Another simple library for unidirectional dataflow architecture with support for fragmented (partial) data. Inspired by collaboration with sarahhenkens and omab and similar libraries.
The goal is to introduce a more functional programming style architecture by eschewing MVC like pattern and adopting a single data flow pattern revolving around resources (Datasets).
╔════════╗
║ Stores ║
╚════════╝
▲ │
│ ▼
╔═════════╗ ╔══════════╗ ╔═════════════════╗
║ Actions ║──────>║ Datasets ║──────>║ Components ║
╚═════════╝ ╚══════════╝ ╚═════════════════╝
▲ │
└────────────────────────────────────────┘
Mutation of data has to happen through calling the Actions that flow through the stack. Actions exist on Datasets, which describe a sub-set of data that is stored in the Data Stores. This keeps the data structure flat and lets the concerns of data manipulation follow the chain to the Stores. This will remove side effects and exhaustive code that may result from Components handling data each on their own.
By forcing the data-flow to be in a single direction, it will be easier to follow how data changes will affect the whole application depending on what actions have been issued.
Here's a diagram that explains what this actually means and how to implement it:
You can currently install the package as a npm package or bower component.
The following command installs reflux as a npm package:
npm install react-partial-store
Short of the Actions, Datasets, Stores, Components already referenced above; there are a few other terms & concepts to be aware of.
Resource refers to objects stored in a Store. Actions perform upon Resources ultimately even though they follow the stack chain through a Dataset.
The stack refers to the flow of an Action through a Dataset up its parental chain until it reaches a Dataset or Store with no parent. This concept is important because Datasets & Stores are typically built upon each other in a given application.
Referencing the Stack, Resolving is the act of collapsing the Stack into a representation of a given Resource or Action upon a Resource. As the Stack resolves, items at the front of the stack will take higher priority than ones at the end and options for each item either concatenate or override each-other as the stack resolves. When an Action finally gets to the Store there will be a clear representation of how that resource should be accessed.
Stores represent a collection of data of one type (For example projects). A Store can be created by calling RPS.createStore with an optional collection type name and/or options object. If multiple createStore commands use the same collection type they will all point to the same data Store and you may get errors telling you so if you tried to change its configuration while doing so.
var ProjectsStore = RPS.createStore("projects");Above is as dry as it gets. Stores can be further configured with the following options:
| Name | Type | Required | Description |
|---|---|---|---|
| type | string | no | The unique name representing the objects in this Store. When no-name is provided it will be considered an anonymous Store. |
A Dataset is typically created through an existing Store since they represent a virtual sub-set of data by calling store.createDataset with an optional options object. They can also be created by calling RPS.createDataset and then later compounded with a Store.
var ProjectsStore = RPS.createStore("projects");
var ProjectsDataset = ProjectsStore.createDataset({
uri: "/projects"
});If a Store is only used once in the overall scope you can use the shorthand method of creating a Dataset off a Store such as:
var ProjectsDataset = RPS.createStore("projects", {
uri: "/projects"
});Datasets can also be chained as seen below.
var ProjectsDataset = RPS.createStore("projects", {
uri: "/projects"
});
var ProjectDataset = ProjectsDataset.createDataset({
uri: "/:projectId",
paramId: "projectId"
});That means when ProjectDataset resolves it would resolve to a URI of /projects/:projectId.
Datasets can be further configured with the following options:
| Name | Type | Required | Description |
|---|---|---|---|
| partial | string | no | When specified, resources that are fetched through this Dataset are associated with the specified partial name unless over-ridden further down the chain. |
| fragments | array(string) | no | Specified fragments that are used to compose objects from this dataset until the "complete" data can be fetched. |
| paramMap | object | no | A lookup table to map params on the Component with params in the uri. |
| paramId | string | no | An identifier to the Component param that will server as the unique id for this Dataset. |
| uri | string | no | A URI that will be resolved and concatenated with other URI's in the stack. Component params may be referenced in the URI by doing :paramName. |
| actions | object | no | An object depicting accessor names and action values that will be provided to the Component referencing this Dataset. This list of actions will be concatenated with other Actions as the stack is resolved. |
| onlyActions | object | no | An object depicting accessor names and action values that will whitelist any Actions found later in the stack. |
An Action can be created by calling RPS.createAction with a function and/or definition. The return value of a function supplied will be appended to the Stack when the Action resolves. Be aware this won't automatically resolve to an end-point action (update / create / delete). If you want to update (PUT) a resource via a custom Action you will need to hook the predefined Action RPS.actions.update by making use of the .hook(fn) method thats inherit on all Actions.
