I'm working on a Backbone.js application with multiple "sections". Each "section" may have multiple routes but share a single collection. Here's a simple example (with sections "a" and "b"): http://jsfiddle.net/scttnlsn/LW4Ny/
In the example, all the collections are fetched when the router is initialized so that they can be shared across multiple routes without the need to re-fetch in every route handler. This seems fine at first but I am wary of continuing this way when the number of shared collections starts to grow. Additionally, it seems silly to be fetching collections for "sections" that may never even be visited by the user- I would much rather load data on demand.
The obvious alternative is to fetch the data in each route handler instead of when the router is initialized. This would mean that only data that is actually needed is fetched, however, it still ends up performing unnecessary fetches when moving between routes in the same "section". There would no longer be any "sharing" of collection data.
What's a good way to handle this situation? I feel like I need to implement some sort of cache-like structure. Are there existing solutions?
Thanks!
-Scott
Assuming I understand the issue... you could skip fetching on initialization (as in the alternative you describe), but also skip instantiation of the collection altogether until needed. Then declare a factory(ish) method:
getOrFetchCollection: function(collectionId) {
var collection = this[collectionId];
if(!collection) {
collection = this[collectionId] = new Backbone.Collection();
collection.url = urlFor(collectionId);
collection.fetch();
}
return collection;
}
Then call it from within each route handler:
a1: function() {
var view = new View({ title: 'a1', collection: this.getOrFetchCollection('a') });
show(view);
}
Related
I am working on a Backbone/Marionette project. This project implements a way to cache data on local memory after loading them from server. Therefore data can be access anytime, anywhere within the project.
This makes me wonder what is the better way to populate data to view in my case:
const ChildView = marionette.View.extend({/*...*/});
const ParentView = marionette.View.extend({
// ...
onRender() {
// 1: pass data to child view from parent view
const childView = new ChildView({
data: this.options.data,
}));
// 2: initialize data when creating new child view
const childView = new ChildView({
data: SomeModel.new({/* some properties */}),
}));
},
// ...
});
new ParentView({
data: SomeModel.new({/* some properties */}),
}).render();
Both methods work correctly. However, the project view structure is pretty deep and complicated so I prefer the second way because with the first one I would need to go up and down a lot to check what data is and where it comes from.
Do you think if there are any possible problems with this method?
I prefer the 1st way, passing data from parent to child, but it depends on what your views are doing.
For me, a big advantage of sharing a data object is that updating it within one view updates it in all other views (this will work if you pass an existing backbone Model, or any object as data). This can save a lot of work... when a user updates their background color (for example), you can update it once in your BackgroundColorChoose view, and know that it is already updated everywhere else that data is in use.
In a sense, it doesn't matter where the data came from, only what it represents (because it can be accessed/modified from within any of your views).
I can imagine scenarios where this approach is not good, but I've found it makes a good baseline to start from (and avoids the need to trust browser-caching)
I seem to often end up in a situation where I am rendering a view, but the Model on which that view depends is not yet loaded. Most often, I have just the model's ID taken from the URL, e.g. for a hypothetical market application, a user lands on the app with that URL:
http://example.org/#/products/product0
In my ProductView, I create a ProductModel and set its id, product0 and then I fetch(). I render once with placeholders, and when the fetch completes, I re-render. But I'm hoping there's a better way.
Waiting for the model to load before rendering anything feels unresponsive. Re-rendering causes flickering, and adding "loading... please wait" or spinners everywhere makes the view templates very complicated (esp. if the model fetch fails because the model doesn't exist, or the user isn't authorized to view the page).
So, what is the proper way to render a view when you don't yet have the model?
Do I need to step away
from hashtag-views and use pushState? Can the server give me a push? I'm all ears.
Loading from an already-loaded page:
I feel there's more you can do when there's already a page loaded as opposed to landing straight on the Product page.
If the app renders a link to a Product page, say by rendering a ProductOrder collection, is there something more that can be done?
