However, the only way default to project-wide authors in curvenote.com at the moment is to clear them which results in authors: []

If i take a markdown file from github e.g. https://github.com/Zabamund/T22_getting_started/blob/master/Installation_and_setup.md

and drop it into a content folder on disk (updating the _toc.yml) then when I start curvenote and check the page the rendered markdown does not look anything like the original, and local images are missing.

Add an integration test suite that

• uses a session based on a generates token, similar to API tests
• tests exporting to all formats
• test exporting pre-configured content from the @curvenote/testing project
• ...

See example in this notebook http://localhost:3000/2022-what-is-mcmc/transform22-what-is-mcmc-solved#Dm7pGhk0NH

Curvenote store large pandas tables in files, they are not immediately available over the API, instead, a download link is provided. The cli should be updated to download the full table (HTML) content when required.

• Style/Design Information
  • numbering
  • hide_authors
• Licensing, open_access, etc.
• DOI

These should be able to be propogated from a few different places:

• curvenote.yml
• each folder, maybe in a frontmatter.yml?
• frontmatter in an article or notebook

Other pieces:

• license, add OSI licenses in addition to CC
• split license defaults into code/text

We will probably always have some content that is wider than our central page, like code cells , text outputs, large figures and tables but currently we are not making best use of space in our layout -- how can can we improve this?

An example is the remix docs which are nice and follow the same left-nav | right-outline schema that we have.

On a 4:3 screen, they make good use of all available horizontal space while not feeling cramped.

https://remix.run/docs/en/v1/guides/envvars

on a 16:9 screen, they maintain the 4:3 layout which works nicely. So this simply makes better use of available space.

BUT, other sites more aligned with scientific content take different approaches:

Quarto sticks with a narrow content layout, but their font sizes are much smaller and light, they also allow large wide or content marked as wide to bust out of the margin:

• https://quarto-dev.github.io/quarto-gallery/page-layout/tufte.html
• https://jjallaire.github.io/hopr/
• https://jjallaire.github.io/visualization-curriculum/
• RHS menu collapses when margin content OR wide content is encountered https://recordit.co/7TPKVWqZbR

Distill - their nav areas scroll, which we don't want but they break margins to deal with wide content which is 😍

• https://distill.pub/
• https://distill.pub/2021/understanding-gnns/
• https://distill.pub/2020/grand-tour/

Objectives:

• What small change to basic styling / layout / gutters could be made to better accommodate content?
• Could we create additional CSS classes allowing us to accommodate wide content like quarto / distill, e.g. busting into RHS margin? and hiding RHS menu when it collides?
• Other ideas?

When exporting an article with references, citations are correctly rendered in the md file but no bib file is produced.

For example, when exporting https://curvenote.com/@stevejpurves/geoscience/phase-and-the-hilbert-transform

Cannot find image, when it is a base64 image.

We should write this to the correct file.

During web builds, curvenotejs reconciles links to other articles in a project, converting the oxa: link to a correct relative link for the build.

For some reason, this is not happening for notebooks, but should work.,

An example is here https://stevejpurves.curve.space/la-palma-earthquakes/notebooks but this can be recreated with any curvenote project, by including an article and notebook and in the article adding a cross-reference to the notebook using [[nb:

Current state:

Need to think through how to have these open/close collapsible. Take a look through firebase docs, jupyterbook as some inspiration.

For any page that is a notebook, we should provide a download link so the user can download that in ipynb format.

This will require some work at build time to generate the link appropriately, currently the ipynb files are only available by making a request to api.curvenote.com. e.g. the notebook at https://curvenote.com/@stevejpurves/la-palma-earthquakes/data-screening-pandas is available on the API endpoint https://api.curvenote.com/blocks/1Bk7uPlcMuaTyKEshESj/w7dhgP7jr4IrWGyDQYfZ/versions/10/download

In terms of design/UI we need somewhere to place that on the page in ourheader blocks - some design/thought is needed / need to look around?

Header blocks for notebooks could look like this if author info has been added:

Or like this if it has not:

Or the author information is potentially empty.

right now - and _ are not valid in the #{{ id }} part of the oxa:link.

There are many times where the conversion of our structure to JB drops some articles from the TOC.

We should ensure that no articles/notebooks are lost when we pull content.

We should be able to distribute this with conda, which will lower the barrier for Jupyter/python users, and can be tested across platforms.

