Hi all,
I spent yesterday evening demo'ing Squeak to a fellow who works with EdUbuntu and OLPC since he wanted to write his educational software in Squeak to save space on the laptop. His kids were there and really got into it after some of the capabilities were made more obvious.
I got some great feedback, and it's always cool for me to show off Squeak games (the RTS that Eddie Cottongim made years ago is perhaps the best of the non-EToys style) and coding to kids, since I don't regularly work with them and most developers are such sticks-in-the- mud about the weird UI.
Anyway, it occurred to me that just a little bit of effort in the UI design department would quickly improve Squeak's draw. Since I know the SqueakMap package loader and it's Squeak's gateway to the community for newbies as they start up Squeak (or should be), I started there.
I've added some buttons to a toolbar at the top and stole MessageNames' search bar to be more clear about the UI.
Current Screenshot (doesn't reflect help-balloon cleanups): http://briantrice.com/Images/Squeak/SMLoader1.jpeg With a filter on: http://briantrice.com/Images/Squeak/SMLoader1Filtered.jpeg
These are only incremental changes, since I don't want to take on a huge-scope project, and don't own this and other packages, so a huge fork would be politically difficult. The best application code I found for Morphic UI was actually the IRC client, although it is a bit spec-heavy.
What do people think of this idea? Particularly the package owners, like Göran, but also whoever is "in charge" of the overall UI face of Squeak (that is, whoever would complain loudest if I wanted to totally re-arrange the world menus).
Also, which projects make this kind of effort obsolete or easier? I know about and like the Services/Commands framework and UI, and Tweak's menubar system, but don't know what they seek to replace. What I don't know is whether there is a simple menu-bar or other such system for easily setting up some shared-styled UI features that get the salient features right in front of the user instead of burying them in context menus.
Okay, I'll stop here for now. Feedback welcome!
-- -Brian http://briantrice.com
I fixed up some of the problems with the previous layout and buttons and added package-specific toolbar buttons: http://briantrice.com/Images/Squeak/SMLoader2.jpeg
Monticello packages (rev. 40 is the relevant one) at: http://briantrice.com/Squeak/
- Installed packages and releases are now marked in bold. - All the main buttons that people need to see are in a button-bar that works. - The code is set up so that if there's a better way (more standard) to construct the toolbar, it will be easy to refactor to that form. - The search bar is obviously marked as such. - Menu entries are more succinct, which means people might actually read them (balloon help explains so I allowed them to be terse).
That seems like it's Good Enough to not put people off.
Comments? Suggestions? Rants?
-- -Brian http://briantrice.com
What about putting the search bar on the top bar, with the buttons? Otherwise it seems pretty good to me!
Romain
Il giorno Nov 22, 2006, alle ore 10:37 AM, Brian Rice ha scritto:
I fixed up some of the problems with the previous layout and buttons and added package-specific toolbar buttons: http://briantrice.com/Images/Squeak/SMLoader2.jpeg
Monticello packages (rev. 40 is the relevant one) at: http://briantrice.com/Squeak/
- Installed packages and releases are now marked in bold.
- All the main buttons that people need to see are in a button-bar
that works.
- The code is set up so that if there's a better way (more
standard) to construct the toolbar, it will be easy to refactor to that form.
- The search bar is obviously marked as such.
- Menu entries are more succinct, which means people might actually
read them (balloon help explains so I allowed them to be terse).
That seems like it's Good Enough to not put people off.
Comments? Suggestions? Rants?
-- -Brian http://briantrice.com
-- Romain Robbes http://romain.robb.es
On Nov 22, 2006, at 6:01 AM, Romain Robbes wrote:
What about putting the search bar on the top bar, with the buttons? Otherwise it seems pretty good to me!
I'm divided on that, since moving the search bar into the top bar would make more room vertically and make the search pane stand out a little more, but it's nice to see it stacked/grouped with the package list, since that's what it's associated with.
For now, I'll leave it where it is, but thanks for the idea.
-- -Brian http://briantrice.com
On Wed, Nov 22, 2006 at 01:37:15AM -0800, Brian Rice wrote:
I fixed up some of the problems with the previous layout and buttons and added package-specific toolbar buttons: http://briantrice.com/Images/Squeak/SMLoader2.jpeg
This looks very nice to me. The buttons give a visual suggestion of what can be done, which seems better for first time users who may not know about hidden menus. A couple of suggestions:
On the filtered view (SMLoader1Filtered.jpeg), I don't see a visual indication of the active filtering criteria. It would be good to show this.
VMMakerTool and the Monticello browsers (and probably others that I'm not thinking of) use a similar button-driven design, but the button shapes and colors seem to vary. A consistent look would be a good thing. Of the UIs that I can think of, I guess that consistency between Monticello and SqueakMap would be the first priority, since they both serve as package loaders from the point of view of user.
