference name: each repository: self] in MCHttpRepository(MCFileBasedRepository)>>goferReferences
Array(SequenceableCollection)>>collect:
MCHttpRepository(MCFileBasedRepository)>>goferReferences
[aRepository goferReferences asSortedCollection asArray] in MetacelloGofer(Gofer)>>basicReferencesIn:
BlockClosure>>on:do:
MetacelloGofer(Gofer)>>basicReferencesIn:
....

Davide

Mac 10.6.8, a recent vm & image 14459
 'NBCoInterpreter NativeBoost-CogPlugin-GuillermoPolito.19 uuid: acc98e51-2fba-4841-a965-2975997bba66 Sep  3 2013, NBCogit NativeBoost-CogPlugin-GuillermoPolito.19 uuid: acc98e51-2fba-4841-a965-2975997bba66 Sep  3 2013, [hidden email]:pharo-project/pharovm.git Commit: 581a11414063fec0e53f047ca90de5e1b6d3b188 Date: 2013-08-28 15:11:08 +0200 By: GuillePolito <[hidden email]> Jenkins build #14636'

seems to work fine too. Refers to an older Zinc version though.
When looking at the differences between 14438 and latest monticello packages for 1.4,
there are a few ugly ones (joining dictionaries, hash of methods).
Could be a vm problem otherwise.

Stephan

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Stephan Eggermont wrote

Mac 10.6.8, a recent vm & image 14459
 'NBCoInterpreter NativeBoost-CogPlugin-GuillermoPolito.19 uuid: acc98e51-2fba-4841-a965-2975997bba66 Sep  3 2013, NBCogit NativeBoost-CogPlugin-GuillermoPolito.19 uuid: acc98e51-2fba-4841-a965-2975997bba66 Sep  3 2013, [hidden email]:pharo-project/pharovm.git Commit: 581a11414063fec0e53f047ca90de5e1b6d3b188 Date: 2013-08-28 15:11:08 +0200 By: GuillePolito <[hidden email]> Jenkins build #14636'

seems to work fine too. Refers to an older Zinc version though.
When looking at the differences between 14438 and latest monticello packages for 1.4,
there are a few ugly ones (joining dictionaries, hash of methods).
Could be a vm problem otherwise.

Stephan

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Stephan,

I downloaded the last stable vm for mac (march 13), but there is no improvement

THERE_BE_DRAGONS_HERE
SubscriptOutOfBounds: 0
5 September 2013 5:08:22.799 pm

VM: Mac OS - intel - 1084 - NBCoInterpreter NativeBoost-CogPlugin-EstebanLorenzano.18 uuid: a53445f9-c0c0-4015-97a3-be7db8d9ed6b Mar 13 2013, NBCogit NativeBoost-CogPlugin-EstebanLorenzano.18 uuid: a53445f9-c0c0-4015-97a3-be7db8d9ed6b Mar 13 2013, git://gitorious.org/cogvm/blessed.git Commit: 412abef33cbed05cf1d75329e451d71c0c6aa5a7 Date: 2013-03-13 17:48:50 +0100 By: Esteban Lorenzano <estebanlm@gmail.com> Jenkins build #14535
Image: Pharo1.4 [Latest update: #14438]

ByteString(Object)>>errorSubscriptBounds:
        Receiver: ''
        Arguments and temporary variables:
                index: 0
        Receiver's instance variables:
''
...

