#python wheels

1 messages · Page 1 of 1 (latest)

analog granite
#

@red birch hoping this helps a bit.

unknown.png
# 
package rentalsapi

import (
  "dagger.io/dagger"
  "dagger.io/dagger/core"
  "universe.dagger.io/docker"
)

dagger.#Plan & {
  client: filesystem: ".build/": write: contents: actions.buildWheels.sub.output
  _base: core.#Source & {
    path: "."
  }
  actions: {
    makeBuilder: docker.#Pull & {source: "alpine:3"}
    buildWheels: {
      read: core.#ReadFile & {
        input: _base.output
        path:  "./Dockerfile.build"
      }
      run: docker.#Run & {
        input:   makeBuilder.output
        always:  true
        workdir: "/rentals/api/src/Rentals-API"
        command: {
          name: "sh"
          args: ["-c", "mkdir -p .build && echo '\(read.contents)' > .build/requirements.txt && ls -al .build"]
        }
      }
      sub: core.#Subdir & {
        input: actions.buildWheels.run.output.rootfs
        path:  "/rentals/api/src/Rentals-API/.build"
      }
    }
  }
}
#

You'll notice that to write to the client (laptop/CI server/etc) we use client: filesystem: ".build/": write: contents: where the contents are the result of an action: actions.buildWheels.sub.output

#

Also, as you did in your example, you can use core.#Source to get at things in the same directory and lower as your Dagger plan. Good for getting at a Dockerfile for example.

I have a very small one and grab it and echo it out into the .build/requirements.txt for an example since I don't have poetry etc running since I didn't build an image, but cheated and just pulled one (alpine:3).

$ rm -r .build

$ cat Dockerfile.build
FROM alpine
   
$ dagger do buildWheels --log-format plain --log-level debug
...
$ cat .build/requirements.txt
FROM alpine
analog granite
#

@red birch wondering about cli.#Load. Do you need to load the builder image into your local Docker daemon in order to build your wheels? Or could you just build them in your Dagger plan?