Updates: Status: Started Comment #10 on issue 3088 by marcus.d...@...: MNU Text>>withBlanksTrimmed http://code.google.com/p/pharo/issues/detail?id=3088 (No comment was entered for this change.)
Updates: Labels: -Milestone-1.2 Milestone-1.3 Comment #11 on issue 3088 by marcus.d...@...: MNU Text>>withBlanksTrimmed http://code.google.com/p/pharo/issues/detail?id=3088 (No comment was entered for this change.)
Hi guys. Just in case someone is interested, I've have started blogging: "Sending messages through small talks"
The blog is here: http://marianopeck.wordpress.com/
Any kind of feedback is more than welcome.
Mariano
Stef wrote:
>Why this hierarchy is wrong?
I did not say that the hierarchy is correct! And agree that
composition is better suited than inheritance here.
But Benjamin said "Workspace is not a model, it's a view basically."
This is wrong since this is more centralized towards what you
see (the workspace view) and not whats behind it.
I'm just at the position that a Browser IS A model and NOT a view
and that it can have one or more different views.
Nothing more said in my post.
Fernando Olivero
>Correct, but this is not enough to advocate for the removal of the
>Browser as a "model" for the view.
+100
Just because the model assembles the view that doesnt mean we
should remove a clean browser model!!!
We should clearly support MVC in the future and also introduce
a View class. Let's think about it:
Model
subclass Browser
View
subclass BrowserView
If we have a browser view - how strongly connected should it
be to Morphic - especially when we think of other kind of UI
bindings like Morphic, SimpleMorphic, GtK, HTML, ...
Or should we name it BrowserMorph?
And is this hierarchy design/naming correct. Let's verify:
It is not necessary that the model assembles the view - but
often wished that you open the view by sending a simple
message. Currently in Pharo you send it to the model:
"Browser open"
If you evaluate the result you SEE is a view, the result
you GET if you inspect it is the model (who shouldnt know
about the views).
But often you also wish to acess the view (who always knows
about its model). So we have a conflict here and have to
use tricks to access the view afterwards.
So what about a different design, lets use the name
"Browser" for the view part:
Model
subclass BrowserModel
View
subclass Browser
If you evaluate "Browser open" you should get a view back
on the browser and if you ask for "model" you should
get the instance of BrowserModel. Looks cleaner but
different from what we currently have.
Maybe we talk all about the same thing and it is more
about naming. We could also be more explicit:
Model
subclass BrowserModel
View
subclass BrowserView
Bye
T.
--
--
Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir
belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de
Ok. Regarding the names, I was arguing in favor of using BrowserMorph instead of Browser for the Morphic side, because it would really a Browser for the Morphic UI. So lets agree on the basic required widgets. On the Nautilus side, what are the widgets needed? In a previous email you mentioned MorphTreeMorph. Lets make a list so i know the status of the SMx port, and what needs to be done at the View level. Basic: 1) Button 2) Menus 3) PluggableTextMorph 4) SystemWindow 5) PluggableListMorph ( and maybe its variants) --------- Nautilus: 5) MorphTreeMorph 6) ? On Thu, Mar 31, 2011 at 10:01 PM, Stéphane Ducasse <stephane.ducasse@...