Nice work!
Dave
On Nov 22, 2006, at 6:16 AM, David T. Lewis wrote:
On Wed, Nov 22, 2006 at 01:37:15AM -0800, Brian Rice wrote:
I fixed up some of the problems with the previous layout and buttons and added package-specific toolbar buttons: http://briantrice.com/Images/Squeak/SMLoader2.jpeg
This looks very nice to me. The buttons give a visual suggestion of what can be done, which seems better for first time users who may not know about hidden menus. A couple of suggestions:
On the filtered view (SMLoader1Filtered.jpeg), I don't see a visual indication of the active filtering criteria. It would be good to show this.
Hmmm. I'll see what I can do, but that might be more difficult to design and implement.
I actually don't use the filters at all in normal usage, just sticking with the search bar, but that could just be a result of the filter system not being approachable enough. It seems like the kind of thing that StarBrowser-style design was made for, which makes me think a bigger re-design is in order, something I'd like to avoid for now.
VMMakerTool and the Monticello browsers (and probably others that I'm not thinking of) use a similar button-driven design, but the button shapes and colors seem to vary. A consistent look would be a good thing. Of the UIs that I can think of, I guess that consistency between Monticello and SqueakMap would be the first priority, since they both serve as package loaders from the point of view of user.
That's a good point about Monticello, especially since I know it greys out buttons and does other context-sensitive things. I'll try to make it consistent with that, at least.
If I can, I'll make both use ToolBuilder, as suggested by Damien.
-- -Brian http://briantrice.com
On 22-Nov-06, at 1:37 AM, Brian Rice wrote:
I fixed up some of the problems with the previous layout and buttons and added package-specific toolbar buttons: http://briantrice.com/Images/Squeak/SMLoader2.jpeg
I like what I see there. Now if only we could finally sort out the ancient paneColor related problems we were all discussing a while ago so that the package tool obeyed the user's colour settings it would really make an improvement!
Adding the search pane is an excellent move. I could see that being a useful thing to add to some other browsers as well.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim 'Profanity: the universal programming language'
So, I've spent a day with the ToolBuilder. I'm mostly impressed, but there are some issues I'll describe as I explain what I've made.
Rev. 44 in the repository on my site (below) has the new functionality in mostly-working order: http://briantrice.com/Squeak/SMLoader-btr.44.mcz
When the package loads, there are SMLoader and SMLoaderPlus. Both of them in their class-side #initialize methods, check for the presence of ToolBuilder, and SMLoader registers itself if ToolBulder isn't present, and otherwise, SMLoaderPlus registers itself. I think this setup will help publish SMLoader without problems in older Squeak versions. I still need to test this in various ways.
Here is a screenshot of SMLoaderPlus: http://briantrice.com/Images/Squeak/SMLoaderPlus1.jpeg
First, I'll state what I liked about ToolBuilder: it removed some grimy nuts-and-bolts-ness from the application code. I didn't have to specify dependents except by saying which selector a given widget should pay attention to. It was easy to specify buttons and their enabling/disabling. ToolBuilder uses some reasonable defaults for a lot of rendering, so my app looks like other ToolBuilder-using apps without any work, and I don't have to maintain that. And it'll presumably work on Tweak and even MVC if that part gets fixed. Despite the long rambling after this, I endorse ToolBuilder and think it is a good kind of way forward.
The search pane... hm, this gave me some grief. So, in this builder design pattern, you construct a tree of specification objects, pass it to the Builder object and tell it to build: them. This seems rather Java-inspired, but that's beside the point. - One issue I had was trying to recreate the "Search" button that in the original, took the search field as its model and called #accept on it to kick off a search. In a builder interface, you can't do this, because you only get the specification objects and not the output. In fact, you're not supposed to be able to access the output. I thought about work-arounds, but they all involved propagating the search field text to the model on every keystroke and then having the button call a method on the model that uses that, or making a trampoline in the model that the search field would watch... except that it can't do that. So the search button had to go, because ToolBuilder can't do that easily. - The other issue was that I couldn't add a help text to the text field. Why isn't #help an attribute of every specification element type? Who do I bother to get that change made to ToolBuilder? Every object in Squeak needs the ability to explain itself, even if the user turns all of it off (smarter help-showing strategies would probably make it less intrusive). - Trying to layout the search pane *with* the buttons on the top bar was an exercise in frustration. Passing a fractionally-specified frame rectangle seems like it should be straightforward in semantics, but I couldn't get any layout to happen that corresponded with my intuition, and I still haven't figured out how exactly the builder code deals with those frame specifications. So I moved the search field back to the package-list pane. What am I missing?
