Glamour-Examples comparison 4.5 v 4.6 - Class GLMBasicExamples
by Ben Coman
btw, I am on Windows 7.
The GLMExamplesBrowser menu item [View browser tree] is interesting...
NEW IN 4.6
Do the following demonstrate new functionality in 4.6 or just a new
example of existing 4.5 functionality ?
#formatAsWords - what does this have to do with composites protocol?
#listDragAndDrop
new to 4.6 - very nice :)
A useful addition would be allowing drop onto background of [Target]
(which might be item + 0)
Dropping onto a number in [Source] causes an error
An additional comment to describe operation of the example would be
useful eg.. "Drag a number from Source into Target dropping on either an
item or the background"
#smalltalkCode
#spawnBrowserActions - it would be useful to also see how a single
selected item could be opened in a new window - not just the whole list)
#spawnBrowserSelectionActions (note, the method of in-comment example is
incorrect)
[*1] operational change...
4.5 has a default window menu ==> 4.6 does not - why the difference ?[3]
[*2]
finder list ==> finder show: [ :a | a list
browser text ==> browser show: [ :a | a text
finder table ==> finder show: [ :a | a table
Browser menu #simpleFinderWithMenu
Filter #multipleFinderWithFilter
Finder #simpleFinder
Populate port action #populatePortAction
Search #multipleFinderWithFilterAndSearch
Toolbar #browserWithToolbar
Tabs with different labels #tabsWithDifferentLabels
#treeWithInitialSelection - actually this gets a MNU
GLMTabulator>>text in 4.6 - needs one more refactoring
[*3] #using seems to be being purged from Glamour. Perhaps to remove a
lot of methods from GLMBrowser & GLMTPresentationBuilder, leaving these
only in GLMCompositePresentation. Examples...
showOn: ... using: [ browser mondrian ==> transmit to: ....
andShow: [ :a | a mondrian
showOn: ... using: [ browser list ==> transmit to: ... andShow:
[ :a | a list
showOn: ... using: [ browser morph ==> transmit to: ... andShow: [
:a | a morph
Action list #simpleActionList
Allowing all nil #allowAllNil
Allowing nil #allowNil
Drop-down #dropDownList [
Drop-down with initial value #dropDownListWithInitialValue
Fix size pane #fixSizePanes
Morph icons #morphIcons
Tags #compoundTaggedTree
Updated selection #listsWithUpdatedSelection
(no glmBrowser pragma) #singleInitialSelection
[*4] #sendTo:from:with: ==> #transmit:port: ; #from: ;
#transformed
Updated selection #listsWithUpdatedSelection
[*6] Transcript ==> Inspector (not really a significant
change?)
Action list #simpleActionList [*3,6]
Menus #staticAndDynamicMenu [*6]
DOES NOT WORK IN 4.6
#interdependentPanes - code is flagged as not working. does not work in
4.5 either.
FIXED IN 4.6, WAS NON-WORKING IN 4.5
Double click #doubleClick
#simpleTable
#textPortsExamples
#accumulator - Accumulator - looks like an example design change to
separate concerns. Double clicking a number in the left pane, in the
right pane brings up...
4.5 - number tab with sub-tabs 'List' & 'Text' as well as a
connect-menu item 'Inspect'
4.6 - number tab only, no sub-tabs or context menu
WORKS FROM EXAMPLE BROWSER BUT NOT FROM METHOD COMMENT
#textSelection - execute from method comment works in 4.5 but fails in
4.6 with unknown selector #textSelectionInterval.
OPERATES THE SAME - CODE DIFFERENCE AS NOTED
Dashboard #dashboard
Dashboard specifying extents #dashboardWithSpecificExtents
Dashboards in dashboard #dashboardsInDashboard
Composite arrangements #differentComposites
Complex morphs (StarBrowser simulation) #starBrowser
EyeSee interactive bar chart #eyeseeBarDiagram
Mondrian painting #mondrianPainting
Multiple actions #multipleActions
Simple Expander #simpleExpander
Smart lists #treeWithAmountFiltering
Stacker #stacker
Tabs with different actions #tabsWithDifferentActions
Three inter-dependent panes #threeInterdependentPanes
Tree with children by level #treeWithChildrenByLevel
Tree with expansion #treeWithExpansion
(no glmBrowser pragma) #taggedTree
(no glmBrowser pragma) #multiInitialSelection [*1,3] - but I am only
able to multi-select contiguous items with the SHIFT key. I am unable
to multi-select non-contiguous items as I would expect using the CTRL
key. Instead a context menu for MorphTreeMorph appears. Is this
unfortunate behaviour just me?
#magritte - This was one a was I was REALLY interested to see regarding
the update to Magritte 3. However the code is identical to 4.5. Both
4.5 and 4.6 produce the same error....
