jolie / leonardo

The Jolie Web Server

Welcome to the Jolie open source project!

What's Jolie?

Jolie is a service-oriented programming language, designed to offer native abstractions for the development of microservices. See our website for more information and installation instructions.

Have you made something in or for Jolie? Please submit it to our awesome Jolie list!

Documentation 📔

See https://docs.jolie-lang.org/ for documentation on how to program with Jolie.

For documentation on the interpreter, see https://github.com/jolie/jolie/wiki/Interpreter-documentation.

Roadmap 📈

You can check our roadmap to see what's staged for the next release.

Contributing 💻

Jolie is an open source project based on community effort. Contributions and constructive discussions are always welcome, and we encourage you to join us! There are many ways in which you can contribute, including coding, improving our documentation, helping out with newcomers, or even event organisation.

Interested in contributing? Please check our contribution and organisation guide.

All interactions regarding the Jolie open source project are expected to follow the code of conduct.

Our contributors 👍

Thank you to all our contributors!

Code status 🧐

Get in touch 👋

Discussions can take place over different media.

• Our GitHub repository, in the forms of discussion threads, issues, and pull requests.
• E-mail, by sending an e-mail to [email protected]. Anybody can send to the group. You can also register to the e-mail group at this link: https://groups.google.com/g/jolie-devel.
• Chat or Audio/Video calls. See below for a link to our Discord server.

Discord

Join our Discord server.

leonardo's Issues

Connection close behaviour

I have tried to get the jo-demo-chuck example (https://github.com/fmontesi/jo-demo-chuck) running on Heroku. So far so good I nearly managed to get it working, but one showstopper remains to fix.

Heroku requires the port to be forwarded to Leonardo, which I managed on behalf of the CMD line in the Dockerfile:

CMD jolie -C DebugHttp=true -C Location_Leonardo=\"socket://localhost:$PORT/\" main.ol

Then Leonardo is listening correctly but it gets stuck on the first request. It was not that easy to figure out where the problem lied, finally I got around to it by exploiting nc:

CMD while true; do nc -v -l -p $PORT; done

So Heroku's HTTP dispatcher opens and closes client socket connections without sending ANY http content, and this happens even before the first ordinary request (leonardo.log). Exactly this constitutes a corner case where Jolie's Leonardo does not know how to deal with it, so it simply hangs around... 😞

2020-09-27T18:42:00.366200+00:00 app[web.1]: listening on [::]:5015 ...
2020-09-27T18:42:00.534676+00:00 app[web.1]: connect to [::ffff:172.18.183.98]:5015 from 37dc1bf3-2e9a-4504-a08c-0ddd927b8098.int.dyno.rt.heroku.com:27766 ([::ffff:10.38.173.79]:27766)
^^^^^^^^^^^^
^^^^^^^^^^^^^^^^
2020-09-27T18:42:00.535432+00:00 app[web.1]: listening on [::]:5015 ...
2020-09-27T18:42:00.542188+00:00 app[web.1]: connect to [::ffff:172.18.183.98]:5015 from ip-172-18-183-97.eu-west-1.compute.internal:30545 ([::ffff:172.18.183.97]:30545)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^

How could we fix that?

Issue with running Leonardo locally on Windows from Git Clone. (Works great when using Docker)

Hi,

I have an issue with Leonardo, when trying to run from a git clone on my local machine. I can get it working with Docker, just fine with the same simple website.

Is this an issue you have seen before? If so, what is the fix?

For comparison, this is my Docker command. Same file location.

Am I misunderstanding something?

Standardisation of Leonardo

As it stands, right now every developer who wants to set up a website needs to fork Leonardo and hack the code to add her own services to it. Granted, this is very easy and that's why we could live with it until now. However, it generates a lot of code duplication: if we make an improvement to Leonardo, sometimes it has to be replicated manually to existing installations.

I would like Leonardo to be installable through JPM and call it a day, with the possibility of getting updates automatically. This means that we have to refactor all configuration outside of the main file leonardo.ol.

Ideally, the code of leonardo should be able to start and be left untouched by the custom configuration of each user. When we reach this, I would feel confident in releasing a "Leonardo v1.0".

I would start by making a list of what we think should be configurable in Leonardo. For example:

• Virtual servers
• Adding services through aggregation and redirection
• Configuring cache policies
• .. ?

The big question is how do we make all of this configurable without touching the main Jolie code in leonardo.ol. :-) Suggestions are welcome.