<ul id="product-order-list">
<li>Ordered 5 days ago. Product 0 (see details)</li>
<li>Ordered 1 month ago. Product 1 (see details)</li>
</ul>
My natural way to handle this link-to-details-page pattern is to define a route which does something along these lines:
routes: {
'products/:productid': 'showProduct'
...
}
showProduct: function (productid) {
var model = new Product({_id: productid});
var view = new ProductView({model: model});
//just jam it in there -- for brevity
$("#main").html(view.render().el);
}
I tend to then call fetch() inside the view's initialize function, and call this.render() from an this.listenTo('change', ...) event listener. This leads to complicated render() cases, and objects appearing and disappearing from view. For instance, my view template for a Product might reserve some screen real-estate for user comments, but if and only if comments are present/enabled on the product -- and that is generally not known before the model is completely fetched from the server.
Now, where/when is it best to do the fetch?
If I load the model before the page transition, it leads to straightforward view code, but introduces delays perceptible to the user. The user would click on an item in the list, and would have to wait (without the page changing) for the model to be returned. Response times are important, and I haven't done a usability study on this, but I think users are used to see pages change immediately as soon as they click a link.
If I load the model inside the ProductView's initialize, with this.model.fetch() and listen for model events, I am forced to render twice, -- once before with empty placeholders (because otherwise you have to stare at a white page), and once after. If an error occurs during loading, then I have to wipe the view (which appears flickery/glitchy) and show some error.
Is there another option I am not seeing, perhaps involving a transitional loading page that can be reused between views? Or is good practice to always make the first call to render() display some spinners/loading indicators?
Edit: Loading via collection.fetch()
One may suggest that because the items are already part of the collection listed (the collection used to render the list of links), they could be fetched before the link is clicked, with collection.fetch(). If the collection was indeed a collection of Product, then it would be easy to render the product view.
The Collection used to generate the list may not be a ProductCollection however. It may be a ProductOrderCollection or something else that simply has a reference to a product id (or some sufficient amount of product information to render a link to it).
Fetching all Product via a collection.fetch() may also be prohibitive if the Product model is big, esp. in the off-chance that one of the product links gets clicked.
The chicken or the egg? The collection.fetch() approach also doesn't really solve the problem for users that navigate directly to a product page... in this case we still need to render a ProductView page that requires a Product model to be fetched from just an id (or whatever's in the product page URL).
Alright, so in my opinion there's a lot of ways that you can fix this. I'll list all that I've thought of and hopefully one will work with you or at the very minimum it will inspire you to find your optimal solution.
I'm not entirely opposed to T J's answer. If you just go ahead and do a collection.fetch() on all the products when the website is loading (users generally expect there to be some load time involved) then you have all of your data and you can just pass that data round like he mentioned. The only difference between what he's suggesting and what I normally do is that I usually have a reference to app in all my views. So, for example in my initialize function in app.js I'll do something like this.
initialize: function() {
var options = {app: this}
var views = {
productsView: new ProductsView(options)
};
this.collections = {
products: new Products()
}
// This session model is just a sandbox object that I use
// to store information about the user's session. I would
// typically store things like currentlySelectedProductId
// or lastViewedProduct or things like that. Then, I just
// hang it off the app for ease of access.
this.models = {
session: new Session()
}
}
Then in my productsView.js initialize function I would do this:
initialize: function(options) {
this.app = options.app;
this.views = {
headerView: new HeaderView(options),
productsListView: new ProductsListView(options),
footerView: new FooterView(options)
};
}
The subviews that I create in the initialize in productsView.js are arbitrary. I was mostly just trying to demonstrate that I continue to pass that options object to subviews of views as well.
What this does is allows every view, whether it be a top level view or deeply nested subview, every view knows about every other view, and every single view has reference to the application data.
These two code samples also introduce the concept of scoping your functionality as precise as you possibly can. Don't try to have a view that does everything. Pass functionality off to other views so that each view has one specific purpose. This will promote reuse of views as well. Especially complex modals.