[Save] ==> error MAWriteError: Not supposed to write to something.
[Cancel] ==> no change. Now without knowing anything about Magritte, I
would expect the text box to revert to the original text.
What would actually be a fantastic example here would be a small address
book of a few entries that browsed a list in one pane and in another the
magritte detail could be edited.
#tableWithIcons
4.6 also adds alternating icons.
4.5 broken & does not transmit to next pane, but same code to 4.6
does work.
selectionAct: [:tree | tree inspect ] ==> selectionAct: [:tree |
tree selection inspect ]
This has been improved in 4.6 to auto-expand when moving from [first
tree] to [second tree]
However in both 4.5 & 4.6, upon initial opening and when switching from
[first tree] to [second tree] an erroneous additional line spacing
occurs as shown in _ treeWithInitialSelection-extra-line-spacing.jpg _
(attached)
#treeWithMenu - might need the same change to selectionAct: as for
#tableWithIcons
#treeWithTags [*1]
#treeWithTagsMoreLevels [*1] - it would be good if after deselecting the
tags, it remembered which node had previously been expanded. When when
I click a tag twice, I expect to get back the same view I had before.
#updateableBrowser [*1] Completely different code and operation. What is
the difference in operation between the 'Upadted automatically' tab and
'Updated via menu' -
#validator [*6]
#validatorDynamic [*6] - but I don't actually see any difference to
operation of #validator method - what am I missing ?
#wizard - it is not clear how to get the selected items
11 years, 6 months
moving to pharo 1.4
by Tudor Girba
Hi,
Now that Moose 4.6 is out, we will move to Pharo 1.4. I will start working on it this week. It would be great to get some help.
Cheers,
Doru
--
www.tudorgirba.com
"One cannot do more than one can do."
11 years, 6 months
Issue 600 in moose-technology: Popup does not work properly
by moose-technology＠googlecode.com
Status: New
Owner: ----
Labels: Type-Defect Priority-Medium Component-Hapao Component-Mondrian
New issue 600 by alexandr...(a)gmail.com: Popup does not work properly
http://code.google.com/p/moose-technology/issues/detail?id=600
It takes times to appear and does not always appears.
I am not sure where the problem comes from. Probably Mondrian. The popup
has problems probably because the visualization takes time to render.
Classes in Hapao are nested within a very large node. This prevent classes
to be catched as a bitmap.
Tested on Pharo 1.3.
Thanks Laurent & Patrick
11 years, 6 months
Problem with getters and setters in VerveineJ
by Santiago Vidal
Hi,
Should VerveineJ mark getters and setters? I saw that the property "kind"
is used for constructors but not for accessors. Is it a bug or should I
compute these values after loading the MSE file? Thanks
Cheers,
Santiago
--
Santiago Vidal
11 years, 6 months
Loading MSE for tests in newest Moose failing
by Dennis Schenk
Hi,
In Moose 4.6 I load a simple MSE for my tests in the following way:
loadIntoMoose: methodName
> | model |
> model := (MooseModel root entityNamed: methodName).
> (model) notNil ifTrue: [^ model].
> model := MooseModel new.
> model name: methodName.
> model importFromMSEStream: (ReadStream on: (self perform: methodName)).
> model install.
> ^ model
Where *self perfom: methodName* returns the MSE as String.
This fails in Moose 4.7 in cacheMooseGroups here:
allPrimitiveTypes
>
> <navigation: 'All primitive types'>
> ^(self allWithType: FAMIXPrimitiveType) sorted: [:a :b | a name < b name];
> yourself
There are three primitive types, but their names are nil.
The peculiar thing is, if I load the MSE file in the Moose Panel manually
it works (same content in the file and in the string!) and I can explore
the three primitive types. They are (in the MSE):
(FAMIX.PrimitiveType (id: 38)
> (name "_unknown_file.void")
> )
> (FAMIX.PrimitiveType (id: 101)
> (name "_unknown_file.int")
> )
> (FAMIX.PrimitiveType (id: 121)
> (name "_unknown_file.boolean")
> )
The only difference between the MSE file and the String I return is that in
the File ' are used and in the String " are used.
So in the MSE file:
(FAMIX.PrimitiveType (id: 38)
> (name '_unknown_file.void')
> )
The String that I return in the method:
^
> '
> ...
> (FAMIX.PrimitiveType (id: 38)
> (name "_unknown_file.void")
> )
> ...
> '
Is this a problem? Does anyone have an explanation for this behavior? Is
there maybe a better way to load moose models for tests?
If anyone wants to look at the code, it is available here:
MCHttpRepository
> location: 'http://www.squeaksource.com/Softwarenaut'
> user: ''
> password: ''
Cheers,
Dennis
11 years, 7 months
Glamour-Examples comparison 4.5 v 4.6 - Class GLMBasicExamples
by Ben Coman
I am just starting to learn Glamour and thought it would be instructive
to see what had changed in the example between 4.5 & 4.6. I thought
someone else might be mildly interested in the results I compiled.
