Weekly Team Meetings: Aug-Dec 2021

[isabela-pf]

Here is a record of the agenda/minutes for our weekly meeting. Everyone is welcome to join.

Let's avoid using this thread for discussion. If you'd like to discuss something in the minutes, open a separate issue and reference this thread.

(Thanks to @jasongrout for starting the last six-month note issue jupyterlab/retrolab#117 that I copied this intro from.)

Meeting Notes

• 22 December 2021
• 15 December 2021
• 8 December 2021
• 1 December 2021
• 24 November 2021
• 17 November 2021
• 10 November 2021
• 3 November 2021
• 27 October 2021
• 20 October 2021
• 13 October 2021
• 06 October 2021
• 29 September 2021
• 22 September 2021
• 15 September 2021
• 8 September 2021
• 1 September 2021
• 25 August 2021
• 18 August 2021
• 11 August 2021
• 04 August 2021

[isabela-pf]

Originally posted by @afshin

JupyterLab Weekly Call - August 4, 2021

Attendees

Name	Organization	GitHub	Favorite Color
A. T. Darian	Two Sigma	@afshin	pink
William Stein	SageMath/CoCalc	@williamstein	White
Vidar T Fauske	JP Morgan Chase	@vidar	Ultramarine
Steve Silvester	Apple	@blink1073	fuchsia
Loïc Huder	ESRF	@loichuder	midnightblue
Martha Cryan	IBM	@marthacryan	yellow
Isabela Presedo-Floyd	Quansight Labs	@isabela-pf	orange!
Frederic Collonval	QuantStack	@fcollonval
Jessica Xu	Quansight	@jess-x
Nick Bollweg	Georgia Tech	@bollwyvl @nrbgt	#000, also #chucknorris
Alex Bozarth	IBM	@ajbozarth	Black
Piyush Jain	AWS	@3coins	blue
Karla Spuldaro	IBM	@karlaspuldaro	scarlet
Mike	UOXF/own behalf	@krassowski	coffee
Gonzalo	Quansight, inc	@goanpeca	blue
Kevin Bates	IBM	@kevin-bates	Dodger Blue

Agenda

• Frederic
  • Should we update binder link in jupyterlab/README.md? (currently pointing to a specific hash on jupyterlab-demo)
    https://github.com/jupyterlab/jupyterlab
    Opening PR to latest stable version
  • New examples on completer and shared models: https://github.com/jupyterlab/extension-examples/ => thanks a lot to @hbcarlos
  • Meeting about notebook performance just after
    Video call link: https://meet.google.com/qbv-zggh-qfk
    Meeting notes are on Monthly Performance Meetings: 2021 benchmarks#68

• Loic
  • Fix error on malformed URLs when rendering Markdown: Markdown url resolver no longer throws for malformed URLs in isLocal check jupyterlab#10773
  • Button to show/hide hidden files to discuss: Add a menu entry to show/hide hidden files in the filebrowser jupyterlab#10769
• Steve
  • JupyterLab 3.1.2
• Isabela
  • Is there a process for closing long abandoned issues or PRs?
    • Past efforts haven't been recieved well.
    • "Closing as Answered" is hooked up to the bot which auto-closes after some time
    • If it has been fixed, make sure to link it to the PR that fixes it and the version of JupyterLab that is before closing.
    • Related problem: locked issues don't show referenced issues after they are locked. This has resulted in duplicate issues and PRs in the past.
• Nick
  • trove classifiers go/no-go
• Alex
  • Martha and I have started work on the Setting UI in parallel to our proposed donation of the metadata service and it's associated editor from Elyra.
  • Double check on the licensing since Elyra and JupyterLab have different licenses.
• Mike:
  • Just highlighting some cool work that may need more eyes on:
    • Add execution progress indicator
    • Removed debug switch, added bug button state update - adds the pressed state
  • Do we need to allow to disable document-wide history tracking to finish off dealing with Undo removes the partial output of currently running cell; cannot redo changes undone when other cell produces outputs
    • there are additional considerations with document-wide history tracking: accidentally pasted secrets (want to disable), accoutnability in RTC (want to keep)
• Gonzalo
  • Language packs
    • Multiple outputs single release tag
    • Individual feedstocks