Now to get back to the actual topic at hand. If you were going to go ahead and load all of the products up front where should you fetch them? Because like you said you don't want a blank page just sitting there in front of your user. So, my advice would be to trick them. Load as much of your page as you possibly can and only block the part that needs the data from loading. That way to the user the page looks like it's loading while you're actually doing work behind the scenes. If you can trick the user into thinking the page is steadily loading then they are much less likely to get impatient with the page load.
So, referencing the initialize from productsView.js, you could go ahead and let the headerView and footerView render. Then, you could do your fetch in the render of the productsListView.
Now, I know what you're thinking. Have I lost my mind. If you do a fetch in the render function then there's no way that the call will have time to return before we hit the line that actually renders the productsViewList template. Well, luckily there's a couple of ways around that. One way would be to use Promises. However, the way I typically do it is to just use the render function as its own callback. Let me show you.
render: function(everythingLoaded) {
var _this = this;
if(!!everythingLoaded) {
this.$el.html(_.template(this.template(this)));
}
else {
// load a spinner template here if you want a spinner
this.app.collection.products.fetch()
.done(function(data) {
_this.render(true);
})
.fail(function(data) {
console.warn('Error: ' + data.status);
});
}
return this;
}
Now, by structuring our render this way the actual template won't load until the data has fully loaded.
While we have a render function here I want to introduce another concept that I use every where. I call it postRender. This is a function where I execute any code that depends on DOM elements being in place once the template has finished loading. If you were just coding a plain .html page then this is code that traditionally goes in the $(document).ready(function() {});. It may be worth noting that I don't use .html files for my templates. I use embedded javascript files (.ejs). Continuing on, the postRender function is a function that I have basically added to my boiler plate code. So, any time I call render for a view in the code base, I immediately chain postRender onto it. I also use postRender as a call back for itself like I did with the render. So, essentially the previous code example would look something like this in my code base.
render: function(everythingLoaded) {
var _this = this;
if(!!everythingLoaded) {
this.$el.html(_.template(this.template(this)));
}
else {
// load a spinner template here if you want a spinner
this.app.collection.products.fetch()
.done(function(data) {
_this.render(true).postRender(true);
})
.fail(function(data) {
console.warn('Error: ' + data.status);
});
}
return this;
},
postRender: function(everythingLoaded) {
if(!!everythingLoaded) {
// do any DOM manipulation to the products list after
// it's loaded and rendered
}
else {
// put code to start spinner
}
return this;
}
By chaining these functions like this we guarantee that they'll run sequentially.
=========================================================================
So, that's one way to tackle the problem. However, you mentioned that you don't want to necessarily load all of the products up front for fear that the request could take too long.
Side Note: You should really consider taking out any information related to the products call that could cause the call to take a considerable amount of time, and make the larger pieces of information a separate request. I have a feeling that users will be more forgiving about data taking a while to load if you can get them the core information really fast and if the thumbnails related to each product takes a little longer to load it shouldn't be then end of the world. That's just my opinion.
The other way to solve this problem is if you just want to go to a specific product page then just implement the render/postRender pattern that I outlined above on the individual productView. However note that your productView.js will probably have to look something like this:
initialize: function(options) {
this.app = options.app;
this.productId = options.productId;
this.views = {
headerView: new HeaderView(options),
productsListView: new ProductsListView(options),
footerView: new FooterView(options)
};
}
render: function(everythingLoaded) {
var _this = this;
if(!!everythingLoaded) {
this.$el.html(_.template(this.template(this)));
}
else {
// load a spinner template here if you want a spinner
this.app.collection.products.get(this.productId).fetch()
.done(function(data) {
_this.render(true).postRender(true);
})
.fail(function(data) {
console.warn('Error: ' + data.status);
});
}
return this;
},
postRender: function(everythingLoaded) {
if(!!everythingLoaded) {
// do any DOM manipulation to the product after it's
// loaded and rendered
}
else {
// put code to start spinner
}
return this;
}
The only difference here is that the productId was passed along in the options object to the initialize and then that's pulled out and used in the .fetch in the render function.