PluggableTreeSpec is strange. It took me a while to figure out what kind of protocol I had to make to populate it, or what objects had to respond to the selectors (sometimes it's the item wrapper object, other times the model). Perhaps there should be a default protocol, a default name for most selectors and a little documentation saying what those are, and that they're override-able in the constructor or mutator methods. (This idea actually reminds me of Rails, and probably rightly so, as they are both builder patterns.) Particularly odd-seeming is how to update the selection programmatically, as happens when the search pane is used - if I call "self changed: #selectedItemWrapperPath", the selector that the package list is watching, it deadlocks trying to redraw until I hit <alt-period>, so I've left that out and the pane doesn't update on its own.
Finally, there doesn't seem to be any documentation on ToolBuilder that is better than the source of the apps created as examples on SqueakSource (not in the image). I learned the most when I found and loaded them. But the classes all have basic, informative comments.
On Nov 22, 2006, at 1:37 AM, Brian Rice wrote:
I fixed up some of the problems with the previous layout and buttons and added package-specific toolbar buttons: http://briantrice.com/Images/Squeak/SMLoader2.jpeg
Monticello packages (rev. 40 is the relevant one) at: http://briantrice.com/Squeak/
- Installed packages and releases are now marked in bold.
- All the main buttons that people need to see are in a button-bar
that works.
- The code is set up so that if there's a better way (more
standard) to construct the toolbar, it will be easy to refactor to that form.
- The search bar is obviously marked as such.
- Menu entries are more succinct, which means people might actually
read them (balloon help explains so I allowed them to be terse).
That seems like it's Good Enough to not put people off.
Comments? Suggestions? Rants?
-- -Brian http://briantrice.com
Brian Rice water@tunes.org writes:
The search pane... hm, this gave me some grief. So, in this builder design pattern, you construct a tree of specification objects, pass it to the Builder object and tell it to build: them. This seems rather Java-inspired, but that's beside the point.
- One issue I had was trying to recreate the "Search" button that in
the original, took the search field as its model and called #accept on it to kick off a search. In a builder interface, you can't do this, because you only get the specification objects and not the output. In fact, you're not supposed to be able to access the output. I thought about work-arounds, but they all involved propagating the search field text to the model on every keystroke and then having the button call a method on the model that uses that, or making a trampoline in the model that the search field would watch... except that it can't do that.
This I think is a mis-designs in the Smalltalk accept/cancel style. I like accept/cancel in general, *except* that it should really apply at the level of an entire pane [1], not at the level of individual widgets.
When I write Squeak dialogs by hand, the best approach I have found is that each entire pane has a variable with a list of individual widgets that need to be accepted before the pane can do its work. Then, I have a method that iterates through this list and calls acceptIfNeeded (or whatever it is called). Then, I call this method at the beginning of every method that observes the input of those fields.
I implement this pattern repeatedly. It would be a candidate for a framework such as ToolBuilder. (Of course, if ToolBuilder has another approach to this problem, then just as well!)
For your search example, the model object behind the search-area pane--or indeed the search-area pane itsealf--could have a variable pointing back to the text widget where the user types in the search phrase. Then the "search" method would accept this pane before doing the actual search.
By the way, always remember when doing this that an "accept" can fail on invalid input! These methods should return a boolean to indicate whether the accept was successful, and if it was not, the calling method should silently quit.
-Lex
[1] I started to write "window", but really what I write just applies to contiguous portions of a window used for one purpose. There is a contingent of Squeak UI designers who prefer to have big multi-function windows instead of the more classical single-purpose windows.
On Nov 24, 2006, at 1:04 AM, Lex Spoon wrote:
Brian Rice water@tunes.org writes:
The search pane... hm, this gave me some grief. So, in this builder design pattern, you construct a tree of specification objects, pass it to the Builder object and tell it to build: them. This seems rather Java-inspired, but that's beside the point.
- One issue I had was trying to recreate the "Search" button that in
the original, took the search field as its model and called #accept on it to kick off a search. In a builder interface, you can't do this, because you only get the specification objects and not the output. In fact, you're not supposed to be able to access the output. I thought about work-arounds, but they all involved propagating the search field text to the model on every keystroke and then having the button call a method on the model that uses that, or making a trampoline in the model that the search field would watch... except that it can't do that.
This I think is a mis-designs in the Smalltalk accept/cancel style. I like accept/cancel in general, *except* that it should really apply at the level of an entire pane [1], not at the level of individual widgets.
I agree about this perspective.
When I write Squeak dialogs by hand, the best approach I have found is that each entire pane has a variable with a list of individual widgets that need to be accepted before the pane can do its work. Then, I have a method that iterates through this list and calls acceptIfNeeded (or whatever it is called). Then, I call this method at the beginning of every method that observes the input of those fields.
Huh. That does seem like a decent way to get the job done in morphic.
I implement this pattern repeatedly. It would be a candidate for a framework such as ToolBuilder. (Of course, if ToolBuilder has another approach to this problem, then just as well!)
