zaydek-old / urql-exchange-graphcache Goto Github PK

An exchange for normalized caching support in urql

@urql/exchange-graphcache is a normalized cache exchange for the urql GraphQL client. This is a drop-in replacement for the default cacheExchange that, instead of document caching, caches normalized data by keys and connections between data.

You can also pass your introspected GraphQL schema to the cacheExchange, which enables it to deliver partial results and match fragments deterministically!

urql is already quite a comprehensive GraphQL client. However in several cases it may be desirable to have data update across the entirety of an app when a response updates some known pieces of data. This cache also provides configurable APIs to:

• resolve Query data from the offline cache
• update Query data after mutations/subscriptions responses
• provide optimistic Mutation responses

⚠️ Note: Documentation for some parts of @urql/exchange-graphcache are still being worked on! For help for features requests, please join our Spectrum.

First install @urql/exchange-graphcache alongside urql:

yarn add @urql/exchange-graphcache
# or
npm install --save @urql/exchange-graphcache

You'll then need to add the cacheExchange, that this package exposes, to your urql Client, by replacing the default cache exchange with it:

import { createClient, dedupExchange, fetchExchange } from 'urql';

import { cacheExchange } from '@urql/exchange-graphcache';

const client = createClient({
  url: 'http://localhost:1234/graphql',
  exchanges: [
    dedupExchange,
    // Replace the default cacheExchange with the new one
    cacheExchange({
      /* config */
    }),
    fetchExchange,
  ],
});

• Normalized resolving and updates
• Schema awareness and deterministic fragment matching
• Partial query results when the cache is schema aware
• Customization using custom resolvers, updates, and keying functions
• Optimistic updates
• Basic offline and persistence support
• Advanced fragment and entity invalidation

This cache defaults to delivering safe results by marking results as incomplete when any field is missing, triggering a network-only operation (a request), when it encounters uncached fields.

Furthermore there's one case in caching where only having the __typename field leads to potentially unsafe behaviour: interfaces. When the cache encounters a fragment that tries to get data for an interface, it can't tell whether the cached type matches the interface. In this case we resort to a heuristic by default. When all fields of the fragment are on the target type, then the fragment matches successfully and we log a warning.

Schema awareness has been introduced to the cache to improve this behaviour. When you pass your API's GraphQL schema to the cache, it becomes able to deliver partial results. When the cache has enough information so that only optional fields in a given query are missing, then it delivers a partial result from the cached data. Subsequently it still issues a network request (like with cache-and-network) to ensure that all information will still be delivered eventually.

With a schema the cache can also match fragments that refer to interfaces deterministically, since it can look at the schema to match fragments against types.

Schema awareness is also an important stepping stone for offline support. Without partial results it becomes difficult to deliver an offline UI safely, when just some bits of information are missing.

Read more about the architecture

You can currently configure:

• resolvers: A nested ['__typename'][fieldName] map to resolve results from cache
• updates: A Mutation/Subscription field map to apply side-effect updates to the cache
• optimistic: A mutation field map to supply optimistic mutation responses
• keys: A __typename map of functions to generate keys with
• schema: An introspected GraphQL schema in JSON format. When it's passed the cache will deliver partial results and enable deterministic fragment matching.

Note that you don't need any of these options to get started

Keys are used when you need a slight alteration to the value of your identifier or when the identifier is a non-traditional property.

Read more

Resolvers are needed when you want to do additional resolving, for example do some custom date formatting.

Read more

The graph cache will automatically handle updates but some things are quite hard to incorporate. Let's say you delete/add an item, it's hard for us to know you wanted to delete or where to add an item in a list.

Read more

Here you can configure optimistic responses, this means that we don't wait for the server to respond but offer the user to instantly replace the data with the variables from the mutation.

Read more

Our way to see what your backend schema looks like, this offers additional functionality.

Read more

Active: Formidable is actively working on this project, and we expect to continue for work for the foreseeable future. Bug reports, feature requests and pull requests are welcome.