> wrote: >> >> >> I agree with it and would like to help with the Nautilus model, and >> the corresponding view for SimpleMorphic. >> >> I belive we should look at CUIS TexProvider hierarchy, since he >> already cleanup this classes, and the intent is similar to the yours. >> As he told in en email: "To clean the hierarchy is ok, but we should >> be very careful to no put MODEL LOGIC in the VIEW. In CUIS the idea is >> to deal with PluggableTextModel that gets plugged a TextProvider or >> CodeProvider." > > I do not see why we need that at all. > For the moment let us finish Nautilus and remove most of the StringHolder subclasses > then after porting that to SimpleMorphic will just be switching to the correct treeMorph > because the model will be the same in Morphic and in SM > >> Somehow similar to what you propose in the pdf, and what stef is >> talking about. Composition instead of inheritance. >> >> So to summarize: >> >> 1) Model logic in specialized code providers > Not only: a browser model should know not only about code but about the current state of the tools: > which view (packages,.... buttons is selected). > >> 2) View on specialized views, moving the current functionality from >> the model to the view, that dont have model logic. > Not really so far we use a good tree and this is enough > >> 3) I suggest keeping the curent names, for instance, Browser is the >> model, BrowserMorph is the view ( in Morphic) > We will see. > > For the moment this is not important. We should get SM cleaned and port Polymorph on top of it > and clean the main widgets. > > >> >> >> Fernando >> pd: I will sync with Juan on this effort, so we can all work together >> on the next Morphic. >> >> >> >> >> On Thu, Mar 31, 2011 at 5:46 PM, Benjamin >> <benjamin.vanryseghem.pharo@...> wrote: >>> >>> On Mar 31, 2011, at 5:51 PM, Torsten Bergmann wrote: >>> >>>>>> Model >>>>>> TextModel >>>>>> Workspace >>>>>> PluggableTextModel >>>>>> TextProvider >>>>>> CodeProvider >>>>>> Browser >>>>>> HierarchyBrowser >>>>>> Debugger >>>>>> Inspector >>>>> >>>>> For me, this hierarchy is bad. >>>>> Workspace is not a model, it's a view basically. >>>> >>>> Take care not to confuse things here. If you think as "Workspace" >>>> as the window that is popping up when you open a workspace >>>> then you are on the wrong track. >>>> What you see is just one (morphic) view on a model/workspace instance. >>>> >>>> >>>> Workspace, Inspector, Browser - all these ARE models and >>>> you can have different looking views on it or open one or more views >>>> on the same model. >>>> >>>> Try >>>> >>>> |browserModel | >>>> browserModel := Browser new. >>>> Browser >>>> openBrowserView: (browserModel openEditString: nil) >>>> label: 'View 1'. >>>> Browser >>>> openBrowserView: (browserModel openEditString: nil) >>>> label: 'View 2'. >>> >>> >>> >>> Browser class>>#newOnClass: aClass label: aLabel >>> "Open a new class browser on this class." >>> | newBrowser | >>> >>> newBrowser := self new. >>> newBrowser setClass: aClass selector: nil. >>> ^ self >>> openBrowserView: (newBrowser openOnClassWithEditString: nil) >>> label: aLabel >>> >>> >>> openOnClassWithEditString: aString >>> "Create a pluggable version of all the views for a Browser, including views and controllers." >>> ^ self openAsMorphClassEditing: aString. >>> >>> >>> openAsMorphClassEditing: editString >>> "Create a pluggable version a Browser on just a single class." >>> ^UIManager default openBrowser: self asMorphClassEditing: editString >>> >>> >>> >>> >>> >>> So here, a Browser instance send openBrowser: asMorphClassEditing: with self as parameter: >>> >>> openBrowser: aBrowser asMorphClassEditing: editString >>> "Create a pluggable version a Browser on just a single class." >>> | window dragNDropFlag hSepFrac switchHeight mySingletonClassList | >>> >>> window := (SystemWindow labelled: 'later') model: aBrowser. >>> dragNDropFlag := true. >>> hSepFrac := 0.3. >>> switchHeight := 25. >>> mySingletonClassList := PluggableListMorph on: aBrowser list: #classListSingleton >>> selected: #indexIsOne changeSelected: #indexIsOne: >>> menu: #classListMenu:shifted: keystroke: #classListKey:from:. >>> mySingletonClassList enableDragNDrop: dragNDropFlag. >>> >>> aBrowser >>> addLowerPanesTo: window >>> at: (0 <at> hSepFrac corner: 1 <at> 1) >>> with: editString. >>> window >>> addMorph: mySingletonClassList >>> fullFrame: ( >>> LayoutFrame >>> fractions: (0 <at> 0 corner: 0.5 <at> 0) >>> offsets: (0 <at> 0 corner: 0 <at> switchHeight) >>> ). >>> >>> aBrowser >>> addMorphicSwitchesTo: window >>> at: ( >>> LayoutFrame >>> fractions: (0.5 <at> 0 corner: 1.0 <at> 0) >>> offsets: (0 <at> 0 corner: 0 <at> switchHeight) >>> ). >>> >>> window >>> addMorph: aBrowser buildMorphicMessageCatList >>> fullFrame: ( >>> LayoutFrame >>> fractions: (0 <at> 0 corner: 0.5 <at> hSepFrac) >>> offsets: (0 <at> switchHeight corner: 0 <at> 0) >>> ). >>> >>> window >>> addMorph: aBrowser buildMorphicMessageList >>> fullFrame: ( >>> LayoutFrame >>> fractions: (0.5 <at> 0 corner: 1.0 <at> hSepFrac) >>> offsets: (0 <at> switchHeight corner: 0 <at> 0) >>> ). >>> >>> window setUpdatablePanesFrom: #(messageCategoryList messageList). >>> ^ window >>> >>> For me, it's not a Model behavior to add panes ... >>> >>> >>> Ben >>> >>> >>> >>>> >>>> Evaluate it, show the windows side by side and then click >>>> in one: two views on the same model and the changes are >>>> propagated to all dependent views. >>>> >>>> So they ARE MODELS: >>>> Take a browser for example. It should know about which class >>>> is selected, which methods to display, ... it's a code holder >>>> and manager thing. Nothing more ... it doesnt even need a UI. >>>> >>>> But it's "view parts" could be (one or more) real windows >>>> layouted in either morphic or a view in Squeaks old MVC or >>>> browser on a Seaside webpage (actually Seaside really has an HTML >>>> based browser). >>>> >>>> The browser model does for instance care if you are working >>>> on the instance or class side (see #metaClassIndicated >>>> and senders) - but it doesnt care if you switch either using >>>> buttons or radio buttons or whatever ... thats up to the view/presentation >>>> layer. >>>> >>>> In general you should be able to also drive this model >>>> without ever really having to open a view ... >>>> >>>> So that even the tools are models (although most people think of them >>>> as windows) is a major difference in Smalltalk's UI design compared to >>>> most UI frameworks in other languages. >>>> >>>> VisualWorks uses a special class ApplicationModel to subclass >>>> for tools and application windows to make this more clear. >>>> >>>> Java's Swing (written by people who designed VisualWorks UI first) has a >>>> similar but more excessive model design JButton -> ButtonModel, ... >>>> The other extreme is VB6/Delphi where you dont have a model at all >>>> and have to write an own if you want separation ;) >>>> >>>> Nothing said about the code quality of the current implementation >>>> in Pharo... >>>> >>>> Bye >>>> T. >>>> -- >>>> Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir >>>> belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de >>>> >>> >>> >>> >> > >
Status: Accepted Owner: Torsten....@... Labels: Milestone-1.3 New issue 3929 by Torsten....@...: Package Filter for MC based repo browser http://code.google.com/p/pharo/issues/detail?id=3929 This changeset adds a package filter to file based repository browser in Monticello. This makes it easier to find something in large repos like PharoInbox, ... Attachments: PackageFilterForMCFileRepositories.1.cs 2.1 KB packageFilter.png 17.2 KB
Updates: Status: FixProposed Comment #1 on issue 3929 by Torsten....@...: Package Filter for MC based repo browser http://code.google.com/p/pharo/issues/detail?id=3929 (No comment was entered for this change.)
On Apr 1, 2011, at 9:16 AM, Fernando Olivero wrote: > Ok. > > Regarding the names, I was arguing in favor of using BrowserMorph > instead of Browser for the Morphic side, because it would really a > Browser for the Morphic UI. > > So lets agree on the basic required widgets. On the Nautilus side, > what are the widgets needed? In a previous email you mentioned > MorphTreeMorph. Lets make a list so i know the status of the SMx port, > and what needs to be done at the View level. > > Basic: > 1) Button > 2) Menus > 3) PluggableTextMorph > 4) SystemWindow > 5) PluggableListMorph ( and maybe its variants) > --------- > Nautilus: > 5) MorphTreeMorph 6) PluggableIconListMorphOfMany (such a good name ...) 7) IconicButton 8) MenuBuilder (menus based on pragma) and I think that's all Ben > > > On Thu, Mar 31, 2011 at 10:01 PM, Stéphane Ducasse > <stephane.ducasse@...