Contributing Galata (JupyterLab UI Testing & Automation Framework)

[mbektas]

Hi everyone, on June 17^th I had presented Galata at weekly dev meeting and there was very positive feedback from core developers. We as Bloomberg are interested in contributing this project to JupyterLab organization, and continue helping its development. We are ready to switch the repo over. What are the next steps in this process? Please let us know.

A brief overview of Galata:

Galata is a UI testing and automation framework custom built for JupyterLab.

• uses headless Chrome to launch JupyterLab
• uses puppeteer to interact with browser
• comes with a high level API to interact with JupyterLab UI effortlessly
• provides testing, capturing and comparison tools
• comes with a built-in HTML report generator for test results
• comes with a built-in http server to serve test reports
• comes with a command line interface to manage tests, configuration and many other tasks

Main use cases

• Visual regression testing of JupyterLab UI components, notebook widget rendering
• Automating end-to-end UI testing
• Automating notebook execution for output capture & inspect

[choldgraf]

whoah! that's awesome! I tend to use the UI testing framework of "demo it locally and then close my eyes and send it to production what's the worst that could happen?" This sounds more rigorous than that :-)

[blink1073]

Fantastic, thanks @mbektasbbg! The steps are outlined in our Readme

[ellisonbg]

This is amazing work - I am strongly in favor of JupyterLab incorporating this repo

On Thu, Jul 16, 2020 at 10:35 AM Steven Silvester notifications@github.com wrote:

Fantastic, thanks @mbektasbbg https://github.com/mbektasbbg! The steps are outlined in our Readme https://github.com/jupyterlab/team-compass#contributing-extensions-to-the-jupyterlab-github-organization

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/jupyterlab/team-compass/issues/77#issuecomment-659562330, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAGXUATZ5N5IFYFLHL5GBDR3425VANCNFSM4O4SRXEQ .

-- Brian E. Granger

Principal Technical Program Manager, AWS AI Platform (brgrange@amazon.com) On Leave - Professor of Physics and Data Science, Cal Poly @ellisonbg on GitHub

[bollwyvl]

I'd be remiss to not mention robot framework here:

https://robotframework.org

I've been using it extensively on open source and day job stuff. I haven't looked over Galata, but from my experience, some differences from the above bullet points:

• tests are written in plain English-ish (humorous, as it's a very international community)
• works with multiple browsers
• python/selenium powered
• doesn't download random browser binaries from the internet
• does not use nodejs anywhere
• has some of the best reporting I've used
• can do far more than "just" browser testing (I've spun up whole hubs with it under test)

We've been putting some of our goodies here for jupyter-specific stuff:

https://github.com/robots-from-jupyter

[krassowski]

Is there a repo for Galata to learn more? Or is it closed-source at the moment?

[mbektas]

Fantastic, thanks @mbektasbbg! The steps are outlined in our Readme

Thanks @blink1073 , will check it out.

[mbektas]

Is there a repo for Galata to learn more? Or is it closed-source at the moment?

yes @krassowski, it is closed-source at the moment.

[mbektas]

I'd be remiss to not mention robot framework here: https://robotframework.org

@bollwyvl thanks for sharing about robot, I will look into it. I need to learn more about robot to provide proper comparison but Galata's main advantage over it would be the deep and low level interaction with JupyterLab since it is custom-built for JupyterLab.

[SylvainCorlay]

I would love it to be elected as a Jupyter subproject.

[mbektas]

Hi everyone, over the past few months I have made several enhancements and fixes to Galata and got it compatible with JupyterLab 3. @jtpio also made contributions and has been reviewing it. I think it is ready to be moved over to jupyterlab org and published for wider audience and feedback now!

[jtpio]

Thanks @mbektasbbg for pushing on this!

I've checked galata over the past weeks and it looks really great. It would be very useful to make it an official subproject, so JupyterLab itself could use it to do end to end and visual regression testing.

So I think we should move forward with making it open source. Then it will be easier to iterate on it and accept contributions from the community. Maybe there could be a quick demo at the next JupyterLab dev meeting? Or be discussed as part of the 4.0 roadmap.

[jtpio]

As a side note, this could even become a more general tool to script and generate screencasts for the documentation: https://github.com/jupyterlab/jupyterlab/issues/8776#issuecomment-784399896.

[krassowski]

Is it possible to checkout galata now, or is it still private/hidden?

[mbektas]

Is it possible to checkout galata now, or is it still private/hidden?

@krassowski it is private at the moment. I will send you invitation.

[krassowski]

Thanks, got it. From a very quick overview it looks like a lightweight alternative to robotframework approach. I think that both have their advantages. I would be interested to see in action. Definitely worth discussing at a dev meeting.

[mbektas]

Here is a screencast of Galata in action.

[jtpio]

During the JupyterLab dev meeting yesterday, it was decided to wait one more week before moving forward with this. So that anyone would get a chance to check it out if they would like to.

The repo is still private, but don't hesitate to comment here if you would like to get access. Thanks!

[blink1073]

We've met the criteria for adoption, @mbektasbbg please feel free to proceed with the transfer.

[mbektas]

Thanks @blink1073 , I will do!

[jtpio]

Closing since galata has now been moved to the jupyterlab organization: https://github.com/jupyterlab/galata

Thanks all!