Additional Discussion

• Frederic
  • buildutils 3.1 (since #10494) depends on (unofficial) npm-cli-login which depends on (official) npm-registry-client that depends on old ssri
    https://github.com/jupyterlab-contrib/jupyterlab_toastify/security/dependabot/yarn.lock/ssri/open
    info This module exists because "_project_#test-jlab-toastify#@jupyterlab#builder#@jupyterlab#buildutils#npm-cli-login#npm-registry-client" depends on it.
    • Is it this? ssri 6.0.1 vulnerability npm/cli#3064
      This is related to that but it seems that they don't release any more the separate package npm-registry-client
      This one: https://github.com/npm/npm-registry-client/issues
    • We can just remove it and use one of answers from: https://stackoverflow.com/questions/54540096/give-credentials-to-npm-login-command-line
  • Language packages on conda-forge: N feedstocks as pypi or single feedstock?
    If single feedstock this forces to use common version for all packages
    https://gitter.im/conda-forge/regro-cf-autotick-bot?at=6109583a7555e33351f04684
    PR on conda-forge recipes: PR: Add package jupyterlab-language-pack-ar-SA conda-forge/staged-recipes#15771 (comment)
    Prerelease versions are available on Pypi.org

Links

• settings UI

  • example of desired customized settings (crusty PR)
  • additional shortcuts UI implementation

• conda-forge multiple outputs

  • dagster

[fcollonval]

Originally posted by @afshin

JupyterLab Weekly Call, August 11, 2021

Attendees

Name	Affiliation	GitHub	Favorite Street Food
A. T. Darian	Two Sigma	@afshin	tteokbokki? toppoki?
Piyush Jain	AWS	@3coins	छोले-पूरी
Frederic Collonval	QuantStack	@fcollonval	French Belgian fries
Nick Bollweg	Georgia Tech	@bollwyvl @nrbgt	DÖNER
Michał Krassowski	Oxford	@krassowski
William Stein	SageMath, Inc.	@williamstein	navajo tacos
Kevin Bates	IBM	@kevin-bates	carne asada tacos
Pete Blois	Google (Colab)	@blois	chili dogs
Zach Sailer	Apple	@Zsailer	tacos, obviously
Max Klein	JPM	@telamonian	https://www.youtube.com/watch?v=Esi7zmmS-H4
Martha Cryan	IBM	@marthacryan	banh mi
Vidar T Fauske	JPMorgan Chase	@vidartf	tacooo
Alex Bozarth	IBM	@ajbozarth	tacos
Isabela Presedo-Floyd	Quansight Labs	@isabela-pf
Carlos Herrero	QuantStack	@hbcarlos
Jason Grout	Bloomberg	@jasongrout
Jessica Xu	Quansight	@jess-x
Karla Spuldaro	IBM	@karlaspuldaro	Brazilian coxinha

Agenda

• William: I've been making an effort to encourage some developers of other Jupyter frontends to come to this meeting. (Hi Pete!) I hope this is OK. I think it's valuable since it's useful to see how problems have been solved, and get feedback about usage based on experience by hosting providers.

• Frederic: new Galata (WIP - PR)

• Max: what's the current state of the version pinning that jupyter lab build does?

  • recently came up at work: our python jlab pkg is at 3.0.14 while our js bundle is at 3.0.16 after a build
  • is there a potential for this to cause issues? Or has our semver discipline around changes to python<->js interactions been strong?

• Piyush

  • Removed debug switch - PR
    Frederic has some comments to use command interface, want to get feedback from others and make sure this is the right direction to go.
  • Removed toolbar scrollbar - PR. Small change, is there any feedback?

• nick

  • broke trove classifiers
  • will ping the upstream...
    • waiting for Bump trove-classifiers from 2021.7.10 to 2021.8.4 pypi/warehouse#9882
  • related to add trove classifer docs, usage jupyterlab#10731

• Adam

  • Question about sharedPackages.

Additional Discussion

• Remote Code Execution in Jupyter Notebook and JupyterLab, please update to latest patch release and read on to know more.
  https://blog.jupyter.org/cve-2021-32797-and-cve-2021-32798-remote-code-execution-in-jupyterlab-and-jupyter-notebook-a70fae0d3239, Feel free to ask questions, I will keep updating the blogpost. Thanks @google for the disclosure and auditing Jupyter.

