When I try (MCWorkingCopy forPackage: (MCPackage named: 'NautilusRefactoring')) unload NautilusUI allInstances do: [ :inst | SystemAnnouncer uniqueInstance unsubscribe: inst]. Smalltalk garbageCollect. (MCWorkingCopy forPackage: (MCPackage named: 'NautilusCommon')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Nautilus')) unload I get the attached system. I do not understand why SpecDebugger has something to do with navigationHistory. Any idea? I also think that we are not good at making sure what we load can be cleanly unloaded. Without that discipline we will never have a modular system. Stef
it's the inspector, not nautilus. => InspectorNavigator >> pushHistory: On 2013-10-02, at 19:20, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
When I try
(MCWorkingCopy forPackage: (MCPackage named: 'NautilusRefactoring')) unload
NautilusUI allInstances do: [ :inst | SystemAnnouncer uniqueInstance unsubscribe: inst]. Smalltalk garbageCollect.
(MCWorkingCopy forPackage: (MCPackage named: 'NautilusCommon')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Nautilus')) unload
I get the attached system. I do not understand why SpecDebugger has something to do with navigationHistory.
Any idea? I also think that we are not good at making sure what we load can be cleanly unloaded. Without that discipline we will never have a modular system.
Stef
<Screen Shot 2013-10-02 at 7.04.02 PM.pdf>
Ok I will check now I do not even understand why and how it is related. Stef
it's the inspector, not nautilus.
=> InspectorNavigator >> pushHistory:
On 2013-10-02, at 19:20, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
When I try
(MCWorkingCopy forPackage: (MCPackage named: 'NautilusRefactoring')) unload
NautilusUI allInstances do: [ :inst | SystemAnnouncer uniqueInstance unsubscribe: inst]. Smalltalk garbageCollect.
(MCWorkingCopy forPackage: (MCPackage named: 'NautilusCommon')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Nautilus')) unload
I get the attached system. I do not understand why SpecDebugger has something to do with navigationHistory.
Any idea? I also think that we are not good at making sure what we load can be cleanly unloaded. Without that discipline we will never have a modular system.
Stef
<Screen Shot 2013-10-02 at 7.04.02 PM.pdf>
Linking the inspector with nautilus is not a good idea to me. Because at the end we can joke the old squeak by saying that it was monolithic but we are building the same. So now I cannot unload Nautilus because when I remove it, it removes NavigationHistory that is used by the inspector and I do not understand why something is failing -> debugger -> fialure of debugger. Stef
Ok I will check now I do not even understand why and how it is related.
Stef
it's the inspector, not nautilus.
=> InspectorNavigator >> pushHistory:
On 2013-10-02, at 19:20, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
When I try
(MCWorkingCopy forPackage: (MCPackage named: 'NautilusRefactoring')) unload
NautilusUI allInstances do: [ :inst | SystemAnnouncer uniqueInstance unsubscribe: inst]. Smalltalk garbageCollect.
(MCWorkingCopy forPackage: (MCPackage named: 'NautilusCommon')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Nautilus')) unload
I get the attached system. I do not understand why SpecDebugger has something to do with navigationHistory.
Any idea? I also think that we are not good at making sure what we load can be cleanly unloaded. Without that discipline we will never have a modular system.
Stef
<Screen Shot 2013-10-02 at 7.04.02 PM.pdf>
So I moved Nautilus-History and retried and I could get a debugger with some announcements not reset So some progress. I will publish a separate package for History Stef On Oct 2, 2013, at 8:19 PM, Stéphane Ducasse <Stephane.Ducasse@inria.fr> wrote:
Linking the inspector with nautilus is not a good idea to me. Because at the end we can joke the old squeak by saying that it was monolithic but we are building the same.
So now I cannot unload Nautilus because when I remove it, it removes NavigationHistory that is used by the inspector and I do not understand why something is failing -> debugger -> fialure of debugger. Stef
Ok I will check now I do not even understand why and how it is related.
Stef
it's the inspector, not nautilus.
=> InspectorNavigator >> pushHistory:
On 2013-10-02, at 19:20, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
When I try
(MCWorkingCopy forPackage: (MCPackage named: 'NautilusRefactoring')) unload
NautilusUI allInstances do: [ :inst | SystemAnnouncer uniqueInstance unsubscribe: inst]. Smalltalk garbageCollect.
(MCWorkingCopy forPackage: (MCPackage named: 'NautilusCommon')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Nautilus')) unload
I get the attached system. I do not understand why SpecDebugger has something to do with navigationHistory.
Any idea? I also think that we are not good at making sure what we load can be cleanly unloaded. Without that discipline we will never have a modular system.
Stef
<Screen Shot 2013-10-02 at 7.04.02 PM.pdf>
The things is that with Nautilus, I pushed more than just Nautilus in the system, like the groups mechanism, the history etc Now, they should be pushed out of Nautilus since there may be (are) used in other part of the system. We should do this clean up phase on all the new tools I think :) (By example NewValueHolder is still part of Spec, where it should be pushed in Tools or something) Ben On Oct 2, 2013, at 8:26 PM, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
So I moved Nautilus-History and retried and I could get a debugger with some announcements not reset So some progress. I will publish a separate package for History
Stef
On Oct 2, 2013, at 8:19 PM, Stéphane Ducasse <Stephane.Ducasse@inria.fr> wrote:
Linking the inspector with nautilus is not a good idea to me. Because at the end we can joke the old squeak by saying that it was monolithic but we are building the same.
So now I cannot unload Nautilus because when I remove it, it removes NavigationHistory that is used by the inspector and I do not understand why something is failing -> debugger -> fialure of debugger. Stef
Ok I will check now I do not even understand why and how it is related.
Stef
it's the inspector, not nautilus.
=> InspectorNavigator >> pushHistory:
On 2013-10-02, at 19:20, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
When I try
(MCWorkingCopy forPackage: (MCPackage named: 'NautilusRefactoring')) unload
NautilusUI allInstances do: [ :inst | SystemAnnouncer uniqueInstance unsubscribe: inst]. Smalltalk garbageCollect.
(MCWorkingCopy forPackage: (MCPackage named: 'NautilusCommon')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Nautilus')) unload
I get the attached system. I do not understand why SpecDebugger has something to do with navigationHistory.
Any idea? I also think that we are not good at making sure what we load can be cleanly unloaded. Without that discipline we will never have a modular system.
Stef
<Screen Shot 2013-10-02 at 7.04.02 PM.pdf>
So victory I could unload completely Nautilus and RB :) Now I will be able to reload it and I will add some unload methods to the configurations and also to key classes. I think that configuration should have a protocol for that. I do not think that we should continue to not have configuration in the image because configurations are project metadata. to reproduce: rename Nautilus-History into History (MCWorkingCopy forPackage: (MCPackage named: 'NautilusRefactoring')) unload. Nautilus cleanUp. KMRepository reset. KMPragmaKeymapBuilder release. NautilusUI allInstances do: [ :inst | SystemAnnouncer uniqueInstance unsubscribe: inst]. Smalltalk garbageCollect. (MCWorkingCopy forPackage: (MCPackage named: 'NautilusCommon')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Nautilus')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Refactoring-Tests-Critics')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Refactoring-Tests-Core')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Refactoring-Tests-Environment')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Refactoring-Tests-Changes')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Refactoring-Spelling')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Refactoring-Critics')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Refactoring-Pharo-Platform')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Refactoring-Changes')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Refactoring-Core')) unload. (MCWorkingCopy forPackage: (MCPackage named: 'Refactoring-Environment')) unload. Stef
participants (3)
-
Benjamin -
Camillo Bruni -
Stéphane Ducasse