Datasource Quickthrough

The source for data items that show up as rows in the table.

  • nGrid only works with 1 datasource: PblDataSource (class)
  • The data source is a container of data items.
  • Internally, PblDataSource always use's an array.
  • Array, Promise<Array> and Observable<Array> are also supported, internally transformed into a PblDataSource.
  • nGrid provides a factory function to create instances of PblDataSource which you will probably use 99.99% of the times (manual creation is possible).

Why PblDataSource

It's not crazy to assume that at the core of every grid out there, the low level rendering of rows is done on an array.

nGrid is no different, so why the external API only works with PblDataSource?

The complete answer to this question is complex, in top-level PblDataSource provides:

  • Normalization and consistency.
  • Well defined setup and teardown routines (safe memory management)
  • Proper scheduling & event handling
  • Unified interface to filtering / sorting / pagination
  • Handling of subset view (virtual scrolling)

In other words, it is an interface that simplifies the work with all of the scenarios we are facing.

It makes it easy to:

  • Update the datasource on-demand (active) or from an event (passive, e.g. next page in pagination clicked, sort clicked, etc...)
  • Deffer when the table is initialized (data-size)
  • Handle client or server scenarios differently but using the same interface

And more...

Datasource Factory Definitions

The definitions are based on the scenario we need to support.

There are 2 main scenarios the support all most all use-cases:

  • Client side: The entire datasource is available in one collection
  • Server side: A portion of the datasource is available
While rare, handling other scenario is possible through custom factory implementations.

For all scenarios there is one mandatory requirement - telling the factory how to fetch data.

Other definitions will determine the desired behavior:

  • How changes in filtering, sorting and pagination are handled
  • Does the datasource die with the table or kept alive so it can move between tables?
  • Does the datasource initialize automatically or we want to wait before fetching data for the first time?
The behavior we define supports the scenario we are in and effect how we fetch data.

Client Side

This is the default scenario, we only need define how to fetch data:

const dataSource = createDS<Person>()
  .onTrigger( (event) => [] ) // or call a remote server to get all data...
  .create();

In the example above we return a simple array but Promise<Array> and Observable<Array> are also supported.

Because all of the collection is available the handling of filtering, sorting and pagination is done for us.

Note that filtering and sorting require matching descriptors but the entire streaming of the data is handled for us.

Server Side

In this mode we only have access to subset of the collection, the best example here is paginated collections. Usually we want to handle all 3, filtering, sorting and pagination.

const dsCustomTrigger = createDS<Person>()
  .setCustomTriggers('filter', 'pagination', 'sort')
  .onTrigger( event => /* get data from a remote server using filter,sort and pagination data stored in "event" */ )
  .create();

We use setCustomTriggers() to opt-in for custom handling. When pagination data changes or when a filter or sort descriptor changes the function we provided to onTrigger() gets invoked with the event object PblDataSourceTriggerChangedEvent.

We use the event object to extract the information we need to make the call to the server, this will include (among other data):

  • Pagination data (current page, current items per page, etc...)
  • Current sort descriptor (The active sorted column and the sort direction)
  • Current filter descriptor (the text to filter by and which fields (columns) to apply it on)
The filter descriptor also accepts a predicate function, avoid using predicates in server side mode as it makes no sense.

Shlomi Assaf © 2019. Code licensed under an MIT-style License. Documentation licensed under CC BY 4.0.

Version 1.0.0-rc.20 using angular 8.2.2 and cdk 8.2.3 (hash: c6af3bd)