I can try a more recent vm, but I don't know where to download it

Moreover, I'm developing on osx 10.8.4

Dave

Stable version from march doesn't sound good. I remember some vm
problems at that time.
http://forum.world.st/Re-Latest-pharo-issue-with-Eliot-s-latest-Cog-td4676466.html
http://forum.world.st/Re-where-to-report-vm-crashes-while-running-Moose-tests-td4676253.html
http://forum.world.st/where-to-report-vm-crashes-while-running-Moose-tests-td4676157.html#a4676251
 
I got these from http://files.pharo.org/vm/pharo/mac/

Stephan_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Stephan Eggermont wrote

Stable version from march doesn't sound good. I remember some vm
problems at that time.
http://forum.world.st/Re-Latest-pharo-issue-with-Eliot-s-latest-Cog-td4676466.html
http://forum.world.st/Re-where-to-report-vm-crashes-while-running-Moose-tests-td4676253.html
http://forum.world.st/where-to-report-vm-crashes-while-running-Moose-tests-td4676157.html#a4676251
 
I got these from http://files.pharo.org/vm/pharo/mac/

Stephan_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Sigh, nothing new even with the last vm

THERE_BE_DRAGONS_HERE
SubscriptOutOfBounds: 0
5 September 2013 8:25:29.232 pm

VM: Mac OS - intel - 1084 - NBCoInterpreter NativeBoost-CogPlugin-GuillermoPolito.19 uuid: acc98e51-2fba-4841-a965-2975997bba66 Sep  3 2013, NBCogit NativeBoost-CogPlugin-GuillermoPolito.19 uuid: acc98e51-2fba-4841-a965-2975997bba66 Sep  3 2013, git@github.com:pharo-project/pharovm.git Commit: 581a11414063fec0e53f047ca90de5e1b6d3b188 Date: 2013-08-28 15:11:08 +0200 By: GuillePolito <guillermopolito@gmail.com> Jenkins build #14636
Image: Pharo1.4 [Latest update: #14438]

ByteString(Object)>>errorSubscriptBounds:
        Receiver: ''
        Arguments and temporary variables:
                index: 0
        Receiver's instance variables:
''
...

The error is always when: Fetching Refactoring-Spelling-EstebanLorenzano.29
see here http://imgur.com/SMYwH0h

Repeatable from a clean Pharo 14438?

Stephan
_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Yes, it is a repeatable problem. I can reproduce it from a clean 14438
_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Repeatable in 14437
_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

14449 no longer shows the problem. (Complains about Zinc-HTTP though)

