Background: I'm currently developing the client side for a web application, using JavaScript, with jQuery and Backbone.js (these are required by the proponent).
This is an application to visualize and edit data, in a graphical mode (through interactive diagrams representing the data, mainly).
Terminology: Said data is under the format of multiple documents, each containing a list of items.
For the purpose of this question, let the items be composed by an identifier, a textual description, and links to items in other documents. Links should be symmetric (i1 -> i2 exists if and only if i2 -> i1 also exists).
The Current Goal: In this phase, the application should be able to read two documents, display both lists side by side, and then draw lines, connecting the items between both documents, according to their links.
These lines should be editable. In other words, the user should be able to create new links, or remove existing ones (reflecting the changes on the item models).
These documents can be somewhat long, say, some dozens of items (maybe a few hundreds, in a realistic scenario). Of course, the page will be scrollable, to allow the user to see everything.
Also, for user convenience, there should be a slider to scale the view (allowing zoom in/out effects, so the user is given a global and a local view, being the latter more adequate for editing and the former for analysis).
Furthermore, the user should be allowed to hide particular items (useful when an item has many links, creating visual rubbish).
What I've managed to do:
Read data and map it to Backbone models and collections;
Display both documents side by side (Backbone views), with item connections;
Allow interactivity on these connections (drag-and-drop to create lines, click to remove), reflecting changes on Backbone models;
Hide particular items;
Scale effects.
I've achieved this using SVG, after coming across jsPlumb.
The Problem at Hands: The application still needs adjustments (emphasis on the scaling effects). Regardless, I found jsPlumb to be comfortable to work with. However, rendering performance seems to be a little lacking, when using the slider (it's possible that the slider steps are too small, thus firing too many events).
The proponent suggested that I try, instead, Sankey diagrams, to represent this kind of data. They also suggested that I try Sankey by tamc, based on Raphaël.js.
Of course, the visual factor is also contributive.
My question(s): Does this library have a good compatibility with Backbone? Possibly, if I use the resulting SVG elements as Backbone views' elements.
Also, does any of the two have a significant rendering performance advantage over the other?
On a final note, are there any other libraries more adequate to this scenario, worth the time of rewritting the application, that I might suggest to the proponents?
The project is going on, and I ended up using Sankey by tamc, with some extra work of my own, to better adapt it to this particular case.
Related
I have a large app with a structure and UI that has been designed to meet the original requirements.
I am now faced with a situation where I have to make an alternative read-only "view" of certain parts of the app for embedding in an iframe (I have no say in this, this is how it has to be).
I'll be referring to these two ways of viewing the app at the "display context".
I am struggling with visualising how to achive this, I can see two solutions both of which have distinct disadvantages:
Have lots of ng-if directives attached to template elements, the ng-if would be bound to the current display context of the app and show/hide elements depending on how it was being used. Even if these were one-time bindings, it would potentially turn the templates into an unreadable mess of nested ng-ifs everywhere
Create entirely separate templates for the two different display contexts. While much cleaner than the first option, it would mean a lot of duplication and maintaining two templates per view whereas previously it would be one.
I should add that the nature of the views is very ng-repeat heavy, a list of data containing categories, each category contains items, each item contains a multitude of data, at each level data may need to be hidden/displayed/manipulated for display depending on the display context. I have made heavy use of directives and components to break things up into logical chunks, however these directives and components are still very much geared towards the original use, rather than the newly required one.
I have also considered creating an entirely new app for this, however I'm not sure this would bring any benefits as I'd be using controllers and injecting services containing tons of stuff that would be never used, I'd also be increasing my duplication problem.
Does anyone have any feeling, suggestions on a good way forward for this as at the moment I'm feeling anything is going to be horrible.
Many thanks
In case a screen of a WPF application contains lots of primitive controls, its rendering becomes sluggish. What are the recommended ways to improve the responsiveness of a WPF application in such a case, apart from adding fewer controls and using more powerful videocard?
Is there a way to somehow use offscreen buffering or something like that?
Our team was faced with problems of rendering performance. In our case we have about 400 transport units and we should render chart of every unit with a lot of details (text labels, special marks, different geometries etc.).
