Overview
Examples
Screenshots
Comparisons
Applications
Download
Documentation
Tutorials
Bazaar
Status & Roadmap
FAQ
Authors & License
Forums
Funding Ultimate++
Search on this site













SourceForge.net Logo

Ctrl Design Concepts

 

When developing own Ctrl's one often is 'reinventing' the wheel, because there are common patterns to do in your Ctrl that Ultimate++ is probably already providing in Ctrl base class. Thus, knowing the base class and some of its key design concepts can make your life easier and the development of your Ctrl faster, while focusing more on the problem than the methods. As always, the best reference for Ctrl is its source code, which is quite large, thats why I try to summarize some of the useful things you can already use. In any case, look at the virtual functions in Ctrl to see what is meant to be implemented or used by you. I won't cover LogPos related things here, it is covered in another documentation.

 

Generally, a Ctrl in U++ is helper to visualize some kind of data. The data, though, is not static, and can be changed from GUI (point and click) perspective or from API perspective, using manipulating functions. The difference is, that GUI interaction should modify the internal data (or state) of Ctrl AND notify application somehow about change, but modifying it using API should NOT generate change notifications. This is a very important design rule that keeps you away from a lot of head ache from recursive invocations when modifying Ctrl's in API.

 

Ctrl Tree

 

Ultimate++ uses a linked list for all the child Ctrl's that have been Add()ed to it, partaking of its drawing space. The Ctrl does NOT own its children, but simply references them (Ptr<Ctrl>). They should be owned by your application, somewhere in a U++ container, i.e. Array<Label> or they are already made members of your application when using Layout files. If a Ctrl is added to another, it is ensured to be properly removed from its previous parent, thus a Ctrl cant be part of 2 trees.

 

GetData / SetData

 

Most Ctrl's you will ever create will only need one single value to visualize or represent. This is true for EditFields, Buttons, Labels, etc. To be able to Get / Set this single value into/from the Ctrl, U++ uses it's own 'polymorphic' Value class (see another documentation), which enables the Ctrl's to receive and handle intrinsic data types internally through one single interface, relieving you from the conversion pain. That's why exists GetData / SetData pair. it is the main door into your Ctrl. Even more complex Ctrl's like TreeCtrl  use it to provide the currently selected index. Think of your Ctrl, which information it could provide as general through this interface. it makes implicit usage easy, also in terms of notification (see next)

 

WhenAction Callback

 

To notify upper layers of some changes, your Ctrl can use internally (or the user externally) the Action() function, which will call WhenCallback. and provide the feedback  This is the Callback that can be set using '<<= THISBACK()' approach, so using it for your own Ctrl is preferable, since it leads to U++ conform short syntaxes. Be careful to only call Action() inside your code upon graphical user interaction. When modifying your Ctrl from API, it should generate no Action(). More or diverse notifications can be provided in your controls using other global Callbacks (or even Callback1<yourtype> or more), if needed. Use the WhenSomething name convention to reflect Event behavior.

 

Updated(), SetModify(), ResetModify(), ClearModify(), IsModified()

 

Often, the control needs to process or calculate other things based on the change of some data inside the control (like maybe some results, cached values or the like, NOT graphical helper data, this is done using Layout() which is invoked when resizing or opening the Ctrl). Use the Updated() virtual function to realize this, because it can be triggered from 'outside' using the Update() function. It also SetModified()'s your Ctrl, so you can check for it. Often, when data is changed, Ctrl needs to be updated somehow calculating its things and then the user needs to be notified. UpdateAction() does this in one step, calling both. If graphical data needs change as well, UpdateActionRefresh() is the chain to go, which will invoke an additional Paint(). ClearModify() acts recursively on all children too.

 

 

Refresh() strategies

 

Each U++ Ctrl can be scheduled for Refresh() explicitly. This does not always cause a Draw() immediately, i.e. in Win32, the control draw area is marked for repaint to be processed as soon as some time is available (the Message Queue decides and fires WM_PAINT to causes the repaint). Sync() causes a manual repaint immediately This is sometimes handy to display immediate changes while Main Thread, which would draw it, is known to be locked for quite a while. This is used in Progress for example, to let the user know, that work in Main Thread is in progress (and cant repaint). More on this topic can be found in the Source Code documentation and the Manual.

 

But When is the right point to call Refresh()? This depends. Think of your control and determine logically, what is considered representation of your data, and what is only setup or helping parametrization.So changing any of the data that renders invalid any portion of the Ctrl's draw space should trigger a Refresh(). Anything else probably not. Helping point: SetData() is probably changing your data for sure, it should generally call Refresh() after manipulation of the Ctrl's data. OTOH, i.e. changing Style should not immediately Refresh() because some other Settings might be necessary to change as well, and implicit calls to Refresh() would equal to performance pain and be in vain.

 

User interactions from GUI perspective, changing your data, should generally repaint your Ctrl, at least in portions. Changing it from API side, should probably not, except for SetData....maybe. Because your application can call Refresh() anytime by itself, it knows best when and why.

 

This all does apply for all Ctrl's in this world, cases differ. While the more static Ctr's like an EditField don't really care about heavy Refresh()ing (you wouldn't feel it), Ctrl's like a Plot Diagram showing some frequent live Data from some Generator would. Thats why in latter case, it might be useful to NOT Refresh() on data arrival (SetData) or change, and have the API determine the time, when to Refresh(), maybe after having inserted or changed a couple of related Plots.

 

Refreshing a control just enough can sometimes be crucial. Take again the Diagram that plots live data.Refreshing it each and every little time due to small bits of info changes can keep your application busy and poor in response. Invent means of 'collecting' data without posting it to the Ctrl, and once every 200 ms, do your work..and let the Ctrl Refresh(), this still yields a good look and keeps the GUI responsive.

 

 

Last edit by cxl on 10/04/2015. Do you want to contribute?. T++