It loads
Comet-Core-lr.51.mcz
Comet-Examples-lr.23.mcz
Comet-Pharo-Core-lr.6.mcz
Comet-Tests-Core-lr.11.mcz
ConfigurationOfGrease-MattSpr.207.mcz
ConfigurationOfOmniBrowser-topa.185.mcz
ConfigurationOfRefactoringBrowser-topa.76.mcz
ConfigurationOfSPort2-DamienCassou.11.mcz
ConfigurationOfSeaside3-DiegoLont.15.mcz
ConfigurationOfZincHTTPComponents-SvenVanCaekenberghe.52.mcz
Grease-Core-NickAger.70.mcz
Grease-Pharo-Core-JohanBrichau.52.mcz
Grease-Slime-topa.15.mcz
Grease-Tests-Core-pmm.80.mcz
Grease-Tests-Pharo-Core-lr.13.mcz
Grease-Tests-Slime-JohanBrichau.11.mcz
JQuery-Core-lr.123.mcz
JQuery-Tests-Core-NickAger.132.mcz
JQuery-Tests-UI-pmm.82.mcz
JQuery-UI-lr.80.mcz
Javascript-Core-pmm.92.mcz
Javascript-Pharo-Core-DaleHenrichs.9.mcz
Javascript-Tests-Core-pmm.66.mcz
Javascript-Tests-Pharo-Core-lr.1.mcz
OB-Standard-EstebanLorenzano.552.mcz
Prototype-Core-lr.25.mcz
Prototype-Tests-Core-lr.13.mcz
RSS-Core-lr.67.mcz
RSS-Examples-lr.26.mcz
RSS-Tests-Core-lr.30.mcz
Scriptaculous-Components-lr.111.mcz
Scriptaculous-Core-pmm.102.mcz
Scriptaculous-Tests-Components-lr.39.mcz
Scriptaculous-Tests-Core-pmm.61.mcz
Seaside-Canvas-JohanBrichau.118.mcz
Seaside-Component-jok.92.mcz
Seaside-Core-JohanBrichau.769.mcz
Seaside-Development-pmm.156.mcz
Seaside-Email-lr.21.mcz
Seaside-Environment-NickAger.77.mcz
Seaside-Examples-lr.23.mcz
Seaside-FileSystem-pmm.26.mcz
Seaside-Flow-pmm.23.mcz
Seaside-HTML5-pmm.28.mcz
Seaside-InternetExplorer-pmm.7.mcz
Seaside-Pharo-Canvas-lr.3.mcz
Seaside-Pharo-Continuation-lr.5.mcz
Seaside-Pharo-Core-JohanBrichau.17.mcz
Seaside-Pharo-Development-lr.57.mcz
Seaside-Pharo-Email-lr.13.mcz
Seaside-Pharo-Environment-lr.6.mcz
Seaside-Pharo-Flow-lr.5.mcz
Seaside-Pharo-Tools-OmniBrowser.pharoOB-NickAger.7.mcz
Seaside-Pharo-Tools-Web-NickAger.39.mcz
Seaside-Pharo-Welcome-NickAger.2.mcz
Seaside-RenderLoop-lr.73.mcz
Seaside-Session-pmm.157.mcz
Seaside-Slime-JohanBrichau.83.mcz
Seaside-Tests-Canvas-dkh.58.mcz
Seaside-Tests-Component-jok.33.mcz
Seaside-Tests-Core-NickAger.257.mcz
Seaside-Tests-Development-pmm.43.mcz
Seaside-Tests-Email-lr.14.mcz
Seaside-Tests-Environment-lr.27.mcz
Seaside-Tests-Examples-lr.9.mcz
Seaside-Tests-FileSystem-lr.13.mcz
Seaside-Tests-Flow-dkh.34.mcz
Seaside-Tests-Functional-pmm.142.mcz
Seaside-Tests-HTML5-pmm.28.mcz
Seaside-Tests-InternetExplorer-lr.9.mcz
Seaside-Tests-Pharo-Continuation-lr.8.mcz
Seaside-Tests-Pharo-Core-JohanBrichau.26.mcz
Seaside-Tests-Pharo-Development-pmm.24.mcz
Seaside-Tests-Pharo-Email-pmm.1.mcz
Seaside-Tests-Pharo-Functional-lr.13.mcz
Seaside-Tests-RenderLoop-lr.11.mcz
Seaside-Tests-Session-lr.49.mcz
Seaside-Tests-Slime-JohanBrichau.43.mcz
Seaside-Tests-Tools-Web-lr.18.mcz
Seaside-Tests-UTF8-lr.10.mcz
Seaside-Tests-Welcome-lr.6.mcz
Seaside-Tools-Core-lr.19.mcz
Seaside-Tools-OmniBrowser-lr.25.mcz
Seaside-Tools-Web-NickAger.113.mcz
Seaside-Welcome-NickAger.40.mcz
Seaside-Widgets-lr.26.mcz
Sport-2.033.mcz
Zinc-Character-Encoding-Core-SvenVanCaekenberghe.21.mcz
Zinc-FileSystem-Legacy-SvenVanCaekenberghe.5.mcz
Zinc-HTTP-SvenVanCaekenberghe.375.mcz
Zinc-Pharo-Forward-Compatibility-SvenVanCaekenberghe.1.mcz
Zinc-Resource-Meta-Core-SvenVanCaekenberghe.17.mcz
Zinc-Seaside-SvenVanCaekenberghe.40.mcz
Zinc-Tests-SvenVanCaekenberghe.194.mcz

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Stephan Eggermont wrote

Repeatable in 14437
_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Thanks for your further investigation Stephan
Dave

When comparing the ConfigurationOf... loaded
In 14438 we have
ConfigurationOfGofer-dkh.39.mcz
which isn't in 14449
Also 14438 needed newer monticello and metacello packages.