In first our implementations we splitted each chart into primitives and composed whole unit's chart via Binding. It was very sad expirience. UI reaction was extremely slow.
So we decided to create one UI element per each unit, and render chart with DrawingContext. Although this was much better in performance aspect, we spent about one month improving rendering.
Some advices:
Cache everything. Brushes, Colors, Geometries, Formatted Texts, Glyphs. (For example we have two classes: RenderTools and TextCache. Rendering process of each unit addresses to shared instance of both classes. So if two charts have the same text, its preparation is executed just once.)
Freeze Freezable, if you are planning to use it for a long time. Especially geometries. Complex unfreezed geometries execute HitTest extremely slow.
Choose the fastest ways of rendering of each primitive. For example, there is about 6 ways of text rendering, but the fastest is DrawingContext.DrawGlyphs.
Use profiler to discover hot spots. For example, in our project we had geometries cache and rendered appropriate of them on demand. It seemed to be, that no improvements are possible. But one day we thought what if we will render geometries one time and cache ready visuals? In our case such approach happened acceptable. Our unit's chart has just several states. When data of chart is changed, we rebuild DrawingVisual for each state and put them into cache.
Of course, this way needs some investments, it's dull and boring work, but result is awesome.
By the way: when we turned on WPF caching option (you could find link in answers), our app hung up.
I've had the same perf issue with a heavily customized datagrid since one year, and My conclusion is:
there is basically nothing you can do
on your side (without affecting your
app, i.e.: having fewer controls or
using only default styles)
The link mentioned by Jens is great but useless in your case.
The "Optimizing WPF Application Performance" link provided by NVM is almost equally useless in my experience: it just appeals to common sense and I am confident you won't learn anything extraordinary either reading. Except one thing maybe: I must say this link taught me to put as much as I can in my app's resources. Because WPF does not reinstanciate anything you put in resource, it simply reuses the same resource over and over. So put as much as you can in there (styles, brushes, templates, fonts...)
all in all, there is simply no way to make things go faster in WPF just by checking an option or turning off an other. You can just pray MS rework their rendering layer in the near future to optimize it and in the meantime, try to reduce your need for effects, customized controls and so on...
Have a look at the new (.NET 4.0) caching option. (See here.)
I have met a similar problem and want to share my thoughts and founds. The original problem is caused by a virtualized list box that displays about 25 complex controls (a grid with a text block and a few buttons inside displaying some paths )
To research the issue I used the VisualStudio Application Timeline that allows to how much time it takes to render each control and PerfView to find out what actually WPF is doing to render each control.
By default it took about 12ms to render each item. It is rather long if you need to update the list dynamically.
It is difficult to use PerfView to analyse what heppens inside since WPF renders item in the parent-child hierarchy, but I got the common understanding about internall processes.
WPF does following to render each item in the list:
Parse template using XAML reader. As far as I can see the XAML parsing is the biggest issue.
Apply styles
Apply bindings
It does not take a lot of time to apply styles and bindings.
I did following to improve performance:
Each button has its own template and it takes a lot of time to render it. I replaced Buttons with Borders. It takes about 4-5ms to render each item after that.
Move all element settings to styles. About 3ms.
Create a custom item control with a single grid in the template. I create all child elements in code and apply styles using TryFindResources method. About 2ms in the result.
After all these changes, performance looks fine but still most time is spent on loding the ListControl.Item template and the custom control template.
4. The last step: replace a ListControl with Canvas and Scrollbar controls. Now all items are created at runtime and position is calculated manually using the MeasureOverride and ArrangeOverride methods. Now it takes <1ms to render each item from which 0.5ms is spent on TextBlock rendering.
I still use styles and bindings since they do not affect performance a lot when data is changed. You can imagine that this is not a WPF solution. But I fave a few similar lists in the application and it is possible not to use templates at all.
I'm developing a Winforms application which has been running for years with an explorer view (TreeView left, screen right). I means that:
All the screens have an hierarchy organization
All the nodes on TreeView have one and only one screen related.
A screen gets activated when a node on treeview gets selected.
One of the advantages is that the user has an ordered stucture and one of the inconveniencies is that with hundreds of screens the user gets confused.
I see other options: use classical menus, use tabs or a mix of everything.
Any advice for a good way to show a lot of screens to user in a user-friendly way?
Update: I'm changed "hundreds screens" by "a lot of screens". The most important thing is not show all at time but that the user can find what they need easily.
Update2: In this proposal, the user only see one screen at time.
Update3: I'm talking about handling multiple screens not showing multiple screens. No MDI, only one ontime.
I have used other applications similar to this is the past, and the major problem is trying to find the exact screen you want. There are two common solutions to this problem, shortcut codes and favorites menu.
With the shortcut codes, allocate a short code (5 or 6 characters) to each screen. The user then inputs this shortcut code into a text box which will then jump to the correct screen. Users will create their own list of often used codes.
For the favorites menu, allow users the ability to be able to create their own menu list in the structure they want. They will find things easier, if they organize it themselves.
Why do you need to show so many seprate screens at once? Why not just show the screen for the currnetly selected node, why are all needed at once?
If it is all tabular data is is probably too much to be consumed all at once, if it is graphical data, could it not be combined?
There may be a valid reason to show all the data at once or there may not, hard to tell from what is provided in your question. With that said, better to keep it simple than overload the user. MDI apps are never easy to use.
Tabs may work for a small set of items but still is not a good UI for hundreds of items.
If you are only showing one element at a time, out of hundreds possible on the tree nodes, then that is fine. The one screen showing at a time would be contextual to the item selected as the user moves through the nodes. Think of the Outlook approach where what is selected in the left pane is displayed in the right pane in whatever form fits the data being displayed.
Have you considered the Office Ribbon?
The Ribbon gives you a lot of flexibility on how to show and
organize functions and it's highly visual.
Here is a good link about the Ribbon and also here
To use the Ribbon you have to license it from Microsoft. You can do that online.
Providing the user with ketboard shotcuts is usually a good thing too.
I also like to provide the user with an "autocomplete" field on the menu
so that they can can find the function by name (or part of it) and be
able to navigate directly to where they want to go.
I general I find trees to be a bad idea, especially if your "hierarchy" is of a small fixed depth.
If you have a small fixed depth, consider replacing the tree with a list. At the top of the list can be drop-downs for filtering based on the node-level properties. It will use up less screen real-estate because it is vertical-only, with no horizontal component.
Clicking on an item can display it in the view (like currently), but it may be a good idea to allow a user to double-click on more than one item which could launch more windows, or tile with the existing displayed items. (I am assuming that currently, the user only sees a single detailed view at once in any given window.)
Actually, it’s hard to beat a hierarchy for organizing large numbers of items. I wouldn’t favor a classical pulldown menu for vast numbers of windows because it would be even harder to keep track of where you are than in a tree (e.g., a tree lets you look in multiple branches at once). But here’s a few alternatives:
I’m not clear how you ended up with so many windows, but maybe it comes from combinations of classes, views, content, and detail, or maybe it comes from using a task-centered UI structure for something far too complex (I’ve more on that at http://www.zuschlogin.com/?p=3). For complex apps, you want a different primary window for each significant class of data object (e.g., invoices, employees). These are listed on one menu, and typically there’s few enough (15 or less) that it can be single non-cascading pulldown menu. The content of each window is set by a separate menu, perhaps by a menu item that opens a dialog that may include a list box (like an Open dialog) or other controls for querying/searching. The “view” of each window (how the data objects are shown, e.g., table versus form) is set by menu items in the View menu. Details for any given object in a window can be shown in a separate pane within the window in a master-detail relation, essentially turning you data objects into a menu for details. A single window can have multiple detail panes for the user to open and close to select the specific detail to show. Tabs may also be used within a single pane to fit subdivisions of content.
You say it’s not important to show all window options at once, but often showing all options at once makes it easiest for users to find what they need. Maybe you need a “home” window that lists all the other windows in organized, labeled, and separated categories. This is will be easier to use than the tree if your users select a window then stick with it for most of the session. Your tree is better if there's frequently selection of windows throughout the session, owing to the overhead of getting to the home window. If all windows/options don’t fit on a single home window, then show only selected common windows for each category on the home window and provide a button or link to show an exhaustive list.