• Releases:

  • 3.2 release issue
  • 4.0 release issue

[afshin]

JupyterLab Weekly Call, August 18, 2021

Attendees

Name	Affiliation	GitHub	Favorite Fictional Villain
A. T. Darian	Two Sigma	@afshin	Randall Flagg
Steven Silvester	Apple	@blink1073	Dr Claw
Vidar T Fauske	JPMorgan Chase	@vidartf
Martha Cryan	IBM	@marthacryan
Layne Sadler	Julia Computing	@aiqc	Lich King (wow/lotr)
Pete Blois	Google Colab	@blois	Dr Strangelove
Frederic Collonval	QuantStack	@fcollonval
Alex Bozarth	IBM	@ajbozarth
William Stein	SageMath	@williamstein	Frank Underwood (House of Cards)
Isabela Presedo-Floyd	Quansight Labs	@isabela-pf	Ozu from The Tatami Galaxy
Brian Granger	AWS/Cal Poly	@ellisonbg
Michał Krassowski	U Oxford	@krassowski	sentiment analysis with Vader
Avish Mehta	Student	@avishmehta68710
Jason Grout	Bloomberg	@jasongrout
Luciano Resende	IBM	@lresende
Karla Spuldaro	IBM	@karlaspuldaro

Agenda

• Isabela
  • Jupyter community call is in two weeks! Yay! More info on discourse
  • Quick feedback on how blind and visually impaired students are working around Jupyter inaccessibility. The main feedback is around getting code out of cells and in a place they can accessibly work on it. This add-on was also mentioned.

• Martha

  • Small Settings UI demo!

• Frederic

  • recurrent Jupyter Performance Meeting set up (thanks to Zach)
    It will be every 4 weeks after this meeting - next will be on September 1st 10am PT
    meeting notes - archived on GitHub
    • (william) I want to try to invite some people from collab/nteract -- how? It's particularly useful, since they have experience with fully implementing other approaches, e.g., Monaco editor, react, etc.
  • JupyterLab review
    • Backport of Firefox toolbar style fix #10849
    • Maintenance camelcase #10500
      Needs to look at the CodeQL Regex error.

• Darian

  • Help wanted for this call!
    • I (@fcollonval) could help with that
    • @ajbozarth is willing to help too

• Brian

  • Do we have a POC for the recent CVE?
    • There are POCs in the Security Advisories for those with access (in comments), we can post them to the security mailing list
  • Jupyter Security Group - https://discourse.jupyter.org/t/project-jupyter-security-subproject/10175/8

• Steve

  • Still working on automated releases, slow going. The Check Release builds are currently failing due to a release of jupyter_releaser this morning.
  • JupyterLab v3.1.7 and v4.0.0a4 released this week

• Vidar

  • I saw this PR had some feedback that caused a revert: #9616. Do we need to report notebook version: https://github.com/jupyterlab/jupyterlab/blob/77c49ea615854a8b7d956ccb4127bace3ca213fc/jupyterlab/labapp.py#L727

• Mike

  • Highlighting issue "Cells not rendering or loading and code seems lost in jupyterlab notebook" #10840 - help in narrowing it down/solving very welcome

Additional Discussion

• Layne: If time allows, the "Ju" in Jupyter. Pluto. History?
  • JUlia PYThon R

[afshin]

JupyterLab Weekly Call, August 25, 2021

Attendees

Name	Affiliation	GitHub	Favorite Flower
A. T. Darian	Two Sigma	afshin	crocus sativus (saffron crocus)
Steve Silvester	Apple	blink1073	sunflower (the radioactive ones in TMNT 2)
Frederic Collonval	QuantStack	@fcollonval
Kevin Jahns	Yjs / QuantStack	@dmonad
Pete Blois	Google	@blois	Dahlia
tonyfast	quansight	@tonyfast	top shelf
Nick Bollweg	Georgia Tech	@bollwyvl @nrbgt	morning glory
William Stein	SageMath, Inc (CoCalc)	@williamstein	Iris
Carlos Herrero	QuantStack	@hbcarlos
Theodore Aptekarev	Independent	@piiq	lavender
Jason Grout	Bloomberg	@jasongrout	all? cool flower
Mehmet Bektas	Splunk	@mbektas	Plumeria
Isabela Presedo-Floyd	Quansight Labs	@isabela-pf	osteospermum
Brian Granger	AWS, Cal Poly	@ellisonbg	?
Vidar T Fauske	JPMorgan Chase	@vidartf	orchids
Alex Bozarth	IBM	@ajbozarth
Martha Cryan	IBM	@marthacryan

Agenda

• Frederic:
  • Translations
    • Documentation translation update policy (strings should not change in a patch release)
    • Add CI to check source strings are not change by a PR
  • CSS Icon deprecation PR10908
  • Galata PR10796 in 3.2.x as @jupyterlab/galata @4.0.0

• Steve

  • JupyterLab 3.1.8 and 4.0.0a5
    • The testutils package was not populated due to #10891, hot fix coming in #10916
  • There is now a separate lint check that uses TypeScript, that does not run by default on commit see #10900 - will be much faster and less intrusive
  • Still working on automated release fixes

• Jason

  • Assigning issues - we have a number of people requesting issues be assigned to them in the GitHub interface so that they can work on them. Sometimes these new contributors are optimistic about fixing an issue and a PR does not eventually come, but the issue is not ever unassigned (thus it appears to be "taken" to other new contributors). I would suggest as a policy that either:
    • in general, we not assign an issue to a person (an exception would be assigning a short list of issues to people in the final phases of a release when there is time pressure to finish that final set of issues, and we are documenting a short time-limited commitment from people to fix issues before a release)
    • or at least that we not assign an issue to a new contributor until after they have submitted a reasonable PR fixing the issue.
    • Based on the discussion in the meeting, I'll put a PR somewhere in the docs that says:
      • we encourage someone that wants to work on an issue to post a comment on the issue
      • we generally don't assign issues to people
      • we encourage draft PRs to be opened early in the development process and linking the draft PR back to the issue
    • Also, we could extend the bot to flag new PRs by first-time contributors for attention.

• Isabela

  • Can we add the Jupyter discourse to the resources at the top of the mailing list?
  • Do we have a "don't show this again" or similar checkbox or settings for some dialogs?

    It was introduced for the renaming untitled files but it was not generic (and the feat. has been reverted).

  • Jupyter Community Call is next week! You can get more info on discourse

• Carlos

  • Undo manager capture timeout
    • We are using the default (500ms), which groups sentences. This is the behavior of every OS and editor but Apple software. Apple capture every character individually. What behavior should we adopt? Or should we make it configurable?
    • (william) Apple maybe doesn't do it every character anymore. CoCalc is similar, but a longer timeout, and I'm not sure I like it.
    • Colab seems to group undos by a similar amount of time.
  • Document or cell level history tracking?
    • The old behavior was a cell level history tracking and with the RTC we made it document-wide. Now cells and typing changes are mixed in the same undo manager. What behavior should we adopt
    • Allow to disable the document-wide history tracking jupyterlab#10791
    • https://discourse.jupyter.org/t/jupyter-command-notebook-undo-cell-action-not-working-as-expected/10484
    • (william) cocalc-jupyter is also document level rather than cell level, and I've not heard a complaint from our users about this in years. User configurable would seem optimal though...

• Brian

  • Cal Poly/Jupyter interns are giving their final presentations today and tomorrow
    • https://discourse.jupyter.org/t/jupyter-internship-jupyterlab-extension-presentations/10469
    • Feel free to join, but please don't Tweet far and wide (we have limited capacity).
    • The plan is to record and post on Youtube for wider consumption
  • Data registry/explorer
    • Piyush (@3coins) and I are starting to work on the data registry/explorer again.
    • The goal is the simplify and focus in on key user stories.
    • Starting to open issues for discussion
      • Use JupyterLab tokens to strongly type dataset interfaces jupyterlab-data-explorer#147
      • Core dataset API jupyterlab-data-explorer#146
      • Simplify data model by removing nested dataset capability jupyterlab-data-explorer#145
    • If you are interested in this, please join the discussion and/or reach out to Piyush or I.

• Vidar

  • File filter vs file "quick-find" mechanism seems to overlap, and could probably benefit from a overhaul. See also Visual indicator that a file filter is in effect jupyterlab#10911

• Martha

  • Based on this spec modified fields that are very long make the modified fields section a little long / awkward? suggestion - indicator next to modified fields / keep plugin settings together

Additional Discussion

• Frederic:
  • Galata PR10796 Demo -> Next week
• Darian / Pete / William
  • In RTC we should look at history to rewind, debug, etc.
    • I was talking to some users yesterday and they asked for a L/R panel that has a list of the timestamped versions of the doc.
  • William mentioned that their current collaborative meetings in CoCalc don't have the problems we were experiencing. What changed? Was it a technical change (refining the cocalc rtc implementation), or a behavioral change (understanding better how to collaborate)?

