SciPy Community Meeting

--- tags: SciPy, meeting --- # 2024 SciPy Community Meeting - Next meeting: **November 13th, 2024 (Wednesday) @ 2pm UTC** - Join via Zoom at https://us06web.zoom.us/j/6345425936?pwd=aDVFQzVmbk9SVU5jU0Jwc0s3YWUrdz09 - If you need a passcode for zoom, `scipy` should work. - Meeting notes archive at https://github.com/scipy/archive/tree/main/community_meetings ### Code of Conduct We want to take a moment to remind you that this meeting, like all project spaces is meant to be open, welcoming, diverse, inclusive, and it's important for us to have a healthy community. Like all SciPy spaces, and everyone participating in them, this meeting will follow our [code of conduct](http://scipy.github.io/devdocs/dev/conduct/code_of_conduct.html). If you haven't read it yet, please take some time to do so later on as it already applies to you. For now, in short, please be kind and generous towards one another. --- ## Agenda for November 13th, 2024 > please add your names (and github handle in parenthesis). This will make it easier to stay in touch later on issues and pull requests (PRs) 😉 > This is optional since these notes will be recorded in our Github repository. If you'd like you can also paste your answer in the zoom chat 😉 **Present:** Melissa (@melissawm), Ralf (@rgommers), Amine (@mathaziz), Evgeni (@ev-br) *Feel free to add items for discussion to this agenda!* - [name=Melissa] Change meeting frequency to monthly. - https://github.com/scientific-python/scientific-python.org/pull/697 - [name=Amine] Interested in contributing, applied mathematics background. - Will take a look at scipy.optimize - Interested in Fortran->C/C++ conversion - [name=Ralf] Items left for 1.15.0 - High-prio: HiGHS, free-threading, OpenBLAS - [1.15.0 milestone issues](https://github.com/scipy/scipy/issues?q=sort%3Aupdated-desc+is%3Aissue+is%3Aopen+milestone%3A1.15.0) - [1.15.0 milestone PRs](https://github.com/scipy/scipy/pulls?q=sort%3Aupdated-desc+is%3Apr+is%3Aopen+milestone%3A1.15.0) ## Agenda for October 30th, 2024 **Present:** Tyler (@tylerjereddy) --- ## Agenda for October 16th, 2024 **Present:** Melissa (@melissawm) - [name=Melissa, Rohit] Who can review https://github.com/scipy/highs/pull/67? ## Agenda for October 2nd, 2024 **Present:** Melissa (@melissawm), Stéfan (@stefanv), Tyler Reddy (@tylerjereddy), Jake Bowhay (@j-bowhay) - [name=Lucas] - New library to share private array API helpers with others: https://data-apis.org/array-api-extra/ - Possibly vendored with SciPy? - [name=Jake] [Better way to document functions that support the Array API](https://github.com/scipy/scipy/issues/21255) (vs not): which array types supported etc. - Keeping the docs and what we test against in sync - array-api-strict - Exception may be immutability: Jax, e.g. - Same challenge holds for dispatching to backends, such as ndimage replacements - [name=Stéfan] Asked about mystmd tutorial experiment, longer term vision of docs tooling - https://github.com/numpy/numpy-tutorials/tree/mystjs - Toolchain right now works: reproducible, testable tutorial content - myst is sparsely maintained - myst different from mystmd (markdown + js), new tools by Rowan Crockett & co. — separation is a bit confusing still - [name=Stéfan] How we can change review culture to best move PRs forward - Fix pedantic recommendations yourself? - Allow finishing PRs that have stalled? - Lack of bandwidth / expertise is limiting - Finding reviewers who understand the topic at depth can be challenging; mechanical reviews are easy - How do we elevate real review needs to domain experts? - X has open PR waiting on review to move on vs Y has open PR but isn't paying attention - NumPy team gets good momentum from fortnightly triage meetings - Sparse WG has been helpful to get a global view of `sparse` (and developer) priorities - SPEC endorsements: SPECs 6, 7, and 8 are ready to look at - Tyler thinks release management one is pretty straightforward - Endorse with PR here: https://github.com/scientific-python/specs - [name=Melissa] Moving the next Newcomers meeting by a week (Melissa will be traveling) --- ## Agenda for September 4th, 2024 **Present:** Lucas (@lucascolley), Tyler (@tylerjereddy) - [name=Lucas, Tyler] - Reproducing CI failures in https://github.com/scipy/scipy/pull/21149 - [name=Lucas] - `xp` assertions