ToolBuilder doesn't do this - it just offers PluggableInputFieldSpec which is just a text input field. The spec could conceivably be extended with an action button and/or title, but it doesn't seem like an easy way to go.
UIManager of course has its dialog protocol, which to me sounds like a ChoiceSpec or something similar, but it's not the same as offering a non-modal open-input choice, or even a modal set of open-(or not)- input choices. Certainly an InputSpec makes sense which would come with its own way to specify validators and sub-validators.
For your search example, the model object behind the search-area pane--or indeed the search-area pane itsealf--could have a variable pointing back to the text widget where the user types in the search phrase. Then the "search" method would accept this pane before doing the actual search.
The problem is that the model (that specifies what it wants built) only holds on to the /spec/ of the widget, not the widget itself. The point of the build process is to decouple that model from the actual structural details of the output, focussing only on the spec. If I reach into the "window" variable output and grab the specific morph in question, I've coupled my app not just to Morphic but to the way in which the ToolBuilder generates Morphic widgets.
But maybe what you're saying is that there needs to be a separate model, perhaps generated for every input field/set. That's interesting, but I don't know what to do with that idea.
By the way, always remember when doing this that an "accept" can fail on invalid input! These methods should return a boolean to indicate whether the accept was successful, and if it was not, the calling method should silently quit.
Yeah, that's a good point.
[1] I started to write "window", but really what I write just applies to contiguous portions of a window used for one purpose. There is a contingent of Squeak UI designers who prefer to have big multi-function windows instead of the more classical single-purpose windows.
There should probably be a term neutral to pane/window which specifies an activity or something. I'll give it some thought.
-- -Brian http://briantrice.com
Brian Rice water@tunes.org writes:
On Nov 24, 2006, at 1:04 AM, Lex Spoon wrote:
When I write Squeak dialogs by hand, the best approach I have found is that each entire pane has a variable with a list of individual widgets that need to be accepted before the pane can do its work. Then, I have a method that iterates through this list and calls acceptIfNeeded (or whatever it is called). Then, I call this method at the beginning of every method that observes the input of those fields.
Huh. That does seem like a decent way to get the job done in morphic.
I implement this pattern repeatedly. It would be a candidate for a framework such as ToolBuilder. (Of course, if ToolBuilder has another approach to this problem, then just as well!)
ToolBuilder doesn't do this - it just offers PluggableInputFieldSpec which is just a text input field. The spec could conceivably be extended with an action button and/or title, but it doesn't seem like an easy way to go.
One approach would be that all button specs optionally trigger this "accept everything" behavior. Buttons that have this option turned on would have an intermediary between the ButtonMorph and the model. This intermediary would accept all fields in the dialog before invoking the user-specified method on the model.
This option could even be turned on by default! It seems like the common case, as well as the safer case.
Certainly an InputSpec makes sense which would come with its own way to specify validators and sub-validators.
Yeah, that's another thing. There are various levels of input validation for dialogs, aren't there? There's syntax of individual inputs (that's not a valid date), things dependent on different fields (your final date is earlier than yoru start date), and the overall final request (you just tried to schedule an appointment at a time that overlaps an existing one). It would be nice to have a generic way to specify these validators.
For your search example, the model object behind the search-area pane--or indeed the search-area pane itsealf--could have a variable pointing back to the text widget where the user types in the search phrase. Then the "search" method would accept this pane before doing the actual search.
The problem is that the model (that specifies what it wants built) only holds on to the /spec/ of the widget, not the widget itself. The point of the build process is to decouple that model from the actual structural details of the output, focussing only on the spec. If I reach into the "window" variable output and grab the specific morph in question, I've coupled my app not just to Morphic but to the way in which the ToolBuilder generates Morphic widgets.
Yes, this coupling is ugly. The intermediary idea I described above seems to avoid this problem, though. The intermediary is part of the UI stuff.
Anyway, one way around this is to make each dialog have its own morph class. Then the dialog object has the pointers to the individual input widgets. This way, you keep the UI and model parts separate.
By the way, thinking of these issues, I hope that ToolBuilder has a way for users to add their own morph classes to a UI. This is an escape hatch for any feature a user wants that is not already covered by the existing system. Even businessy plain-gray UI's will occasionally benefit from a custom widget here and there.
-Lex
I have fixed a bug in the #initialize code which I was somehow missing before. SMLoader seems to load fine now in 3.7-3.9 for me, modulo the oddities that are hopefully soon to be fixed in the use of ImageSegments for SqueakMap's database (you'll see those ByteSymbol/ ByteString/etc. notices). The new revision is 45.
On Nov 23, 2006, at 1:57 PM, Brian Rice wrote:
Rev. 44 in the repository on my site (below) has the new functionality in mostly-working order: http://briantrice.com/Squeak/SMLoader-btr.44.mcz
When the package loads, there are SMLoader and SMLoaderPlus. Both of them in their class-side #initialize methods, check for the presence of ToolBuilder, and SMLoader registers itself if ToolBulder isn't present, and otherwise, SMLoaderPlus registers itself. I think this setup will help publish SMLoader without problems in older Squeak versions. I still need to test this in various ways.