(notation... 4.5-code ==> 4.6 code)
I'll begin by thanking those that took the time to provide these
examples. They are a great way to learn the system. The addition in
4.6 of GLMExamplesBrowser is nice, and the menu item [View browser tree]
provides an interesting rendition of each example.
_EXAMPLES NOW WORKING IN 4.6, WAS NOT IN 4.5_
Double click #doubleClick
Simple table #simpleTable
Text ports #textPortsExamples
Icons #tableWithIcons - 4.5 did not transmit to next pane, but same
code to 4.6 does work.
Also 4.6 refactors.... selectionAct: [:tree | tree inspect ]
==> selectionAct: [:tree | tree selection inspect ]
Also 4.6 adds alternating icons.
_EXAMPLES THAT OPERATE THE SAME IN 4.6 AS 4.5 - CODE IDENTICAL _
* Tree with menu #treeWithMenu - but might need the same change to
selectionAct: as for "Icons #tableWithIcons"
* Dashboard #dashboard
* Dashboard specifying extents #dashboardWithSpecificExtents
* Dashboards in dashboard #dashboardsInDashboard
* Composite arrangements #differentComposites
* Complex morphs (StarBrowser simulation) #starBrowser
* EyeSee interactive bar chart #eyeseeBarDiagram
* Mondrian painting #mondrianPainting
* Multiple actions #multipleActions
* Simple Expander #simpleExpander
* Smart lists #treeWithAmountFiltering
* Stacker #stacker
* Tabs with different actions #tabsWithDifferentActions
* Three inter-dependent panes #threeInterdependentPanes
* Tree with children by level #treeWithChildrenByLevel
* Tree with expansion #treeWithExpansion
* Tree with tags #treeWithTags
* (no glmBrowser pragma) #taggedTree
* Tree withTags more levels #treeWithTagsMoreLevels [*1] -
enhancement... it would be good if after deselecting the tags, it
remembered which node had previously been expanded. When when I click a
tag twice, I expect to get back the same view I had before.
* Magritte presentation #magritte - This was one a was I was REALLY
interested to see regarding the update to Magritte 3. However 4.6 code
is identical to 4.5 and produce the same error....
[Save] ==> error MAWriteError: Not supposed to write to something.
[Cancel] ==> no change. Now without knowing anything about Magritte, I
would expect the text box to revert to the original text.
What would be a REALLY great would be a Magritte 3 example here of
something like a tiny address book of a few entries that in one pane
browsed the list and in another pane showed the detail via Magritte
_EXAMPLE THAT OPERATE THE SAME AS 4.5 - CODE REFACTORED AS NOTED
?? Perhaps someone could comment on the general nature of these
refactorings...
1. #using seems to be being purged from Glamour. Perhaps(?) to remove a
lot of methods from GLMBrowser & GLMTPresentationBuilder, leaving these
only in GLMCompositePresentation. Examples...
showOn: ... using: [ browser mondrian ==> transmit to: ....
andShow: [ :a | a mondrian
showOn: ... using: [ browser list ==> transmit to: ...
andShow: [ :a | a list
showOn: ... using: [ browser morph ==> transmit to: ... andShow:
[ :a | a morph
* Action list #simpleActionList
* Allowing all nil #allowAllNil
* Allowing nil #allowNil
* Drop-down #dropDownList [
* Drop-down with initial value #dropDownListWithInitialValue
* Dynamic validator #validatorDynamic
* Fix size pane #fixSizePanes
* Morph icons #morphIcons
* Tags #compoundTaggedTree
* Updated selection #listsWithUpdatedSelection
* Validator #validator - but I don't actually see any difference to
operation of "Dynamic Validator #dynamicValidator" method - what am I
missing ?
* (no glmBrowser pragma) #singleInitialSelection
2. Clean up similar to [1.]
finder list ==> finder show: [ :a | a list
browser text ==> browser show: [ :a | a text
finder table ==> finder show: [ :a | a table
* Browser menu #simpleFinderWithMenu
* Filter #multipleFinderWithFilter
* Finder #simpleFinder
* Populate port action #populatePortAction
* Search #multipleFinderWithFilterAndSearch
* Toolbar #browserWithToolbar
* Tabs with different labels #tabsWithDifferentLabels
* (no glmBrowser pragma) #treeWithInitialSelection - however this
gets a MNU GLMTabulator>>text in 4.6 - requires one more browser text
===> a text
3. #sendTo: from: with: ==> #transmit: port: ; #from: ;
#transformed
* Updated selection #listsWithUpdatedSelection
4. Transcript ==> Inspector (not really a significant change?)
* Action list #simpleActionList
* Menus #staticAndDynamicMenu
_EXAMPLE CODE AND OPERATION CHANGED SIGNIFICANTLY FROM 4.5 TO 4.6 _
* Accumulator #accumulator
* Updateable browser #updateableBrowser - Completely different
code and operation. What is the difference in operation between the
'Upadted automatically' tab and 'Updated via menu' -
_NEW EXAMPLES IN 4.6_
?? Do the following demonstrate new functionality in 4.6 or just a new
example of existing 4.5 functionality ?
