hpi-swa / ipolyglot Goto Github PK

https://gist.github.com/fniephaus/d0e95ff0ac2c8c17870a07f5d1d9f898

We want to check if the kernel setup within GraalVM works; to do that, we connect a Jupyter notebook frontend to the kernel and test common functionality.

Using jp-kernel try running polyglot code in GraalVM.

The script installing the polyglot kernel is still called ijsinstall; let's rename the file and the references to it.

Errors are returned inconsistently depending on the language used in the kernel cell.
Consistent and useful error messages are desired.

Our basic approach to inform to kernel which language a notebook cell is using will be to use magic commands. If time permits we want to be able to detect the language automatically.

For best UX, users should not have to deal with exporting and importing variables explicitly using the polyglot API provided in GraalVM.
Instead, in some way (that has to be determined yet) we want to deliver a way to import and export variables for each notebook cell.

Ideas include:

1. Including a UI cell to select all available variables for importing and exporting next to each cell; see:

2. Introducing a global bindings object that can be accessed from all cells.

3. Introducing several bindings objects (one for each language) that can be accessed from each cell. (This would be more explicit in mentioning the origin of each variable.)

So that we don't need to set flags --polyglot --jvm in submodules

As a basis for our IPolyglot Jupyter kernel we want to use IJavaScript.
As a first step, we want to run the kernel in GraalVM.

It seems to make sense to have #4 done as a prerequisite so that the programming language is known. Then, depending on the programming language, set the character sequence appended to the beginning of a line (e.g. // or #) according to the programming language. This sequence should be added automatically when hitting cmd + / in the Jupyter Notebook. We might be able to set this in the kernelspec?

We've access to a VM now, set up a jupyter notebook server there.

While attempting to utilize ipolyglot via the docker instance, I got this error from Step #2: docker build --tag=ipolyglot .

Error:

npm ERR! code ELIFECYCLE
npm ERR! errno 1
npm ERR! [email protected] postinstall: `patch-package && ./bin/ipolyglotinstall.js --spec-path=full`
npm ERR! Exit status 1
npm ERR!
npm ERR! Failed at the [email protected] postinstall script.

The full output can be found in this gist

I was able to reproduce this three times in a row.

OSX 10.13.6
Docker Desktop: 2.3.0.3
Docker Engine: 19.03.8

in setJupyterInfoAsync you parse version just executing "jupyter --version" and call
context.frontend.version = stdout.toString().trim(); context.frontend.majorVersion = parseInt( context.frontend.version.split(".")[0] );
But output like that cause error
$ jupyter --version
jupyter core : 4.6.1
jupyter-notebook : 6.0.2
qtconsole : 4.6.0
ipython : 7.9.0
ipykernel : 5.1.3
jupyter client : 5.3.4
jupyter lab : 1.2.3
nbconvert : 5.6.1
ipywidgets : 7.5.1
nbformat : 4.4.0
traitlets : 4.3.3

We want to improv devX by providing an inspector view element to observe the state of the bindings variable used to share state between notebook cells.

If we can use JuypterLab instead, we may use this for our planned frontend extensions and meta data passing.

• remove mock data
• fix variable updating

Syntax highlighting is an important aspect of user experience. We want to introduce syntax highlighting in the notebook frontend for supported languages.

Using magics, we want to support interpreting all languages supported by GraalVM.

If this works well, we want to switch to automatic language detection. #3

Currently, the inspector is a simple html list. In order to better inspect element, we want to make it a tree view.

The kernel was installed according to the installation tutorial, but the attempt to run failed