Here is a screenshot of SMLoaderPlus: http://briantrice.com/Images/Squeak/SMLoaderPlus1.jpeg
-- -Brian http://briantrice.com
Brian Rice wrote:
The search pane... hm, this gave me some grief. So, in this builder design pattern, you construct a tree of specification objects, pass it to the Builder object and tell it to build: them. This seems rather Java-inspired, but that's beside the point.
That's a funny comment given how little exposure I had to Java ;-)
- One issue I had was trying to recreate the "Search" button that in the
original, took the search field as its model and called #accept on it to kick off a search. In a builder interface, you can't do this, because you only get the specification objects and not the output. In fact, you're not supposed to be able to access the output. I thought about work-arounds, but they all involved propagating the search field text to the model on every keystroke and then having the button call a method on the model that uses that, or making a trampoline in the model that the search field would watch... except that it can't do that. So the search button had to go, because ToolBuilder can't do that easily.
Yes, that is true.
- The other issue was that I couldn't add a help text to the text field.
Why isn't #help an attribute of every specification element type? Who do I bother to get that change made to ToolBuilder? Every object in Squeak needs the ability to explain itself, even if the user turns all of it off (smarter help-showing strategies would probably make it less intrusive).
The best person to bother would be me. And I agree that text fields should have a #help to be used with it. If you post a fix to mantis just assign it to me.
- Trying to layout the search pane *with* the buttons on the top bar was
an exercise in frustration. Passing a fractionally-specified frame rectangle seems like it should be straightforward in semantics, but I couldn't get any layout to happen that corresponded with my intuition, and I still haven't figured out how exactly the builder code deals with those frame specifications. So I moved the search field back to the package-list pane. What am I missing?
Not sure. The layouts should work just as intuitively as you thought. Can you send an example that didn't work the way you expected it?
PluggableTreeSpec is strange. It took me a while to figure out what kind of protocol I had to make to populate it, or what objects had to respond to the selectors (sometimes it's the item wrapper object, other times the model).
Actually you seem to be misunderstanding something more fundamental about the tree widgets here. From the programmer's point of view there is no "wrapper object" (this is only part of the Morphic implementation but doesn't exist in Tweak for example) and the messages are *always* sent to the model and *always* pass the item along to query for. So, for example, to query the children of an item the tree sends the <getChildren> selector, a one-argument message passing the item alongside it. This avoids the need for any of the wrappers. For example, instead of using a hierarchy of FileDirectoryWrappers you would use something like here:
MyFileList>>getDirectoryRoots "Answer the roots for the directory tree" ^Array with: FileDirectory root
MyFileList>>getChildrenOf: directory "Answer the children of the directory" ^directory directoryNames collect:[:str| directory directoryNamed: str]
MyFileList>>getLabelOf: directory "Answer the label for the directory" ^directory localName
etc. Again, the idea is to abstract away the current (Morphic) implementation of trees and replace them with a set of pluggable selectors that can be used to query for the relevant properties.
Perhaps there should be a default protocol, a default name for most selectors and a little documentation saying what those are, and that they're override-able in the constructor or mutator methods. (This idea actually reminds me of Rails, and probably rightly so, as they are both builder patterns.) Particularly odd-seeming is how to update the selection programmatically, as happens when the search pane is used - if I call "self changed: #selectedItemWrapperPath", the selector that the package list is watching, it deadlocks trying to redraw until I hit <alt-period>, so I've left that out and the pane doesn't update on its own.
Yes, you'd be interfering with the Morphic implementation in this case. What you want to use is <getSelectedPath> selector from the PluggableTreeSpec, e.g.,
MyModel>>getSelectedDirectoryPath "Answer the path for the currently selected directory" ^selectedDirectory pathParts
and then:
MyModel>>changeDirectoryTo: newDirectory selectedDirectory := newDirectory. self changed: #getSelectedDirectoryPath.
Cheers, - Andreas
On Nov 24, 2006, at 1:10 PM, Andreas Raab wrote:
Brian Rice wrote:
So, in this builder design pattern, you construct a tree of specification objects, pass it to the Builder object and tell it to build: them. This seems rather Java-inspired, but that's beside the point.
That's a funny comment given how little exposure I had to Java ;-)
Fair enough. It's just my reaction to "Builder build:" idioms where the Builder class is only there for that build: protocol and is stateless (okay, there's a #parent slot, but that wouldn't conflict with any other protocol except if the name clashed). It's "nouning a verb", if you will. Usually that means it's fine to move the protocol to another class; the concrete one that comes to mind is the UIManager, which notably already also is subclassed for each UI framework. In fact, I'd like to make that a concrete suggestion, tempered by the political expense of getting users of the protocol to switch (but that seems like one class reference per user, so probably not difficult).