work seems to be converging - Currently at the NumFOCUS Summit (next few days) --- ## Agenda for August 21st, 2024 **Present:** Lucas (@lucascolley), Melissa (@melissawm) - [name=Lucas] Improving `xp` assertions - RFC describing problems at https://github.com/scipy/scipy/issues/21044 - My attempt at an implementation at https://github.com/scipy/scipy/pull/21413 - Various possible implementations for the same test coverage and various opinions on what we need to test - complicated! - [name=Lucas] Dask progress - https://github.com/scipy/scipy/pull/20956 CI is now showing all remaining errors for Dask - Various issues upstream at Dask and array-api-compat, but not too many big issues! --- ## Agenda for August 7th, 2024 **Present:** Melissa, Tyler - [name=Melissa] Had to leave for a conflicting work meeting, please take notes in case you come by :) --- ## Agenda for July 24th, 2024 **Present:** Lucas, Jake, Ralf, Evgeni - [name=Lucas] - New vendored code tracker: [gh-21232](https://github.com/scipy/scipy/issues/21232) - fine as an issue 👍 - GPU CI: https://betatim.github.io/posts/github-action-with-gpu/ - do we want to follow yet or not? - yep, we can go ahead once we sort out finances - [name=Jake] - Array API documentation - Whilst there is a centeral list of functions that support the array API prehaps it would also be nice to have something in the docstring that indicates this? - For functions that dispatch, clear documentation on which backends will dispatch and which will use SciPy implementation - Array API support for generalised eigenvalue problems --- ## Agenda for July 10th, 2024 **Present:** Melissa (@melissawm), Tyler (@tylerjereddy), Lucas (@lucascolley) - [name=Lucas] - Should we update `CODEOWNERS` at some point? - Some people may be pinged incorrectly. We're not sure exactly how to ask about it - @lobpcg is looking for reviewers on `sparse.linalg` PRs - Dask support - went to their community meeting but it was empty (https://github.com/scipy/scipy/pull/20956) - [name=Melissa] Blockers to move towards spin? --- ## Agenda for June 26th, 2024 **Present:** *nobody* --- ## Agenda for June 12th, 2024 **Present:** Ralf (@rgommers), Tyler (@tylerjereddy), Lucas (@lucascolley), Jake (@j-bowhay) - [name=rgommers] Strategy for PRs that add array API support (discussion in [gh-20935](https://github.com/scipy/scipy/pull/20935), [gh-20946](https://github.com/scipy/scipy/pull/20946)) - try to keep refactoring in separate commits (or separate PRs) to array API conversion - [name=Lucas] - almost free from exams, plan to contribute regularly over the summer. Open to suggestions for priorities. I plan to implement `Dask` support, tie up some loose ends from issues I have opened, and continue helping with the array API support effort. Would be nice to work on the backlog of first-time contributor PRs too. - I've got funding from my college to travel to EuroScipy 2024 in Poland (end of August) in the summer 🎉 - [name=rgommers] Accelerate LAPACK 3.11 is in beta, - [name=Jake] Scientific Python Summit - Dispatching: https://hackmd.io/yI1iAqekQIq0a4jLS9WPyw - Package for functions which sit between the standard and consumer libraries? - Switching from NumPy to PyTorch? https://github.com/scientific-python/summit-2024/issues/18 - Supply chain security: https://github.com/scientific-python/specs/pull/325 --- ## Agenda for May 29th, 2024 **Present:** Melissa (@melissawm), Shiv Katira (@shivkatira), Sebastian (github: SebastianMunozMora), Ralf Gommers (@rgommers), Rohit (@haozeke) - [name=Melissa] FYI: Proposed release schedule https://discuss.scientific-python.org/t/proposed-release-schedule-for-scipy-1-14-0/1151/4 - [name=Melissa] Newcomer building issue (windows) - Codespaces may be an alternative - [name=Rohit] HiGHS - Will wait for next release. --- ## Agenda for May 15th, 2024 **Present:** Melissa (@melissawm), Tyler (@tylerjereddy) - [name=Melissa] [Release schedule](https://discuss.scientific-python.org/t/proposed-release-schedule-for-scipy-1-14-0/1151) - Proposed scheduled will be kept for now, unless there are objections in the next few days. --- ## Agenda for April 17th, 2024 **Present:** Melissa (@melissawm), Agriya (@agriyakhetarpal), Tyler (@tylerjereddy) - [name=Lucas] I can't make it but here are a few things: - I started looking