See:

• https://conda-forge.org/docs/maintainer/adding_pkgs.html
• https://github.com/conda-forge/typescript-feedstock/blob/main/recipe/meta.yaml
• https://docs.conda.io/projects/conda-build/en/latest/resources/define-metadata.html

Speed today:

• The CLI already sends the version headers
• Minimum version check on the minor.minor version.
• CLI reports back to user that they need to run npm i curvenote@latest

bug in process images is not using the provided convertFn

https://github.com/curvenote/curvenotejs/blob/289625a843aa5f6cae2da382fa5050c396ed8044/src/export/tex/index.ts#L88

Would it be better for the RHS navigation to be just full height, and if longer than the viewport you need to scroll the page?

Currently, there is no engine field set. Which version of node is this compatible with? v14.0?

Increase the tap target size. Likely also need to show the hover tooltip faster.

Using a local github repo, be able to read the paths of folders (md, notebooks) and check the contents on/off to create various sections to show in the web interface.

• large example repo https://github.com/GeoStat-Examples/gstools-transform22-tutorial

• #93
• #97
• #94
• #95
• Get old init & init from curvenote working on the new config
• Add should write project config on add
• Watch new config files and rebuild
• Correctly handle frontmatter from new configs
• Unit tests!!

Fast sharing of a single notebook

when we have a link with underscores, these seem to get escaped in myst

and then those backslashes get escaped (URL encoded) again when curvespace is built leading to a broken link

There is currently a list of domains in the curvenote.yml, this should be respected with you say curvenote deploy!

There should be some handling for non-curve.space domains (simple for now).

Currently during web builds, we do nothing to find and process local images that may be referenced in a markdown cell in a jupyter notebook.

We should start doing that when:

• the image included in available on a url
• the image included is available to us on disk relative to the notebook

Also, images can be referenced:

• in markdown syntax ![curvenote logo](./curvenote.png)
• using an HTML <img /> tag - (see also #55)

All these scenarios should be supported.

Currently, it seems in some cases that the build process correctly constructs links to local images when specified in markdown but the images are never made available in the public folder. e.g. https://transform.softwareunderground.org/2021-practical-python/t21-practical-seismic-with-python

Note: currently when we add notebooks to curvenote.com, images are lost :( we need more work on version control to fix that, so for this issue we have to manually make available and locally referenced image in order to test.

Example:

in: https://climasoma.curve.space/report/eu-policy-instruments-driving-soil-management

In this notebook: https://github.com/cgre-aachen/gempy_workshops/blob/master/workshops/Transform2022/Transform2022-What%20is%20MCMC.ipynb

see the elements in the first block do not render in curvespace

• Add titles to callouts in the app
• align / standardize schema with myst / jupyterbook / sphinx
• round trip nicely from app to myst

a curvenote article with a table containing 2 images:

produces a model like this:

that does not render on a curvespace site

Currently in web builds we do nothing to handle html-in-markdown properly for notebooks, meaning that we end up rendering HTML source as plain text content.

It is common for notebook authors to use HTML in markdown cells to achieve things like this:

example notebook: https://github.com/gimli-org/transform2021/blob/main/1_Creating_a_subsurface_model.ipynb

Which currently renders in curvenote (and curvespace) as:

Objective

Modify notebook processing during the web build process to handle HTML in markdown cells.

The processNotebook function is where the specialised processing for notebooks happens at the moment. https://github.com/curvenote/curvenotejs/blob/feat/web/src/web/cache.ts#L46

When content/markdown cells are encountered we should process these (using schema/mdast) to sanitize (<script> removal) render the HTML appropriately.

Note: rendering images included in HTML, both local and via URL should be handled here 👉🏻 #56

IN this notebook: https://github.com/cgre-aachen/gempy_workshops/blob/master/workshops/Transform2022/Transform2022-What%20is%20MCMC.ipynb

this plain link:

is rendered as plain text

the result is that any local changes to existing content folder are blown away.

this can probably be teamed with supporting curvenote sync pull <folder-name>

Currently in node 16 at least there seems to be some problems with the prosemirror dependencies.

• curvenote init
• more logging on the init pull.
• check node version >=14
• check that node 16 works
• check npm version (only on web install) I think anything <8 will fail? need to test. (@Yxwww says since npm v7)
• ensure oxa IDs are in the repo
• check that init runs without logging in
• Add -v as well as an option

Init

• your.curve.space --> rowanc1.curve.space
• Remove Would you like to add additional projects now?
• Auto pull content
• only say start (not install)

Emergent

• choose a license
• remove extra title for our renderer
• get rid of serve
• we are going to deploy things to this website, please confirm.
• issue #59
• issue #56
• Export to docx / pdf / latex should work

Final Changes

• grey out Import from Local curvenote init option in all cases
• grey out Use Demo content curvenote init option in all cases
• release v1 to npm 🚀

• should be beautiful
• can be in place at the root os a domain/site
• can be in place at the root of a set of nested content
• still looks good without thumbnails on the cards
• looks even better with thumbnails