To continue this mental experiment a little, I'd suggest: - Moving ToolBuilder protocol and #parent instvar to UIManager (and recurse on subclasses). - Changing users of ToolBuilder to UIManager. - Changing associated SUnit classes to use UIManager in their names. - Renaming ToolBuilderSpec to UISpec.
How does this sound? I can assist if appropriate.
- One issue I had was trying to recreate the "Search" button that
in the original, took the search field as its model and called #accept on it to kick off a search. In a builder interface, you can't do this, because you only get the specification objects and not the output. In fact, you're not supposed to be able to access the output. I thought about work-arounds, but they all involved propagating the search field text to the model on every keystroke and then having the button call a method on the model that uses that, or making a trampoline in the model that the search field would watch... except that it can't do that. So the search button had to go, because ToolBuilder can't do that easily.
Yes, that is true.
Maybe it can be fixed by detecting that a model is a Spec and then remembering that in an IdentityDictionary and then mapping it to the output. I don't know how feasible that is, but it's seems like it'd work reasonably.
- The other issue was that I couldn't add a help text to the text
field. Why isn't #help an attribute of every specification element type? Who do I bother to get that change made to ToolBuilder? Every object in Squeak needs the ability to explain itself, even if the user turns all of it off (smarter help-showing strategies would probably make it less intrusive).
The best person to bother would be me. And I agree that text fields should have a #help to be used with it. If you post a fix to mantis just assign it to me.
I'll do that, probably with enough of a changeset to start from.
- Trying to layout the search pane *with* the buttons on the top
bar was an exercise in frustration. Passing a fractionally- specified frame rectangle seems like it should be straightforward in semantics, but I couldn't get any layout to happen that corresponded with my intuition, and I still haven't figured out how exactly the builder code deals with those frame specifications. So I moved the search field back to the package- list pane. What am I missing?
Not sure. The layouts should work just as intuitively as you thought. Can you send an example that didn't work the way you expected it?
Okay, here's my explanation: - I've got buildSearchPaneWith: to make the search pane, and buildButtonBarWith: that builds the buttons. - Right now, buildWith: calls the search-pane code and composes it with the other top-level panes to make the whole window. - If I move the buildSearchPaneWith: call into buildButtonBarWith: (before the buttons in my case), then frame: calls don't really work as well. - Usually I get a search pane that is much taller than the buttons, and/or the buttons will not clear past the end of the search pane. - When I had an extra search button, the button wouldn't size to fit its own text and I couldn't figure out what frame: call to get it to fit right. - On the other hand, I was putting the search field and button into their own panel and then putting the panel into the button panel, but even without that extra wrapper, I had the problems with the search field on that row.
PluggableTreeSpec is strange. It took me a while to figure out what kind of protocol I had to make to populate it, or what objects had to respond to the selectors (sometimes it's the item wrapper object, other times the model).
Actually you seem to be misunderstanding something more fundamental about the tree widgets here. From the programmer's point of view there is no "wrapper object" (this is only part of the Morphic implementation but doesn't exist in Tweak for example) and the messages are *always* sent to the model and *always* pass the item along to query for. So, for example, to query the children of an item the tree sends the <getChildren> selector, a one-argument message passing the item alongside it. This avoids the need for any of the wrappers. For example, instead of using a hierarchy of FileDirectoryWrappers you would use something like here:
Aha. The old SMLoader code uses wrappers so I just didn't think I needed to disregard them. FileListPlus actually was the code that I was looking at as an example, but I didn't get this.
MyFileList>>getDirectoryRoots "Answer the roots for the directory tree" ^Array with: FileDirectory root
MyFileList>>getChildrenOf: directory "Answer the children of the directory" ^directory directoryNames collect:[:str| directory directoryNamed: str]
MyFileList>>getLabelOf: directory "Answer the label for the directory" ^directory localName
etc. Again, the idea is to abstract away the current (Morphic) implementation of trees and replace them with a set of pluggable selectors that can be used to query for the relevant properties.
Gotcha. Thanks.
Perhaps there should be a default protocol, a default name for most selectors and a little documentation saying what those are, and that they're override-able in the constructor or mutator methods. (This idea actually reminds me of Rails, and probably rightly so, as they are both builder patterns.) Particularly odd- seeming is how to update the selection programmatically, as happens when the search pane is used - if I call "self changed: #selectedItemWrapperPath", the selector that the package list is watching, it deadlocks trying to redraw until I hit <alt-period>, so I've left that out and the pane doesn't update on its own.