var ProjectsStore = RPS.createStore("projects");
var ActionArchive = RPS.actions.update.hook({
uri: "/archive"
})
var ProjectsDataset = ProjectsStore.createDataset({
uri: "/projects"
});
var ProjectDataset = ProjectsDataset.createDataset({
uri: "/:projectId",
paramId: "projectId",
actions: {
archive: ActionArchive
}
});Now when using this Dataset in your Component, simply call this.project.archive() to invoke the update (PUT) that will resolve to /projects/:projectId/archive.
Let's see this as a callback intead that appends json into the response:
var ActionArchive = RPS.actions.update.hook(function() {
console.info("archiving project: %i", this.projectId);
return {
archive: true
};
});
When Providing an options object to an Action you can use the following:
| Name | Type | Required | Description |
|---|---|---|---|
| uri | string | no | A URI that will be resolved and concatenated with other URI's in the stack. Component params may be referenced in the URI by doing :paramName. |
Making a React Component work with RPS requires just changing your React.createClass over to RPS.createClass and providing a datasets object denoting accessor names to Dataset objects.
To bring this full circle, lets look at a relatively simple demo:
var React = require("react")
, RPS = require("react-partial-store");
var ProjectsStore = RPS.createStore("projects");
var ProjectsDataset = ProjectsStore.createDataset({
uri: "/projects"
});
var ProjectDataset = ProjectsDataset.createDataset({
uri: "/:projectId",
paramId: "projectId"
});
var ComponentProjectShow = RPS.createClass({
datasets: {
project: ProjectDataset
},
render: function() {
var projectDataset = this.project
, project = projectDataset.data;
if (this.isLoading()) {
return (
<div>Loading Project...</div>
);
}
return (
<div>
<label>Title</label>
<span>{project.title}</span>
<label>Description</label>
<span>{project.description}</span>
<label>Options</label>
<span>
<a href="#" onClick={this._delete}>Delete Project</a>
</span>
</div>
);
},
_delete: function() {
this.project.delete()
.then(function() {
window.location = "/projects"
});
}
});
var ComponentProjectsList = RPS.createClass({
datasets: {
projects: ProjectsDataset
},
render: function() {
var projectsDataset = this.projects
, projects = projectsDataset.data;
if (this.isLoading()) {
return (
<div>Loading Projects...</div>
);
}
var contentProjects = []
for (var i = 0; i < projects.length; i++) {
var project = projects[i];
contentProjects.add((
<li key={project.id}>
<a href={"/projects/" + project.id}>
{project.title}
</a>
</li>
));
}
return (
<ul>
{contentProjects}
</ul>
);
}
});If you are familiar with other Stores in React you may notice that you aren't accessing data through this.props. Instead all dataset accessors are mounted right on the component itself.
You may have also noticed that the .delete action (any action through a React Component) returns a Promise that can be followed for a success/failure. But wait, why is the delete action even available in the first place; we didn't specify any actions option for our Store or Datasets? This is because by default Datasets offer the basic create, update, delete actions. If you want to disable them you can pass onlyActions: {} to your Dataset or Store as a simple empty white-list.
You will also see one of several helper methods for components in use via this.isLoading(), this will check all datasets to see if any are actively waiting for data.
If you want to further modify a Dataset solely for use in the component but don't want to create a new external dataset, you can also pass the datasetConfig object on the Component and then specify the dataset you wish to config along with an options object that accepts all options a Dataset normally would accept.
For example:
var ComponentProjectsList = RPS.createClass({
datasets: {
projects: ProjectsDataset
},
datasetConfig: {
projects: {
onlyActions: {}
}
}
render: function() {
...
}
});While we saw how Datasets can be chained off each other, this hierarchy is often needed to be easily referenced at each point along it. One approach could be to manually build/store references to each Dataset or you can make use of Dataset Subsets.
Subsets allow you to add a Dataset to an existing Dataset as a subset via a literal reference instead of just creating a Dataset that parental references another. As an example:
var Projects = RPS.createStore("projects", {
uri: "/projects"
});
Projects.addDataset("project", {
uri: "/:projectId",
paramId: "projectId"
});
...
var ComponentProjectShow = RPS.createClass({
datasets: {
project: Projects.project
},
render: function() {
...
}
});
var ComponentProjectsList = RPS.createClass({
datasets: {
projects: Projects
},
render: function() {
...
}
});This allows greater Dataset re-usability componed with the need to maintain references to various Datasets across the application.
We mentioned partials earlier, but didn't see them in action. Out of the box RPS will work with non-fragmented resources. But let's add some optimization and fragment some data.
This can be accomplished one of two ways.
From our above example we can rewrite the Datasets and declare them fragmented and associated as a partial.
var ProjectsDataset = ProjectsStore.createDataset({
partial: "minimal",
uri: "/projects"
});
var ProjectDataset = ProjectsDataset.createDataset({
uri: "/:projectId",
paramId: "projectId",
fragments: ["minimal"]
});
We can see that any resources going through our ProjectsDataset will be associated as a partial called minimal. The server can then return a minimalistic represntation of a project for the index view. The other Dataset says it accepts fragments of a given partial to allow a pre-render before the complete data is fetched and rendered completey.
This can be expanded in a React Component as the following:
var ComponentProjectShow = RPS.createClass({
datasets: {
project: ProjectDataset
},
render: function() {
var projectDataset = this.project
, project = projectDataset.data;
if (this.isLoading() && !this.hasData()) {
return (
<div>Loading Project...</div>
);
}
return (
<div>
<label>Title</label>
<span>{project.title}</span>
<label>Description</label>
<span>{project.description}</span>
<label>Options</label>
<span>
<a href="#" onClick={this._delete}>Delete Project</a>
</span>
</div>
);
},
_delete: function() {
this.project.delete()
.then(function() {
window.location = "/projects"
});
}
});
We just added an extra conditional && !this.hasData() (another one of those component helper methods mentioned earlier). So if the user is coming from the index view into the show view seen above, isLoading() will return true because the complete representation of the resource hasn't been loaded yet but hasData() will return true since it contains a fragment of existing data. Any properties on the resource object that are not in the fragment of data will be undefined and can be handled appropriately for a default value.
With this approach you can leave the Datasets without configuration seen above:
var ProjectsDataset = ProjectsStore.createDataset({
uri: "/projects"
});
var ProjectDataset = ProjectsDataset.createDataset({
uri: "/:projectId",
paramId: "projectId"
});
And it will rely on the server denoting any partial existence when data is returned (with a specific json format containting a _partial property for each resource in the json representation; minimal on the index uri /projects and full for the show uri /projects/:projectId).
There are times it's convenient to preload data on an actual page-load. To support this you can make use of the RPS.prefetchCache toolset and directly put any preloaded data by Store type name into cache by using its .setEntries or .setQueries function. That data will be checked before any request is made out to the server.
An example usage can be seen below in an example of a script at the start of the App. Of course this generally is populated from a JSON/Script embed from the html template and can be easily used in those scenarios as well.
An array of objects each with explicit collection type:
RPS.prefetchCache.setEntries([
{
_partial: "minimal",
_type: "projects",
id: 1,
title: "The Big Bad Wolf"
},
...
]);
An array of objects with explicit passed type:
RPS.prefetchCache.setEntries("projects", [
{
_partial: "minimal",
id: 1,
title: "The Big Bad Wolf"
},
...
]);
Another supported feature is allowing one resource request to potentially fill in other resource stores via embedded data in the response.
TODO FINISH/EXAMPLES