[afshin]

JupyterLab Weekly Call, September 1, 2021

Attendees

Name	Affiliation	GitHub	Favorite Dinosaur
A. T. Darian	Two Sigma	@afshin	peregrine falcon
Steven Silvester	Apple	@blink1073	stegasaurus
F. Collonval	QuantStack	@fcollonval
W. Stein	SageMath	@williamstein	Pyrex-o-suarus
Pete Blois	Google	@blois	Apatosaurus
Alex Bozarth	IBM	@ajbozarth	raptor
Michał Krassowski	Oxford	@krassowski
Mehmet Bektas	Splunk	@mbektas
Gonzalo Peña-C	Quansight	@goanpeca	T-rex
Jessica Xu	Quansight	@jess-x
Martha Cryan	IBM	@marthacryan
Karla Spuldaro	IBM	@karlaspuldaro
Jason Grout	Bloomberg	@jasongrout
Jeremy Tuloup	QuantStack	@jtpio
Rick McGeer	engageLively	@rickmcgeer	Troodon
Brian Granger	AWS/Cal Poly	@ellisonbg

Agenda

• Frederic:
  • Galata #10796 Demo
    • The one regression compared to previous Galata is that this new implementation does not have the ability to connect to an external browser session, which we discourage anyway.
    • We are also in the process of archiving the galata repo: Archive the repository galata#81
  • Benchmark in CI #10936
  • Doc theme change to use pydata theme #10834
• Steve
  • JupyterLab v3.1.10 and v4.0.0a7
  • jupyterlab-probot Demo (using Probot)
• Vidar
  • Building a project with TS 4.4 and deps on jupyterlab packages complain about a few jupyterlab and lumino typings. It mainly seems related to lib.d.ts changes.
• Jason
  • Nomenclature for "document-wide history": Enable disabling document-wide history tracking jupyterlab#10949 (comment)
  • (ws: I guess in CoCalc we call that "the TimeTravel History".)
  • It would be great if the name of this concept could also be applied to search, i.e., the scope of the search being in a single cell vs through the whole document.
  • Please comment on the PR with any ideas or preferences you have.
• Brian
  • Misformatting of markdown lists? On some lists in this doc, the second level items have a blank line above them. Looks like it was due to the empty lines between the primary items...
  • Dataset definition
    • See dataset interface below
    • Simplified data registry interface, added unit tests jupyterlab-data-explorer#148
    • https://github.com/jupyterlab/jupyterlab-data-explorer/blob/4beec8bf7a5d2cfb68985d4452fd6215f6f424bf/packages/dataregistry/src/dataset.ts
• Mike: PyPI classifiers for JupyterLab are available
  • See Get a Framework :: Jupyter :: Lab (:: 3) classifier jupyterlab#9538, add trove classifer docs, usage jupyterlab#10731

Additional Discussion

• Darian / Pete / William (from last week)
  • In RTC we should look at history to rewind, debug, etc.
  • I was talking to some users yesterday and they asked for a L/R panel that has a list of the timestamped versions of the doc.
  • William mentioned that their current collaborative meetings in CoCalc don't have the problems we were experiencing. What changed? Was it a technical change (refining the cocalc rtc implementation), or a behavioral change (understanding better how to collaborate)? [Answer from William: mostly just fixing bugs. You didn't have these problems when using hackpad? Example bug right now: every time somebody types the preview on the right keeps scrolling back to the top, which is annoying and hard to use.]
  • The coolest undo interface Jason has used is the undo tree interface in Emacs or Vim
  • A user might want to use global history vs local history in different phases in their work. For example, if they are iterating quickly in different cells, they may want a local cell undo. If they are trying to revert the document to the working state from yesterday, a global rollback mechanism is valuable.