Yes, you'd be interfering with the Morphic implementation in this case. What you want to use is <getSelectedPath> selector from the PluggableTreeSpec, e.g.,
MyModel>>getSelectedDirectoryPath "Answer the path for the currently selected directory" ^selectedDirectory pathParts
and then:
MyModel>>changeDirectoryTo: newDirectory selectedDirectory := newDirectory. self changed: #getSelectedDirectoryPath.
Unfortunately, after making this refactoring, I still have the same problem. I'll elaborate: - I removed any uses of ItemWrappers. - I changed the #selectedItemWrapperPath to #selectedItemPath, returning an Array of the same labels used to build the tree to begin with. - There was a variation in terms of bolding some strings, making them Texts, but removing this variation did not change the outcome (requiring a user-halt).
What I'm reading implicitly is that the path needs to be a sequence of the keys used to drill down to the item. That it is not the sequence of objects themselves stored with those keys. It seems to bear out with the usage, but maybe I'm way off base for reasons of misinterpretation.
When I call "self changed: #selectedItemPath", I get a lock-up which I have to halt which puts me in a stack with PluggableTreeMorph>>update: (on #selectedItemPath) and Array>>at:ifAbsent: at the top.
I've made my latest snapshot available at: http://briantrice.com/Squeak/SMLoader-btr.46.mcz
Uncomment line 3 of SMLoaderPlus>>selectedItem: to see the bug. If someone could look at this with some Morphic knowledge, it'd probably be more effective than my analysis.
-- -Brian http://briantrice.com
On Nov 24, 2006, at 4:42 PM, Brian Rice wrote:
On Nov 24, 2006, at 1:10 PM, Andreas Raab wrote:
Brian Rice wrote:
- The other issue was that I couldn't add a help text to the text
field. Why isn't #help an attribute of every specification element type? Who do I bother to get that change made to ToolBuilder? Every object in Squeak needs the ability to explain itself, even if the user turns all of it off (smarter help- showing strategies would probably make it less intrusive).
The best person to bother would be me. And I agree that text fields should have a #help to be used with it. If you post a fix to mantis just assign it to me.
I'll do that, probably with enough of a changeset to start from.
Filed as report #5522 with a changeset including Morphic-rendering support that seems to work. See: http://bugs.impara.de/view.php?id=5522
Thanks!
-- -Brian http://briantrice.com
The error turned out to be that the tree participates in the change- update mechanism which causes infinite recursion. To wit: - Calling #selectedItem: in the model tells the tree that #selectedItemPath has updated - The tree updates its path and then *by design* sends #selectedItem: to the model
In order to avoid a deadlock, a ~~ test in #selectedItem: on the new- and-old values guards against further change-notification.
I am unsure what better pattern to use, since none of the other ToolBuilder-using tools uses another widget to update their trees. Perhaps it is a bug?
In any case, I'll release my fixes shortly.
On Nov 24, 2006, at 4:42 PM, Brian Rice wrote:
Unfortunately, after making this refactoring, I still have the same problem. I'll elaborate:
- I removed any uses of ItemWrappers.
- I changed the #selectedItemWrapperPath to #selectedItemPath,
returning an Array of the same labels used to build the tree to begin with.
- There was a variation in terms of bolding some strings, making
them Texts, but removing this variation did not change the outcome (requiring a user-halt).
What I'm reading implicitly is that the path needs to be a sequence of the keys used to drill down to the item. That it is not the sequence of objects themselves stored with those keys. It seems to bear out with the usage, but maybe I'm way off base for reasons of misinterpretation.
When I call "self changed: #selectedItemPath", I get a lock-up which I have to halt which puts me in a stack with PluggableTreeMorph>>update: (on #selectedItemPath) and Array>>at:ifAbsent: at the top.
-- -Brian http://briantrice.com
On Nov 24, 2006, at 4:42 PM, Brian Rice wrote:
On Nov 24, 2006, at 1:10 PM, Andreas Raab wrote:
- One issue I had was trying to recreate the "Search" button that
in the original, took the search field as its model and called #accept on it to kick off a search. In a builder interface, you can't do this, because you only get the specification objects and not the output. In fact, you're not supposed to be able to access the output. I thought about work-arounds, but they all involved propagating the search field text to the model on every keystroke and then having the button call a method on the model that uses that, or making a trampoline in the model that the search field would watch... except that it can't do that. So the search button had to go, because ToolBuilder can't do that easily.
Yes, that is true.
Maybe it can be fixed by detecting that a model is a Spec and then remembering that in an IdentityDictionary and then mapping it to the output. I don't know how feasible that is, but it's seems like it'd work reasonably.
It's time for me to backtrack again - I was wrong about this in the general sense. There *is* a Dictionary mapping widgets to outputs. More specifically, it maps the names of widgets to outputs, which is good enough, and accessible via widgetAt:(ifAbsent:). I missed this because the abstract ToolBuilder class doesn't have a dictionary instvar, just the protocol, even if all concrete subclasses use it. Also, there is no code making use of this feature, but it works fine.
