feat(travis): upload docs and client to S3

This change uses Travis CI's artifact feature to store build information
in S3. Specifically we'd like a client and a doc build so that we can test
and view changes for a specific build with these artifacts, as the rest of
the build artifacts are Docker images, uploaded to Quay.io and DockerHub.

Author: Matthew

.travis.yml

@@ -34,10 +34,18 @@ install:
   - GLIDE_HOME=/home/travis/.glide make -C client bootstrap
 script:
   - make test
-  - make -C client/ test
-  - make -C docs/ test
+  - make -C client/ build test
+  - make -C docs/ build test
 deploy:
   provider: script
+  # ensure client/doc builds aren't removed
+  # see https://docs.travis-ci.com/user/deployment/#Uploading-Files
+  skip_cleanup: true
   script: _scripts/deploy.sh
   on:
     branch: master
+addons:
+  artifacts:
+    paths:
+      - client/deis
+      - docs/_build

README.md

@@ -37,7 +37,8 @@ $ kubectl get pods --namespace=deis
 Once this is done, you can SSH into the minion running the controller and run the following:
 
 ```
-$ curl -sSL http://deis.io/deis-cli/install.sh | sh
+$ curl -sSL https://s3-us-west-1.amazonaws.com/bacongobbler/bacongobbler/workflow/19/19.1/client/deis
+$ chmod 755 deis
 $ sudo mv deis /bin
 $ kubectl get service --namespace=deis deis-workflow
 $ deis register 10.247.59.157 # or the appropriate CLUSTER_IP
@@ -50,6 +51,12 @@ $ deis pull deis/example-go -a madras-radiator
 Creating build... ..o
 ```
 
+If you want to retrieve the latest client build, check
+[the latest builds on Travis CI](https://travis-ci.org/deis/workflow/builds), notice the last build
+number that went green and use the following URL to retrieve the client build:
+
+    <https://s3-us-west-1.amazonaws.com/get-deis/deis/workflow/$BUILD_NUM/$BUILD_NUM.1/client/deis>
+
 If you want to hack on a new feature, build the deis/workflow image and push it to a Docker
 registry. The `$DEIS_REGISTRY` environment variable must point to a registry accessible to your
 Kubernetes cluster. You may need to configure the Docker engines on your Kubernetes nodes to allow