If you’re talking 100’s of windows, maybe you should have Search, perhaps in addition to a menu-based browse approach to getting to a window.
In any case, providing easy access to the few most commonly used windows is a good idea. Such windows can be explicitly selected by the designer, based on user research, or selected by the the user (favorites), but it also typically works well to make it automatic with an algorithm that uses some combination of frequency and recency of use.
I am developing a windows forms application. It was basically evovled with a mix of BDUP and prototyping.
I have about 1500 lines of code (excluding IDE generated partial class... 1465 to be exact) in the form and the form has 6 tabs (9 subtabs). Does not have more than 10 controls in each form so multiple form solution would be an overkill.
I have a set of entity classes that when serialized give me an XML representation that I store in an XML file. I've encapsulated this with a Repository pattern so I could move the storage to a database in future. The form uses the entity classes (for save/edit) and the Repo factory (add, retrieve and save).
Now, my problem is that much of the 1500 lines of code deals with interaction between UI elements (making a choice in a Combo disables some elements, or displays different items in a grid, handle tab transitions (next/back buttons), loading menus (each distinct item in the xml file repository becomes a menu item), new/edit mode etc (I've three distinct sets of new/edit on the same form).
What would be best approach here to move the element interaction out? Let's say, I may decide to make it web-based UI in future.
More importantly what are the composite refactorings I can apply?
What patterns I should refactor to / towards?
Thanks for helping out.
Note: I'm reading Refactoring to Patterns... Specifically I wanted to have a "howto"/tips on Refactoring to MVC...
Martin Fowler describes different UI architectural patterns in this article. It's fairly long, but well worth reading. You will then be able to determine which patterns and architectural models fit your scenario the best.
At the end of the "Model View Controller" section of that article, Fowler mentions these items, which seem applicable for you:
Make a strong separation between presentation (view & controller) and domain (model) - Separated Presentation.
Divide GUI widgets into a controller (for reacting to user stimulus) and view (for displaying the state of the model). Controller and view should (mostly) not communicate directly but through the model.
Have views (and controllers) observe the model to allow multiple widgets to update without needed to communicate directly - Observer Synchronization.
See the article and its links for more information. Good luck!
You should look into the Model-View-Controller (MVC) design pattern to abstract your business logic from your display logic.
There is also another flavor of the same thing called the Model View Presenter.
Both of these are extensively documented on the net.
Regards,
Joon
Background
WinForms application using NHibernate. Application is in MDI style and each MDI child form opens a new NHibernate session at Load which remains open for the life of the form.
Question
My application is basically an "order management" or "purchasing" system. One particular form uses a lot of "lookup" lists. Like a list of products, a list of vendors, a list of locations, a list of UnitsOfMeasurement, a list of PriceQuotes, etc.
Lots of lists, that all get loaded when the form is constructed.
Problem: I need the lookup lists, but I need the form to be a bit faster to load. The form is taking too long to perform all the lookups. How can I get better performance and keep my lookup lists?
My Thoughts
Can I load the lookup lists once and hold on to them for the life of the application, and periodically check to see if the lists are stale?
Can I load just the text description for the lists, and instead of holding a bunch of IList, IList, etc, I could hold a bunch of IList, and then when I save, perform the Gets against NHibernate to get the real object.
Is there some other approach that I just haven't thought of?
You should definitely cache slowly changing data to improve performance. How often you need to check for stale data depends on the type of data and your business, e.g. units of measure probably doesn't change as frequently as a list of products. You should also provide a method for manually refreshing lists so that the user can refresh them if something appears to be missing.
If you need the business objects in the list in order to perform a database operation, you can call ISession.Lock(obj) to lock the object into the current ISession. One thing to be aware of is that the lock doesn't automatically cascade to child objects: I think there's a mapping setting to do that or you can do it manually.
Are you sending lists of full objects to your UI? I recently worked on an app using DTO's between the data layer and the UI so I'm not sending the full object, just a description and an identifier. That could help you trim out some unneeded data. So basically when the screen loads a service call is made, nhibernate gets all of the objects I want for my list box, then the UI binds to the list. I bound my listbox display member to the description and the value member to the identifier.