[Pharo-project] Why keymap modifiers change receiver instead of return new instance?
Hello. I found this: shortcut := $a ctrl. self assert: shortcut == shortcut shift Why modifiers change receiver state instead of create new instance? To me it is very bad unexpected behaviour. There is another example: shortcut := $a ctrl. self assert: shortcut == (KMModifier shift + shortcut) Yes. Such operation modified argument of message #+. What the crapp! Why it not returns new instance? I hope you agree to change this behaviour
On Sun, Apr 28, 2013 at 7:57 PM, Denis Kudriashov <dionisiydk@gmail.com>wrote:
Hello.
I found this:
shortcut := $a ctrl. self assert: shortcut == shortcut shift
Why modifiers change receiver state instead of create new instance? To me it is very bad unexpected behaviour.
Hmm, I really never cared before because it was not a problem :$.
There is another example:
shortcut := $a ctrl. self assert: shortcut == (KMModifier shift + shortcut)
Yes. Such operation modified argument of message #+. What the crapp! Why it not returns new instance?
I hope you agree to change this behaviour
The change makes sense, I'm for it :)
On 28 April 2013 20:08, Guillermo Polito <guillermopolito@gmail.com> wrote:
On Sun, Apr 28, 2013 at 7:57 PM, Denis Kudriashov <dionisiydk@gmail.com> wrote:
Hello.
I found this:
shortcut := $a ctrl. self assert: shortcut == shortcut shift
Why modifiers change receiver state instead of create new instance? To me it is very bad unexpected behaviour.
Hmm, I really never cared before because it was not a problem :$.
There is another example:
shortcut := $a ctrl. self assert: shortcut == (KMModifier shift + shortcut)
Yes. Such operation modified argument of message #+. What the crapp! Why it not returns new instance?
I hope you agree to change this behaviour
The change makes sense, I'm for it :)
+1 -- Best regards, Igor Stasenko.
Hello.
I found this:
shortcut := $a ctrl. self assert: shortcut == shortcut shift
Why modifiers change receiver state instead of create new instance?
I can understand since they modify the shortcut.
To me it is very bad unexpected behaviour.
Why?
There is another example:
shortcut := $a ctrl. self assert: shortcut == (KMModifier shift + shortcut)
Yes. Such operation modified argument of message #+. What the crapp! Why it not returns new instance?
I hope you agree to change this behaviour
2013/4/28 stephane ducasse <stephane.ducasse@free.fr>
Hello.
I found this:
shortcut := $a ctrl. self assert: shortcut == shortcut shift
Why modifiers change receiver state instead of create new instance?
I can understand since they modify the shortcut.
To me it is very bad unexpected behaviour.
Why?
Because shortcut1 := $a ctrl. shortcut2 := $a ctrl shift created different shortcuts. And it is fully unexpected that I can not refactor it to: shotcut2 := shortcut1 shift Because after that my shortcut1 become broken. It was happen in TxTextMorph where selection and movement shortcuts created at same time TxTextMorph>>moveEditorOn: aShortcut by: aSelector self on: aShortcut do: [ self moveEditorBy: aSelector withSelection: false ]. self on: selectionModeModifier + aShortcut do: [ self moveEditorBy: aSelector withSelection: true]. To fix it I add deep copy of given shortcut before modifier operation.
There is another example:
shortcut := $a ctrl. self assert: shortcut == (KMModifier shift + shortcut)
Yes. Such operation modified argument of message #+. What the crapp! Why it not returns new instance?
I hope you agree to change this behaviour
Thanks for the explanation but I do not get it
Because
shortcut1 := $a ctrl. shortcut2 := $a ctrl shift
created different shortcuts.
I thought you wanted the inverse: You wrote:
Why modifiers change receiver state instead of create new instance?
And it is fully unexpected that I can not refactor it to:
shotcut2 := shortcut1 shift
ahhh I got it.
Because after that my shortcut1 become broken.
It was happen in TxTextMorph where selection and movement shortcuts created at same time
TxTextMorph>>moveEditorOn: aShortcut by: aSelector
self on: aShortcut do: [ self moveEditorBy: aSelector withSelection: false ]. self on: selectionModeModifier + aShortcut do: [ self moveEditorBy: aSelector withSelection: true].
To fix it I add deep copy of given shortcut before modifier operation.
There is another example:
shortcut := $a ctrl. self assert: shortcut == (KMModifier shift + shortcut)
Yes. Such operation modified argument of message #+. What the crapp! Why it not returns new instance?
I hope you agree to change this behaviour
Yes evil mutations! Let's ban these 2013/4/28 Stéphane Ducasse <stephane.ducasse@inria.fr>
Thanks for the explanation but I do not get it
Because
shortcut1 := $a ctrl. shortcut2 := $a ctrl shift
created different shortcuts.
I thought you wanted the inverse: You wrote:
Why modifiers change receiver state instead of create new instance?
And it is fully unexpected that I can not refactor it to:
shotcut2 := shortcut1 shift
ahhh I got it.
Because after that my shortcut1 become broken.
It was happen in TxTextMorph where selection and movement shortcuts created at same time
TxTextMorph>>moveEditorOn: aShortcut by: aSelector
self on: aShortcut do: [ self moveEditorBy: aSelector withSelection: false ]. self on: selectionModeModifier + aShortcut do: [ self moveEditorBy: aSelector withSelection: true].
To fix it I add deep copy of given shortcut before modifier operation.
There is another example:
shortcut := $a ctrl. self assert: shortcut == (KMModifier shift + shortcut)
Yes. Such operation modified argument of message #+. What the crapp! Why it not returns new instance?
I hope you agree to change this behaviour
On 28 April 2013 22:01, Denis Kudriashov <dionisiydk@gmail.com> wrote:
2013/4/28 stephane ducasse <stephane.ducasse@free.fr>
Hello.
I found this:
shortcut := $a ctrl. self assert: shortcut == shortcut shift
Why modifiers change receiver state instead of create new instance?
I can understand since they modify the shortcut.
To me it is very bad unexpected behaviour.
Why?
Because
shortcut1 := $a ctrl. shortcut2 := $a ctrl shift
created different shortcuts.
And it is fully unexpected that I can not refactor it to:
shotcut2 := shortcut1 shift
Because after that my shortcut1 become broken.
It was happen in TxTextMorph where selection and movement shortcuts created at same time
TxTextMorph>>moveEditorOn: aShortcut by: aSelector
self on: aShortcut do: [ self moveEditorBy: aSelector withSelection: false ]. self on: selectionModeModifier + aShortcut do: [ self moveEditorBy: aSelector withSelection: true].
Wow, that's beatiful piece of code! This is how it should look like: clear and understandable from a first sight. So, we should fix the shortcuts api to let this beauty live :)
To fix it I add deep copy of given shortcut before modifier operation.
There is another example:
shortcut := $a ctrl. self assert: shortcut == (KMModifier shift + shortcut)
Yes. Such operation modified argument of message #+. What the crapp! Why it not returns new instance?
I hope you agree to change this behaviour
-- Best regards, Igor Stasenko.
participants (6)
-
Denis Kudriashov -
Guillermo Polito -
Igor Stasenko -
Nicolas Cellier -
stephane ducasse -
Stéphane Ducasse