Skip to content

datatrails/go-datatrails-common-api

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

55 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

go-datatrails-common-api

Common public api definitions for the DataTrails platform

Finding and including proto files for depdendecnies

tools/go.mod is the source of truth for all proto providing dependencies. That file alone specifies both the upstream version we are using and is used, via go install, to make the .proto files available locally

This corresponds to the practice recommended by grpc-gateway and elsewhere

  1. *** ALWAYS *** Use tools/go.mod to specify the dependency version.
  2. Add the package to the go install command in the apis:preflight task
  3. If necessary, add a var for the new path in any-api and then add a reference to that var in the PROTO_INC var.

Following this practice removes the need for dual maintenance of dependency versions in the builder image. It also produces a build cycle that is significantly faster.

Cross repository builds in docker while using go.work to refer to locally modified sources don't work. And this setup is essential for an efficient workflow.

bootstrap proto files

The proto's for protoc itself, the googleapis, and the grpc_health proxy are needed by almost everything and are also don't apear to be compatible with the tools/go.mod approach

For this reason we curl the proto's and make them available in our aggregate proto-includes archive

Workflow for updating common apis

Ensure the go tool chain is setup on your host

It is inordinately cumbersome to set this all up for docker. It can't be achieved without imposing significant inneficiencies in the workflow. For any engineer working on the api's it is reasonable to expect they can install the correct version of go for their host, and likely already have.

Use an avid/src/go.work

First point avid at a clone of this repo so that it's imports will resolve directly against the sources in ./api

Familiarize with the workflow sub tasks

The general use entry points all use Docker and do not support efficiently and iteratively developing avid and other dependents against a local clone of this repo.

For this repo

You can do everything in the builder image using task all. This uses the blessed builder image and requires no host setup other than docker.

If you are iterating it is a lot more efficient to use the sub tasks, because they all run on your host.

task apis:bootstrap -> task:apis:generate -> task:apis:test -> task:apis:export

The source of truth for the host build is .github/workflows/ci.yml as that is how the CI pipeline works. It does not use our builder image.

Command When
task apis:bootstrap once after clone or rebase
task apis:clean:generated once after clone or rebase
task apis:generate every time the .proto changes
task apis:test every time the .proto changes
task apis:export every time the .proto changes

If you want to iterate on just the helper go code and there tests, do one round of

apis:bootstrap -> apis:clean:generated

Then just iterate using task apis:generate and task apis:test

For avid

See the README.md in avid/src/api/README.md

Build one api against locally cloned go-datatrails-common-api

The protos can be included exactly as they are found from a clone of go-datatrails-common-api. Without needing to generate, build or export. Eg,

task apis:assetsv2-api \
    DATATRAILS_COMMON_API="../../go-datatrails-common-api"

It is necessary however to run task apis:bootsrap after cloning go-datatrails-common