The Cro Development Tools

Cro includes tools to help developers work more efficiently. Currently the tools are available through a command line interface; in the future a web interface will be added also. It is entirely possible to use Cro without these tools. They aim to provide some reasonable defaults, but will not be suitable for every project.

Stubbing Services

A new service can be stubbed using the cro stub command. The general usage is:

cro stub <service-type> <service-id> <path> ['links-and-options']

Where

• service-type is the type of service to create

• service-id is the ID of the service (to be used with other cro commands; this will also be used as the service's default descriptive name in .cro.yml)

• path is the location to create the service

• links-and-options specifies links to other services that should be added to the stub, together with options specific to the service type

If the links and options are not specified, then they will be requested interactively. To provide the options, place them in quotes using Raku colonpair-like syntax, where :foo enables an option, :!foo disables an option, and :foo<bar> is the option foo with the value bar. For example:

cro stub http foo services/foo ':!secure :websocket'
cro stub http bar services/bar ':!secure :websocket'

The stubbed services take port and certificate configuration from environment variables, and when there are relations between services their addresses are also injected using environment variables. This is convenient when setting up container deployment.

Links cause the stubbed service to include code that creates some kind of "client" that can communicate with another endpoint. These go in with the options, having the form :link<service-id:endpoint-id>. The service-id is the id field from the target .cro.yml, and endpoint-id is the id field of an entry in the endpoints list of that .cro.yml file.

cro stub http foo services/foo ':link<flash-storage:http>'

HTTP Services

The http service type stubs in a HTTP service, using Cro::HTTP::Router and served by Cro::HTTP::Server. By default, it stubs an HTTPS service that will accept HTTP/1.0, HTTP/1.1 and HTTP/2.0 requests.

cro stub http flashcard-backend backend/flashcards

The following options may be supplied:

• :secure: generates an HTTPS service instead of an HTTP one (:!secure is the default); implies :http1 :http2 by default, using ALPN to negotiate whether to use HTTP/2

• :!http2: generates a service without HTTP 2 support

• :!http1: generates a service without HTTP 1 support

• :websocket: adds a dependency to the Cro::WebSocket module and adds a stub web socket example

Running Services

cro run [<service-id> ...]

The cro run command is used to run services. It automatically sets up file watching and restarts services when there are source changes to the services (with a debounce to handle a stampede of changes, for example due to fetching latest changes of a running service from version control or saving many files in an editor). To run all services (identified by searching for .cro.yml files in the current working directory and its subdirectories), use:

cro run

To run a specific service, write its service-id (which must appear as the id field in a .cro.yml file in the current working directory or one of its subdirectories):

cro run flashcard-backend

It's also possible to list multiple services:

cro run flashcard-backend users frontend

The output of the services will be displayed, prefixed with the service-name. Sending SIGINT (hitting Ctrl+C) will kill all of the services.

Ports are automatically allocated and the environment variable set per the .cro.yml for the service. The host environment variable will be set to localhost by default, however can be specified with the --host option:

cro --host=dev-vm run

Tracing Services

cro trace <service-id-or-filter>

The cro trace command is much like cro run, except it turns on pipeline debugging in the services. This makes it possible to see the traffic that each service is receiving and sending, and how it is being interpreted and affected by middleware.

The amount of output may be slightly overwhelming, so it can be filtered by the message type name. This is done by checking if any name component is, case-insensitively, equal to the filter. Inclusive filters are expressed as :name, and exclusive filters as :!name. For example, to exclude all of the TCP message messages from the trace, do:

cro trace :!tcp

To see only HTTP messages, do:

cro trace :http

To restrict that further to just requests, do:

cro trace :http :request

Anything not starting with a : is taken as a service-name. The order is unimportant, so these are equivalent:

cro trace :http flashcard-backend
cro trace flashcard-backend :http

The --host option may be specified as for cro run:

cro --host=dev-vm trace

Serving Static Content

cro serve <host-port> [<directory>]

Sometimes it is useful to set up a HTTP server to serve some static content. Serve the current directory on port 8080 of localhost with:

cro serve 8080

Or specify a directory to serve:

cro serve 8080 static_content/

An IP address to bind to may also be provided before the port number:

cro serve 192.168.0.1:8080 static_content/

Working with service links

The cro link subcommand is used to manage the links section of .cro.yml files. These describe how one Cro service uses another, resulting in the injection of environment variables specifying the host and port where the service can be found. In production, these would be set by a container engine such as Kubernetes, by some kind of configuration management system, or even just hardcoded into a wrapper script.

To add a service link, use add:

cro link add <from-service-id> <to-service-id> [<to-endpoint-id>]

Where from-service-id is the id of the .cro.yml that whose links should be modified, to-service-id is the id of the .cro.yml of the service that will be consumed, and to-endpoint-id is the id of an endpoint in that service's .cro.yml. This command will, provided there is a link template matching the protocol of the service linked to, produce some stub code that you can paste into your service code at the appropriate place (Cro is not so crazy as to think it can edit your code under you!)

If to-endpoint-id is not specified, and the to-service-id service has only one endpoint, then that one will be used by default. Otherwise, the ambiguity will be whined about.

To regenerate the code for an existing link, do:

cro link code <from-service-id> <to-service-id> [<to-endpoint-id>]

To remove a link, use:

cro link rm <from-service-id> <to-service-id> [<to-endpoint-id>]

Which simply removes the entry from the links section of the .cro.yml that is identified by from-service-id.