* Format #formatAsWords - what does this have to do with composites
protocol it is located in?
* Smalltalk code #smalltalkCode
* Spawn browser actions #spawnBrowserActions - it would be useful
to also see how a single selected item could be opened in a new window -
not just the whole list)
* Spawn browser selection actions #spawnBrowserSelectionActions
(note, the method of in-comment example is incorrect)
*(no glmBrowser pragma) #listDragAndDrop - very nice :) However
dropping onto a number in [Source] causes an error. Also a useful
additional example would dropping on to the [Target] background (which
might use item + 0)
_ISSUES WITH 4.6 EXAMPLES_
* Two inter-dependent panes #interdependentPanes - code is flagged
as not working. does not work in 4.5 either.
* (no glmBrowser pragma) #treeWithInitialSelection - 4.6 has a
nice additional functionality of auto-expand when moving from [first
tree] to [second tree]. However in both 4.5 & 4.6, upon initial opening
and also when switching from [first tree] to [second tree] an erroneous
additional line spacing occurs as shown in _
treeWithInitialSelection-extra-line-spacing.jpg _ (attached)
* Text selection #textSelection - execute from method comment works
in 4.5 but fails in 4.6 with unknown selector #textSelectionInterval.
However it works from Example Browser
* (no glmBrowser pragma) #wizard - It is not clear how to get hold
of the selected items
* (no glmBrowser pragma) #multiInitialSelection - I am only able to
multi-select contiguous items with the SHIFT key. I am unable to
multi-select non-contiguous items as I would expect using the CTRL key.
Instead a context menu for MorphTreeMorph appears. Is this unfortunate
behaviour just me?
cheers, -ben
11 years, 7 months
the future of moose
by Tudor Girba
Hi,
A project lives through the activity around it. There is a significant potential around Moose, both for research and for industry, but, to make it reality, we need to work together.
One challenge around Moose is to communicate what it does and who is it for. I want to reorganize and position Moose clearly. There is much confusion in the space of analysis tools, but Moose is different:
- First, Moose is a platform for software and data analysis with a distinct goal: to make crafting of analyses practical and affordable. This sets it apart from other tools that aim to provide one-click-analyses.
- Second, Moose has a philosophy behind: humane assessment. This approach provides an economical meaning to the goal of crafting custom analyses. We are first in this area, and I believe this part of software engineering will become important in the future.
- Third, Moose has a research philosophy, too: collaborative & tool-centric research:
http://www.themoosebook.org/book/philosophy/research
The first step is to align the internal structure with the main message. I need help here. My intent is reflected in the abstract architecture described here:
http://www.humane-assessment.com/minibook/moose
http://www.themoosebook.org/book/introduction/nutshell
The second step is to rebuild the web presence. I need help here, too.
The third step is to rebuild collaborations. The analysis space is large, and to make a meaningful impact we need to work (more) closely. This is a call for collaboration:
1. Be a client and use Moose to build your tools. And let us know how it goes.
2. Be a developer and develop the engines of Moose. We have some, but we need more :).
3. Be a documenter and help us document the usage of Moose. There is a ton of things to describe: from technical details, to how to use it in practical settings.
4. Be an evangelist and spread the word about Moose.
5. Surprise us :)
Cheers,
Doru
--
www.tudorgirba.com
"Yesterday is a fact.
Tomorrow is a possibility.
Today is a challenge."
11 years, 7 months
glamorous toolkit is back in the moose image
by Tudor Girba
Hi,
The Glamorous Toolkit, including the Glamorous Inspector, is now in the Moose image. There was a tiny problem of loading due the fact that SystemAnnouncements are already in the base image.
Cheers,
Doru
--
www.tudorgirba.com
"Be rather willing to give than demanding to get."
11 years, 7 months
System browser (ctrl + F)
by Usman Bhatti
I've just downloaded moose-dev-4.7 and the first change i noticed was the
absence of a very useful feature in System browser: search bar. The change
comes from the move from pharo 1.3 to pharo 1.4. Sure, the same thing can
be done with ctrl + F but the search bar was much more easily accessible
and fewer windows in the system.
May be there is some custom browser that I can load to avoid changing my
habits.
regards,
usman
11 years, 7 months