Proposed dataset interface

https://github.com/jupyterlab/jupyterlab-data-explorer/blob/4beec8bf7a5d2cfb68985d4452fd6215f6f424bf/packages/dataregistry/src/dataset.ts

export interface Dataset<T extends JSONObject, U extends JSONObject> {
  /**
   * Unique identifier for the dataset, for in-memory
   * datasets, a unique uuid might be provided.
   * This id should be unique across a jupyter server instance.
   */
  id: string;
  abstractDataType: string;
  serializationType: string;
  storageType: string;
  /**
   * Output value for the dataset
   */
  value: T;
  /**
   * Additional properties for the dataset
   * that help serialize or query data
   */
  metadata: U;
  title: string;
  description: string;
  tags?: { [key: string]: string };
  version?: number;
}

[afshin]

JupyterLab Weekly Call, September 8, 2021

Attendees

Name	Affiliation	GitHub	Favorite Ancient Monument
A. T. Darian	Two Sigma	@afshin	Göbekli Tepe
William Stein	SageMath, Inc.	@williamstein	The Mount Baker Volcano
Frederic Collonval	QuantStack	@fcollonval
Loïc Huder	ESRF	@loichuder	Angkor Wat
Brian Granger	AWS/Cal Poly	@ellisonbg
Kevin Jahns	Yjs / Quantstack	@dmonad
Alex Bozarth	IBM	@ajbozarth
Steven Silvester	Apple	@blink1073	Barbican London Wall
Carlos Herrero	QuantStack	@hbcarlos
Jason Grout	Bloomberg	@jasongrout	just found https://en.wikipedia.org/wiki/Megalith for good reading :)
Jessica Xu	Quansight	@jess-x
Youness Bennani	Bloomberg	@bsyouness	Ek Balam
Isabela Presedo-Floyd	Quansight Labs	@isabela-pf	Jantar Mantar
Mehmet Bektas	Splunk	@mbektas	Galata Tower :)
Pete Blois	Google	@blois	2001 Monolith
Karla Spuldaro	IBM	@karlaspuldaro
Martha Cryan	IBM	@marthacryan
Rick McGeer	engageLively	@rickmcgeer	Stonehenge
Max Klein	JPM	@telamonian
Nick Bollweg	Georgia Tech	@bollwyvl @nrbgt	rosetta stone

Agenda

• Loic

  • Working on null file format (part 2) => null fileFormat raise Format is invalid error 400 jupyterlab#10757
  • New jupyterlab-hdf5 release bringing support for external/soft links. JLab3 release on the works.

• Brian

  • Enable disabling document-wide history tracking jupyterlab#10949 (comment)

  • In cocalc, William solves the simultaneous cell edit/move problem by just maintaining a "position" metadata field on each cell, and when a cell is moved, the position is just updated, i.e., it doesn't actually change an array datastructure. Caveat: you have to ensure the positions are unique. The position is a float so it is easy to insert between two other positions. Periodically, one of the processes renumbers all the cells as integers to take care of precision issues (it looks like maybe in the code now that is done by the client doing the cell move, since they can change anything globally anyways).

  • Regarding the discussion about moving objects in CRDTs: There is a discussion about it on the yjs discussion https://discuss.yjs.dev/t/moving-elements-in-lists/92 (Brian & William). Moving elements in a CRDT is not trivial but definitely possible. However, it always comes with tradeoffs. There is a solution described that sounds fairly similar to William's solution.

• Isabela

  • Jupyter accessibility workshops are starting on October 2 and 9! We would like to have some more event moderators from the community. (More specific announcements to come!)

• Max

  • yarn v3 upgrade!
    • almost done
    • Upgrade jlpm cmd to yarn 3.x jupyterlab#10929
    • one last problem: jlpm run vs yarn run (vs $npm_execpath run) in ensure-buildutils.js and other related scripts that might run as part of pip install ... before jlpm is properly installed
      • under discussion: move jlpm entrypoint to jupyterlab_server

• Frederic

  • End of 3.1.x to start 3.2.x?
    Need to sync with @goanpeca to list if all 3.2.x feat. are ready.
  • Performance: status bar item position unstable - maybe related to #9910
  • Translation:
    • Settings are now translated before being served (JLab server 2.8+)
    • What about the documentation?
      It should be its own project on Crowdin. Ideally an intermediate step could be to get the translated text from the
      UI text; like sentence You can access that feature through the File menu.

Additional Discussion

• Duplicate Design System CSS GitHub label? Can we remove?
  • Design System CSS
  • tag: Design System CSS
  • used in labeler