> wrote: >>> >>> >>> I agree with it and would like to help with the Nautilus model, and >>> the corresponding view for SimpleMorphic. >>> >>> I belive we should look at CUIS TexProvider hierarchy, since he >>> already cleanup this classes, and the intent is similar to the yours. >>> As he told in en email: "To clean the hierarchy is ok, but we should >>> be very careful to no put MODEL LOGIC in the VIEW. In CUIS the idea is >>> to deal with PluggableTextModel that gets plugged a TextProvider or >>> CodeProvider." >> >> I do not see why we need that at all. >> For the moment let us finish Nautilus and remove most of the StringHolder subclasses >> then after porting that to SimpleMorphic will just be switching to the correct treeMorph >> because the model will be the same in Morphic and in SM >> >>> Somehow similar to what you propose in the pdf, and what stef is >>> talking about. Composition instead of inheritance. >>> >>> So to summarize: >>> >>> 1) Model logic in specialized code providers >> Not only: a browser model should know not only about code but about the current state of the tools: >> which view (packages,.... buttons is selected). >> >>> 2) View on specialized views, moving the current functionality from >>> the model to the view, that dont have model logic. >> Not really so far we use a good tree and this is enough >> >>> 3) I suggest keeping the curent names, for instance, Browser is the >>> model, BrowserMorph is the view ( in Morphic) >> We will see. >> >> For the moment this is not important. We should get SM cleaned and port Polymorph on top of it >> and clean the main widgets. >> >> >>> >>> >>> Fernando >>> pd: I will sync with Juan on this effort, so we can all work together >>> on the next Morphic. >>> >>> >>> >>> >>> On Thu, Mar 31, 2011 at 5:46 PM, Benjamin >>> <benjamin.vanryseghem.pharo@...> wrote: >>>> >>>> On Mar 31, 2011, at 5:51 PM, Torsten Bergmann wrote: >>>> >>>>>>> Model >>>>>>> TextModel >>>>>>> Workspace >>>>>>> PluggableTextModel >>>>>>> TextProvider >>>>>>> CodeProvider >>>>>>> Browser >>>>>>> HierarchyBrowser >>>>>>> Debugger >>>>>>> Inspector >>>>>> >>>>>> For me, this hierarchy is bad. >>>>>> Workspace is not a model, it's a view basically. >>>>> >>>>> Take care not to confuse things here. If you think as "Workspace" >>>>> as the window that is popping up when you open a workspace >>>>> then you are on the wrong track. >>>>> What you see is just one (morphic) view on a model/workspace instance. >>>>> >>>>> >>>>> Workspace, Inspector, Browser - all these ARE models and >>>>> you can have different looking views on it or open one or more views >>>>> on the same model. >>>>> >>>>> Try >>>>> >>>>> |browserModel | >>>>> browserModel := Browser new. >>>>> Browser >>>>> openBrowserView: (browserModel openEditString: nil) >>>>> label: 'View 1'. >>>>> Browser >>>>> openBrowserView: (browserModel openEditString: nil) >>>>> label: 'View 2'. >>>> >>>> >>>> >>>> Browser class>>#newOnClass: aClass label: aLabel >>>> "Open a new class browser on this class." >>>> | newBrowser | >>>> >>>> newBrowser := self new. >>>> newBrowser setClass: aClass selector: nil. >>>> ^ self >>>> openBrowserView: (newBrowser openOnClassWithEditString: nil) >>>> label: aLabel >>>> >>>> >>>> openOnClassWithEditString: aString >>>> "Create a pluggable version of all the views for a Browser, including views and controllers." >>>> ^ self openAsMorphClassEditing: aString. >>>> >>>> >>>> openAsMorphClassEditing: editString >>>> "Create a pluggable version a Browser on just a single class." >>>> ^UIManager default openBrowser: self asMorphClassEditing: editString >>>> >>>> >>>> >>>> >>>> >>>> So here, a Browser instance send openBrowser: asMorphClassEditing: with self as parameter: >>>> >>>> openBrowser: aBrowser asMorphClassEditing: editString >>>> "Create a pluggable version a Browser on just a single class." >>>> | window dragNDropFlag hSepFrac switchHeight mySingletonClassList | >>>> >>>> window := (SystemWindow labelled: 'later') model: aBrowser. >>>> dragNDropFlag := true. >>>> hSepFrac := 0.3. >>>> switchHeight := 25. >>>> mySingletonClassList := PluggableListMorph on: aBrowser list: #classListSingleton >>>> selected: #indexIsOne changeSelected: #indexIsOne: >>>> menu: #classListMenu:shifted: keystroke: #classListKey:from:. >>>> mySingletonClassList enableDragNDrop: dragNDropFlag. >>>> >>>> aBrowser >>>> addLowerPanesTo: window >>>> at: (0 <at> hSepFrac corner: 1 <at> 1) >>>> with: editString. >>>> window >>>> addMorph: mySingletonClassList >>>> fullFrame: ( >>>> LayoutFrame >>>> fractions: (0 <at> 0 corner: 0.5 <at> 0) >>>> offsets: (0 <at> 0 corner: 0 <at> switchHeight) >>>> ). >>>> >>>> aBrowser >>>> addMorphicSwitchesTo: window >>>> at: ( >>>> LayoutFrame >>>> fractions: (0.5 <at> 0 corner: 1.0 <at> 0) >>>> offsets: (0 <at> 0 corner: 0 <at> switchHeight) >>>> ). >>>> >>>> window >>>> addMorph: aBrowser buildMorphicMessageCatList >>>> fullFrame: ( >>>> LayoutFrame >>>> fractions: (0 <at> 0 corner: 0.5 <at> hSepFrac) >>>> offsets: (0 <at> switchHeight corner: 0 <at> 0) >>>> ). >>>> >>>> window >>>> addMorph: aBrowser buildMorphicMessageList >>>> fullFrame: ( >>>> LayoutFrame >>>> fractions: (0.5 <at> 0 corner: 1.0 <at> hSepFrac) >>>> offsets: (0 <at> switchHeight corner: 0 <at> 0) >>>> ). >>>> >>>> window setUpdatablePanesFrom: #(messageCategoryList messageList). >>>> ^ window >>>> >>>> For me, it's not a Model behavior to add panes ... >>>> >>>> >>>> Ben >>>> >>>> >>>> >>>>> >>>>> Evaluate it, show the windows side by side and then click >>>>> in one: two views on the same model and the changes are >>>>> propagated to all dependent views. >>>>> >>>>> So they ARE MODELS: >>>>> Take a browser for example. It should know about which class >>>>> is selected, which methods to display, ... it's a code holder >>>>> and manager thing. Nothing more ... it doesnt even need a UI. >>>>> >>>>> But it's "view parts" could be (one or more) real windows >>>>> layouted in either morphic or a view in Squeaks old MVC or >>>>> browser on a Seaside webpage (actually Seaside really has an HTML >>>>> based browser). >>>>> >>>>> The browser model does for instance care if you are working >>>>> on the instance or class side (see #metaClassIndicated >>>>> and senders) - but it doesnt care if you switch either using >>>>> buttons or radio buttons or whatever ... thats up to the view/presentation >>>>> layer. >>>>> >>>>> In general you should be able to also drive this model >>>>> without ever really having to open a view ... >>>>> >>>>> So that even the tools are models (although most people think of them >>>>> as windows) is a major difference in Smalltalk's UI design compared to >>>>> most UI frameworks in other languages. >>>>> >>>>> VisualWorks uses a special class ApplicationModel to subclass >>>>> for tools and application windows to make this more clear. >>>>> >>>>> Java's Swing (written by people who designed VisualWorks UI first) has a >>>>> similar but more excessive model design JButton -> ButtonModel, ... >>>>> The other extreme is VB6/Delphi where you dont have a model at all >>>>> and have to write an own if you want separation ;) >>>>> >>>>> Nothing said about the code quality of the current implementation >>>>> in Pharo... >>>>> >>>>> Bye >>>>> T. >>>>> -- >>>>> Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir >>>>> belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de >>>>> >>>> >>>> >>>> >>> >> >> >
Updates: Status: FixToInclude Comment #1 on issue 3896 by marcus.d...@...: 1.3 core has wrong theme enabled http://code.google.com/p/pharo/issues/detail?id=3896 add the following to the next postscript: GLMUITheme beCurrent
RSS Feed502 | |
|---|---|
1579 | |
1588 | |
1326 | |
1105 | |
856 | |
1156 | |
1006 | |
818 | |
769 | |
1052 | |
1377 | |
1425 | |
1847 | |
1448 | |
1647 | |
1348 | |
1120 | |
1000 | |
845 | |
1064 | |
1251 | |
963 | |
1039 | |
1476 | |
2268 | |
2343 | |
2223 | |
2241 | |
1986 | |
1579 | |
1507 | |
1460 | |
1717 | |
1243 | |
1080 | |
1555 | |
1529 | |
1695 | |
1217 | |
2706 | |
2036 | |
1098 | |
1097 | |
1173 | |
1343 | |
1439 | |
863 | |
1155 | |
1075 | |
1162 | |
1169 | |
597 | |
654 | |
707 | |
862 | |
777 | |
201 | |
425 | |
587 | |
94 |