=========================================================================
In conclusion, I hope this helps. I'm not sure I've answered all of your questions, but I think I made a pretty good pass at them. For the sake of this getting too long I'm going to stop here for now and let you digest this and ask any questions that you have. I imagine I'll probably have to do at least 1 update to this post to further flush it out.
You started saying:
I have a listing of items in one Collection view
So what does a collection have..? Models..!
When you do collection.fetch() you retrieve all the models.
When the user selects an item, just pass the corresponding model to the item view, something like:
this.currentView = new ItemView({
model: this.collection.find(id); // where this points to collection view
// and id is the id of clicked model
});
This way, there there won't be any delay/ improper rendering.
What if your collections end point returns huge volume of data..?
Then implement common practices like pagination, lazy loading etc.
I construct a Product model with the given ID
To me that sounds wrong. If you have a collection of products, you shouldn't be constructing such models manually.
Have the collection fetch your models before rendering the list view. This way all the problem you mentioned can be avoided.
Here's my scenario:
I have a layout template that needs to check to see if a User belongs to at least one Team. If not, then display a div across the entire site. A user can only see the teams they belong to, so I created a simple publication that works: (code samples are CoffeeScript)
Meteor.publish 'teams', ->
return null if !#userId
Teams.find {'members._id': #userId}
This works great and Teams.find().fetch() in console gives expected results.
However, if I put this code in, say, the Template.layout.rendered, it doesn't work.
Template.layout.rendered = ->
teams = Teams.find().fetch()
hasTeams = teams.length > 0
if !hasTeams
...do stuff..
Obviously this doesn't work because the Teams find is async and not loaded when it needs to make the decision. With a normal template / page I would just use the IronRouter waitOn() but what do I do on the layout?
I could do a waitOn in my router, but since the data is "global" and going to be used everywhere, and because a user could deep link into the site all over the place, I don't want to add that waitOn to EVERY single route.
So what is the proper pattern? How do I get the meteor client to load global data and wait for the data before running the route?
More thinking and searching found the answer right here on SO: struggling to wait on subscriptions and multiple subscriptions
I changed my Router.configure to this:
Router.configure
layoutTemplate: 'layout'
waitOn: ->
return [
Meteor.subscribe('teams')
]
Multiple subscriptions can be added to the return array, and I believe it will wait on all of them.
I had a similar issue with iron router subscribing to a chat in two different publications with waitOn.
Meteor.subscribe('chats')
Only when I switched positions of the following publish blocks, would it let me see any data on the route. The following is the correct order for the two.
Meteor.publish("chats", function () {
return Chats.find();
});
Meteor.publish("chats", function(id) {
return Chats.find({eventRoom: id});
});
In the data-driven paradigm of Backbone, the backbone views/routers should subscribe to model changes and act based on the model change events. Following this principle, the application's views/routers can be isolated from each other, which is great.
However, there are a lot of changes in the states of the applications that are not persisted in the models. For example, in a to-do app, there could be buttons that lets you look at tasks that are "completed", "not completed", or "all". This is an application state not persisted in the model. Note that the completion state of any task is persisted, but the current filter in the view is a transient state.
What is a good way to deal with such application state? Using a plain, non-backboned state means that the views/routers cannot listen to the changes in this state, and hence become difficult to code in the data-driven paradigm.
Your buttons filter example can be properly solved using Model events.
I suppose your buttons handlers have access to the tasks Collection. Then filter the collection and trigger events over the selected Models like:
model.trigger( "filter:selected" )
or
model.trigger( "filter:un-selected" )
The ModelView can be listening to these events on its Model and acts accordingly.
This is following your requirements of respecting the not use or "attributes that are not persistent" like selected but I don't have any trauma to use special attributes even if they shouldn't be persistent. So I also suggest to modify the selected attribute of your Models to represent volatile states.
So in your buttons handlers filter the collection and modify the selected attribute in your Models is my preferred solution:
model.set( "selected", true )
You can always override Model.toJSON() to clean up before sync or just leave this special attributes to travel to your server and being ignored there.
Comment got long so I'll produce a second answer to compare.
First, I feel like "completed", "not completed" are totally item model attributes that would be persisted. Or maybe if your items are owned by many users, each with their own "completed" "not completed" states, then the item would have a completedState submodel or something. Point being, while #fguillen produced two possible solutions for you, I also prefer to do it his second way, having models contain the attributes and the button / view doing most of the work.
To me it doesn't make sense for the model to have its own custom event for this. Sounds to me like a filter button would only have to deal with providing the appropriate views. (Which items to show) Thus, I would just make the button element call a function that runs a filter on the collection more or less directly.
event: {
'click filterBtnCompleted':'showCompleted'
},
showCompleted: function(event) {
var completedAry = this.itemCollection.filter(function(item) {
return item.get('completed');
});
// Code empties your current item views and re-renders them with just filtered models
}
I tend to tuck away these kind of convenience filter functions within the collection themselves so I can just call:
this.ItemCollection.getCompleted(); // etc.
Leaving these attributes in your model and ignoring them on your server is fine. Although again, it does sound to me like they would be attributes you want to persist.
One more thing, you said that using plain non-backboned states sacrifices events. (Grin :-) Not so! You can easily extend any object to have Backbone.Event capabilities.
var flamingo = {};
_.extend(flamingo, Backbone.Events);
Now you can have flamingo trigger and listen for events like anything else!
EDIT: to address Router > View data dealings -------------------//
What I do with my router might not be what you do, but I pass my appView into the router as an options. I have a function in appView called showView() that creates subviews. Thus my router has access to the views I'm dealing with pretty much directly.
// Router
initialize: function(options) {
this.appView = options.appView;
}
In our case, it may be the itemsView that will need to be filtered to present the completed items. Note: I also have a showView() function that manages subviews. You might just be working directly with appView in your scenario.
So when a route like /items/#completed is called, I might do something like this.
routes: {
'completed':'completed'
},
completed: {
var itemsView = ItemCollectionView.create({
'collection': // Your collection however you do it
});
this.appView.showView(itemsView);
itemsView.showCompleted(); // Calls the showCompleted() from View example way above
}
Does this help?
In this stackoverflow post i read about filtering backbone collections and using subsets.
One answer (by sled) recommends using backbone.subset.js (usage example).
I could not find any further resources on backbone.subset.js and I failed implementing it into my project.
It seems like backbone.subset.js is the perfect solution for what i'm trying to achieve.
(Having one "parent" collection that holds all models at all times, and depending on user input filtering the relevant models from the parent collection into a backbone.subset collection.)
My "parent" collection, holding all tasks:
var TasksAll = Backbone.Collection.extend({
url: '/tasks', // the REST url to retrieve collection data
model: Task // the models of which the collection consists of
});
var allTasks = new TasksAll();
Now i want to create a subset collection for e.g. tasks where task.status = 0:
var TasksTrash = new Backbone.Subset({
superset: allTasks,
filter: function(Task) {
return Task.isTrash();
}
});
var trashTasks = new TasksTrash();
Whereas inside the Task model, the method "isTrash" returns true if:
this.get('status') == 0
a) Are there any more resources on backbone.subset.js?
b) How do I implement above scenario?
c) Can I pass 'superset' and 'filter' options as params to the Backbone.Subset init function?
d) I looked into the backbone.subset.js code, when I 'reset' my parent Collection my subset Collections should be updated straight away, right?
PS: I'm fairly new to Backbone. Thanks for your help.
Looking at the source for backbone-subset, it looks as though there is a pre-initialization hook which you could utilize in order to make the 'sieve' or filter available as an option or argument:
https://github.com/masylum/Backbone.Subset/blob/master/backbone.subset.js#L50
As for providing parent as an argument, there is an outstanding patch to add that exact functionality:
https://github.com/masylum/Backbone.Subset/pull/5
With it, you can pass in parent as an option, if it is not an option the library will fall back to looking for it on the object Prototype