Stephan
_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Stephan Eggermont wrote

When comparing the ConfigurationOf... loaded
In 14438 we have
ConfigurationOfGofer-dkh.39.mcz
which isn't in 14449
Also 14438 needed newer monticello and metacello packages.

Stephan
_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Stephan,
I've just downloaded 14438 and 14449 from http://files.pharo.org/image/14/

But, looking at monticello browser:

* 14449 has ConfigurationOfGofer-dkh.36.mcz not ConfigurationOfGofer-dkh.39.mcz
* 14438 has no Gofer configuration.

Otherwise, my local 14438 image has ConfigurationOfGofer-dkh.36.mcz hmm... I'm puzzled!

Dave

Ok, I wasn't clear.
I was looking at the packages downloaded to the package cache
_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Stephan Eggermont wrote

Ok, I wasn't clear.
I was looking at the packages downloaded to the package cache
_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

I see, but in the package cache you can have several versions of a package, only monticello shows the right version that image is using. Am I wrong?

Dave

Dave wrote
>I see, but in the package cache you can have several versions of a package,
>only monticello shows the right version that image is using. Am I wrong?

This was with fresh images, so the package cache shows the packages downloaded
while running

Gofer new
        url:'http://smalltalkhub.com/mc/Seaside/MetacelloConfigurations/main';
        package: 'ConfigurationOfSeaside3';
        load.
(Smalltalk at: #ConfigurationOfSeaside3) project load: #stable

at least up to the crash.

Configurations only check if Metacello is loaded, not which version is loaded
You might want to upgrade Metacello first.

Stephan_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Dave,

Thanks for fixing my script, I was rather fast in typing a reply.

There can be several problems remaining, that might be a bit more difficult to find out:
        1) Metacello might be outdated. The configuration of Seaside is quite complex, and therefor needs a recent version. Loading a newer version of Metacallo before upgrading Seaside might help.
        2) Your package cache might be polluted. Trying to upgrade an image, always put things in your package cache first (downloads them there) and wrong packages here sometimes lead to trouble. But I understand that Stephan tested with a clean image, so this is not the case.
        3) There might be a problem with one of the (newer) used configurations. One of the trouble I ran into was that the currentVersion was not determined correctly, so I had to specify Seaside3 differently. I do not know if all configurations are well. Finding out where the problem here is, is harder. But maybe you should try running:
                (ConfigurationOfSeaside3 project version: #stable)

But I will try to find some time to look into your problem.

Good luck,
Diego

On Sep 6, 2013, at 12:11 PM, Dave wrote:

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Dave,

I realize I should have sent you this reply before.

First thing I always check if I run into trouble is whether Metacello can determine what packages to load. If not, I know the problem lies in the configurations, if it can determine, it allows me to load the packages manually in the same order as metacello.

And I always forget what method to call. I have to look it up. Below here only traverses the configuration itself, but is a good first test.
        (ConfigurationOfSeaside3 project version: #stable) expandToLoadableSpecNames: #( default )

Cheers,
Diego

On Sep 6, 2013, at 12:11 PM, Dave wrote:

_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Stephan Eggermont wrote

Dave wrote
>I see, but in the package cache you can have several versions of a package,
>only monticello shows the right version that image is using. Am I wrong?

This was with fresh images, so the package cache shows the packages downloaded
while running

Gofer new
        url:'http://smalltalkhub.com/mc/Seaside/MetacelloConfigurations/main';
        package: 'ConfigurationOfSeaside3';
        load.
(Smalltalk at: #ConfigurationOfSeaside3) project load: #stable

at least up to the crash.

Configurations only check if Metacello is loaded, not which version is loaded
You might want to upgrade Metacello first.

Stephan_______________________________________________
seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Ok, I see.
Dave