Anzer platform and language

Both the provided platform and described language implement convenient way for explicit and type-safe composition of cloud functions.

For example the following scheme describes two services and their communication.

package test

type Source = {
    name  :: String
    price :: Float
}
type Prepared = {
    name        :: String
    description :: String
    category    :: Integer
    price       :: Float
}
type Product = {
    id          :: Integer
    name        :: String
    description :: String
    category    :: Integer
    price       :: Float
}
type Error = {
    error :: String
}
github.com/project/prepare[go] :: Source -> Either Prepared Error
github.com/project/save[go] :: Prepared -> Either Product Error
create = prepare >>= save
invoke(create,)

Anzer utility currently supports only Apache OpenWhisk and Golang.

See the documentation for details.

As quick as possible start

Setup

1. Install docker for your system.
2. Download and install wsk utility from github page.
3. Find working Apache OpenWhisk instance.

With IBM CloudFunctions

1. Register at cloud.ibm.com.
2. Follow the instruction at cloud.ibm.com/openwhisk/learn/cli.
3. Check the settings with wsk action list. Also ~/.wskprops must contain something like the following.

APIHOST=eu-gb.functions.cloud.ibm.com
NAMESPACE=_
AUTH=short_token
APIGW_ACCESS_TOKEN=big_token

Download Anzer

Binaries

Download suitable binary from realizes page. For example for linux:

sudo curl -L "https://github.com/tariel-x/anzer/releases/download/latest/linux_amd64_anzer" -o /usr/local/bin/anzer
sudo chmod +x /usr/local/bin/anzer
anzer help

For OS X:

sudo curl -L "https://github.com/tariel-x/anzer/releases/download/latest/darwin_amd64_anzer" -o /usr/local/bin/anzer
sudo chmod +x /usr/local/bin/anzer
anzer help

Install from sources

git clone https://github.com/tariel-x/anzer.git
cd anzer
go install
anzer help

Download example project

git clone https://github.com/tariel-x/anzer-example.git
cd anzer-example

Make copy of the envs.yaml.dist and edit according to your CouchDB settings.

cp envs.yaml.dist envs.yaml
nano envs.yaml
anzer build -p wsk -i etl.anz --envs envs.yaml

The output of utility would be name of functions sequence in OpenWhisk.

Create POST API

wsk -i api create /etl post /guest/etl_sequence --response-type json

Make request like the following:

curl -X POST \
  http://your.whisk.com:9001/api/23bc46b1-71f6-4ed5-8c54-816aa4f8c502/etl \
  -H 'Content-Type: application/json' \
  -d '{
    "name": "Opel",
    "model": "Astra;J",
    "phone": "79095544445",
    "price": 500.50,
    "photos": [
    	"http://origin.org/path/image.jpeg"	
    ],
    "year": 2014
}'

The expected result is:

{
    "car": {
        "brand": "Opel",
        "model": "Astra"
    },
    "err": null,
    "id": 38081
}

Detailed quick start

Detailed quick start.

Roadmap

1. Full support of OR-types and pattern-matching in the scheme.
2. Remake subsystem for building functions.