I have a GTK widget, in this case, a TreeView. It starts off pretty small and compressed, as there's no text in it besides the columns names. As I add things, it grows horizontally to cover the text and vertically to cover the extra rows. If I then take those away, it retains its expanded size.
It's kind of annoying for your window to always be resizing as you add things. My question is - how can I "pre-"size the widget? Like one way would be to fill it with junk text that I think is the biggest size it can get, and then remove the text, but that won't look very pretty. Is there a better way?
Simplest answer: treeview.set_size_request(width, height) (but then your tree view won't grow when it needs to afterwards.)
Is your tree view in a box? Have you tried packing it with expand=True and fill=True?
window.set_default_size(width, height) on your whole gtk.Window is the best solution, because that "pre-"sizes the window as you say. It can still grow, or be resized smaller by the user.
Related
I have a number of UI elements like panels, edit fields, buttons, labels etc. so panels contain other panels which contain input fields, editors and so on. Most of the elements are editable and/or resizable which means whenever I change anything, a lot of adjacent UI elements should change their width, height and x/y position on the pane. It works fine with a small number of elements but incredibly slow when the number of elements is thousands.
Is there a fast layout algorithm which can be used in this case? Note that I cannot use any existing layout managers and should come up with my own implementation.
I'd suggest taking a leaf out of the Android playbook and have a larger 'grid' and keep everything sized in modular multiples - this avoids you needing to solve the knapsack problem everytime!
For example, instead of having a button with an width of 80 and a height of 40 you store this as metadata as {2:1} (assuming your layout grid is 40^40 squares).
This way if you have a work panel with space of, say, {2:12} this could be filled with two objects of size {2:6} or maybe 3 of size {2:4}.
It's pretty simple to fit-to-max too as any available space can just be scaled up (say you delete a {1:1} item you can just expand the one next to it to take the space etc - you can of course create your own rules around whether objects can scale in single directions etc.
The other advantage of this approach is that you can easily manage different screen sizes and resolutions too while still keeping the same framework and look and feel.
Since Jason Smith didn't recommend Relative layout and said that we should use Absolute layout I have a question how can we deal with RelativeToView concept?
Absolute Layout sets proportional coordinates and sizes of the elements within itself relative to itself not to each other as RelativeLayout. What to do if I need some elements to be relative to each other? Creating additional Grids and StackLayouts? I would rather use RelativeLayout then or I am missing something.
Decided to add a simplest example and consider we are talking ONLY about Relative and Absolute layouts, no Stack, no Grid. I have 2 buttons and I want to place them as shown in the picture
With absolute layout I could define the position of the top button and say it's height 10% of the screen. Now I could shift the bottom button by saying it starts at 11% of the screen. BUT this will change my top button height. If I want my top button to be it's natural "auto" size I cannot do that. So, how can I put my bottom button under top one if I have no idea how much top button occupies on screen? I know how to do it with Relative but how I can do it with Absolute Layout?
It looks like the solution is nest bunch of layouts https://developer.xamarin.com/guides/xamarin-forms/user-interface/layouts/absolute-layout/
Is that the only way? Is that performance still better than Relative layout?
By its nature RelativeLayout is powerful and offers layouting options that no other Layout on its own does. But that power comes at a cost in performance. Resolving the constraint dependencies consistently and obtaining a final layout doesn't come cheaply.
The point is not necessarily that RelativeLayout should never be used, rather that often times other Layouts can do the job, and yes, even 2-3 nested Layouts can be more performant than a single equivalent RelativePanel.
Grid in particular is a powerful option with which similar effects can be achieved by astute use of Auto, Star, and/or absolute-sized rows and columns as appropriate, plus RowSpan and ColumnSpan, plus element margins, etc.
To consider your specific example, I don't know of a way to achieve what you want with an AbsoluteLayout, at least without the added complication of attached properties. But it seems like a natural fit for a Grid with RowSpan="Auto" on the first row. From a diagram alone I can't tell exactly what other constraints you're going for.
My app needs to show several buttons, without overlap, and preferably without scrolling or zooming. They must be big enough to poke with a finger and read the text. Button width depends on its text length, and the height is constant. The screen size is known.
Each button represents a food about which I know some nutritional information. I'll calculate a protein:carb ratio and a fat content, both ranging from 0% to 100%.
I want to put the buttons close to a position that reflects their nutritional content: e.g. protein-rich at the top, carby at the bottom, fatty on the right and lean on the left. So cake would be bottom right and meats would be somewhere on the top edge.
Often, there'll be overlap and I'll have to nudge them away from each other.
The puzzle is to invent an algorithm for that nudging. The desiderata in order of priority are:
1) Readable and pokeable size, no overlap.
2) No scrolling or zooming required, although it'll happen when there are so many buttons that they could never fit on the screen even if we didn't care where they were.
3) Buttons should be close to where the user would look based on knowing the nutritional content of the food.
Incidentally, I'm using JS on a smartphone, not prolog or the like.
(There are some seeming dupes, but no solutions. One is about diagonal stalks, another just advocates throwing it at a game engine, but most are devoid of answers.)
Ther MArVL group at Monash University does work on constraint-based layout work. Some of their software might be applicable to your problem.
What is the best way to prevent overlapping text on coreplot? Especially when some of the slices may be quite small? The overlapping text appears at the bottom, I think that it runs out of space to put it anywhere and thus must pile one label on top of the other.
I could reduce the size of the pie chart, but is there a more elegant way of achieving this?
Mark
Core Plot doesn't do anything about overlapping labels right now. You have to use your knowledge of the data and remove some of the labels that might be too close together.
I have several objects that are static in size that hug the top and bottom of the screen. However, there is a blank area in the middle of the screen that I would like to stretch with orientation change.
I have solved an issue by adding an extra clear view [paddedView] that can be stretched but wanted to know if there was an easier way to do this without the paddedView just with Visual Format Language.
#"V:|-20-[topLabel(40)]-15-[anotherTopLabel(40)]-[paddedView]-[bottomView(73)]|";
Instead of a view used only for padding, you can either specify a lower priority to a distance or specify the constraint as "greater than or equal to".
Since what you want is a flexible space, the last option sounds like the best one:
#"V:|-20-[topLabel(40)]-15-[anotherTopLabel(40)]-(>=15)-[bottomView(73)]|"
The number 15 is of course just an example.