at updating the core dev guide, https://github.com/scipy/scipy/pull/20199. I'm a bit out of my depth in [the distributing section](https://scipy.github.io/devdocs/dev/core-dev/index.html#distributing), if someone would like to take a look at that. We should also write down a list of code which is vendored, but not obviously so (not a git submodule), so that we know when to contribute fixes upstream rather than into SciPy (inspired by some confusion in gh-20227). - Second part can be a follow up? - JAX PR is pretty much just blocked on [this array API issue](https://github.com/data-apis/array-api/issues/609#issuecomment-2041230122). - [name=Pamphile] cannot make it but: define a schedule for dropping the mailing list (read only mode) - No big objections - Proposal: freezing the mailing list next week (Thursday April 25). - Melissa to send a reminder to the mailing list - [name=Melissa, Agriya] MathJax - When setting up JupyterLite features/converting .md files to executable notebooks, we noted MathJax is not loading properly. It looks like the MathJax path is hardcoded in MyST-NB (see https://github.com/executablebooks/MyST-NB/issues/590) - A potential solution would be to implement a fallback logic: if we are connected to the internet, use the regular MathJax; if offline, use the scipy vendored MathJax. This means notebooks would only be broken for offline local docs builds. - Agriya will be opening a PR for this based on the discussion and further actions can be decided upon in the PR thread - [name=Melissa, Agriya] Proposal for Jupytext live notebooks in the docs - https://melissawm.github.io/scipy/tutorial/stats/sampling.html (enables opening up document in a new tab as a Jupyter notebook) - Doesn't open the notebook in the correct format because we need an extra step to convert .md files into .ipynb files - Agriya is investigating a fix for this as well --- ## Agenda for April 3rd, 2024 **Present:** Melissa (@melissawm), Lucas (@lucascolley), Evgeni (@ev-br), Ralf (@rgommers) - [name=Lucas] Separating docs releases from library releases? - Comment from Pamphile: https://github.com/scipy/scipy/issues/19973#issuecomment-2024229349 - Seems like there are no better alternatives. E.g. storage limitations on GitHub. - [name=Lucas] Problem with make dist https://github.com/scipy/scipy/issues/20379 - [name=Melissa] Moving mailing list to scientific-python discourse - https://discuss.scientific-python.org/c/contributor/scipy/ - There is also https://discuss.scientific-python.org/c/user/scipy - Stefan already did some homework back then when we first proposed the move. We are just waiting for the official go and we can move. (cf previous threads on our ML) - Will need PRs to update scipy.org and the docs to point to discourse instead of the mailing list. - [name=Pamphile] xz backdoor: - we need to bring more transparence and guarantee with our release process. It should be fully auditable by anyone. Trusted publisher mechanism should also be used to publish on PyPi. - needs more security on who can actually push to pypi, otherwise any tags get automatically published as new releases - Pamphile: if you use trusted publisher, it needs a GH environment and you can (and should) restrict the environment to only run if certain maintainers approve it. On GH you can also restrict who can push a tag if you want to restrict that part. - we need to be carefull with the addition of new maintainers. We need, at the very least, to do real ID checks. Many orgs like Debian for instance have strong requirements. - [name=Ralf] Work on nogil compatibility has started on NumPy (Lysandros, Nathan) - SciPy probably next --- ## Agenda for March 20th, 2024 **Present:** Melissa (@melissawm), Tyler (@tylerjereddy), Jake (@j-bowhay), Lucas (@lucascolley) - [name=Tyler] Release on track, first release candidate out, waiting on NumPy. - Issues with openblas, but nothing urgent. - [name=Lucas] Mailing list migration? - Melissa reached out to Stéfan - [name=Lucas] SciPy hackmd account? - Unsure on right way to go with passwords, maybe pending discussion of https://scientific-python.org/specs/spec-0006/ - [name=Melissa] Contributor docs versioning (or lack thereof) - Looking into it with Napari, but no elegant solution. Maybe pending a solution on the pydata-sphinx-theme side. --- ## