This commit:
* updates the introduction to include necessary background information
* start with an operation list then move to comparison
* places the setup basic instructions higher up in the flow
* uses npx instead of a global apollo
* adds schema tags to the instructions with an image
* defines more of the current cli's behavior
* shifts describing specific schema changes to general strategies around
proper evolution of a schema
* adds learnings from dogfooding
* ignoring exit code
* simplifies the CircleCI setup
* add rollover recomendation
* Label alternative strategies for maintaining an api and make
non-breaking changes section less opinonated
* ease example into example of field change
* simplify description of behavior between Notice and Error
Wording changes include:
* passive -> active voice
* remove second person
* cutdown extra information
* pull-request => pull request
* cli => CLI
* published => pushed
* Remove the 2nd person language
* remove unused cicleCI comment
* add GraphQL service to glossary
* publish -> register
* add reference to schema tag
We now point users at publishing their schema on every deploy and add a
note that it can be advantageous to publish from master.
By default, Apollo Server and Apollo Client take care of most aspects of client awareness automatically, with the requirements being:
* Using recent versions of Apollo Server and Apollo Client which support client awareness automatically.
* The `name` and `version` be set on the `ApolloClient` constructor.
As currently written, the variables used in this _Advanced setup_ section have caused some to believe that those are the headers that should be used, whereas they were merely suggestions. The _Setup_ section above should still be observed for most cases and I hope the additional clarity in this re-wording will help avoid future confusion.
An "application programming interface" need not be capitalized in its full form, but I believe API should remain capitalized.
With all due respect to SCUBA, I'm not willing to let every form initialism land in the dictionary in lower-case.
Add documentation for new service: check flags:
* validationPeriod
* queryCountThreshold
* queryCountThresholdPercentage
* Remove broken link
* Run prettier
* (op-reg) Clarify that the operation registry is for Enterprise and Team plans.
* (op-reg) Drop the word "grab" and go for something more friendly.
* (op-reg) Add a "What to expect" heads-up on what the installation necessitates.
* op-reg: "should" > "can".
* op-reg: Refer to "Engine" as "Apollo Engine".
* op-reg: Complete sentences with correct words are great.
* Update docs/source/platform/operation-registry.md
I felt like the text saying to include schema stuff "inside the gql function" may not be quite clear enough for folks who haven't seen template strings or perhaps haven't seen template strings used as a function in that style.
* (op-reg) Update for `forbidUnregisteredOperations` predicate function support.
The Apollo Server operation registry plugin — available for Apollo Engine
Team and Enterprise users, now supports passing a predicate function rather
than a boolean on the `forbidUnregisteredOperations` parameter.
This new parameter can be used to selectively enforce the safe-listing
depending on contextual differences — for example, the presence of a header.
This documentation adds the implementation details for that new feature
along with an example of how to utilize it.
* Clarify predicate function with a link.
A predicate function is a function which returns `true` or `false`, which is explained within the same sentence, but a link would be helpful to those who might not understand that terminology. It's worth understanding the term, and a link helps make that easier.
* Address feedback from @peggyrayzis.
Initializers are no longer part of the AC local state API. We're
now recommending that people use `cache.writeData` directly, since
it's already part of the Apollo Cache API and is more flexible.