What has to happen is that the model that calls the builder should retain the builder or at least its widgets dictionary after building. I am writing up a ToolModel class to make an easy way to do this and hopefully a few other things by default.
I am approaching the point where I may be able to write an article about the use of ToolBuilder, probably on Squeak People or I may blog it elsewhere.
-- -Brian http://briantrice.com
Brian Rice wrote:
It's time for me to backtrack again - I was wrong about this in the general sense. There *is* a Dictionary mapping widgets to outputs. More specifically, it maps the names of widgets to outputs, which is good enough, and accessible via widgetAt:(ifAbsent:). I missed this because the abstract ToolBuilder class doesn't have a dictionary instvar, just the protocol, even if all concrete subclasses use it. Also, there is no code making use of this feature, but it works fine.
If you want to push this up to ToolBuilder, feel free to send me a patch. Or even easier, get yourself an account on squeaksource.com and I'll add you as another developer.
I am approaching the point where I may be able to write an article about the use of ToolBuilder, probably on Squeak People or I may blog it elsewhere.
That'd be very nice.
Cheers, - Andreas
On Dec 5, 2006, at 11:22 PM, Andreas Raab wrote:
Brian Rice wrote:
It's time for me to backtrack again - I was wrong about this in the general sense. There *is* a Dictionary mapping widgets to outputs. More specifically, it maps the names of widgets to outputs, which is good enough, and accessible via widgetAt: (ifAbsent:). I missed this because the abstract ToolBuilder class doesn't have a dictionary instvar, just the protocol, even if all concrete subclasses use it. Also, there is no code making use of this feature, but it works fine.
If you want to push this up to ToolBuilder, feel free to send me a patch. Or even easier, get yourself an account on squeaksource.com and I'll add you as another developer.
I'm still working on code for this. My SqueakSource initials are "btr", same as in the image. Thanks!
-- -Brian http://briantrice.com
Brian Rice wrote:
If you want to push this up to ToolBuilder, feel free to send me a patch. Or even easier, get yourself an account on squeaksource.com and I'll add you as another developer.
I'm still working on code for this. My SqueakSource initials are "btr", same as in the image. Thanks!
I just added you to http://www.squeaksource.com/ToolBuilder.html - thanks for the contributions (and keep 'em coming ;-)
Cheers, - Andreas
-----Original Message----- From: Andreas Raab Sent: Wednesday, December 06, 2006 2:23 AM
Brian Rice wrote:
It's time for me to backtrack again - I was wrong about this in the general sense. There *is* a Dictionary mapping widgets to outputs. More specifically, it maps the names of widgets to outputs, which is good enough, and accessible via widgetAt:(ifAbsent:). I missed this because the abstract ToolBuilder class doesn't have a dictionary instvar, just the protocol, even if all concrete subclasses use it. Also, there is no code making use of this feature, but it works fine.
If you want to push this up to ToolBuilder, feel free to send me a patch. Or even easier, get yourself an account on squeaksource.com and I'll add you as another developer.
I am approaching the point where I may be able to write an article about the use of ToolBuilder, probably on Squeak People or I may blog it elsewhere.
How about weeklysqueak.wordpress.com?
Ron Teitelbaum
Hi Brian,
Also, which projects make this kind of effort obsolete or easier? I know about and like the Services/Commands framework and UI, and Tweak's menubar system, but don't know what they seek to replace. What I don't know is whether there is a simple menu-bar or other such system for easily setting up some shared-styled UI features that get the salient features right in front of the user instead of burying them in context menus.
if you are using 3.9, you may want to have a look at TestRunner (included in 3.9). Lukas Renggli wrote it and it is very good I think. It uses the new tool framework, ToolBuilder. I think this is the right new way of implementing tool. It should be compatible with new frameworks.
Tell me when you trust your implementation, I will include it in Squeak-dev image.
Thanks for your work.
Bye
On Nov 22, 2006, at 10:44 AM, Damien Cassou wrote:
Hi Brian,
Also, which projects make this kind of effort obsolete or easier? I know about and like the Services/Commands framework and UI, and Tweak's menubar system, but don't know what they seek to replace. What I don't know is whether there is a simple menu-bar or other such system for easily setting up some shared-styled UI features that get the salient features right in front of the user instead of burying them in context menus.
if you are using 3.9, you may want to have a look at TestRunner (included in 3.9). Lukas Renggli wrote it and it is very good I think. It uses the new tool framework, ToolBuilder. I think this is the right new way of implementing tool. It should be compatible with new frameworks.
Good point. I'll take a look.
Tell me when you trust your implementation, I will include it in Squeak-dev image.
Thanks for your work.
Thanks. This is my first Squeak contribution in a while.
-- -Brian http://briantrice.com
squeak-dev@lists.squeakfoundation.org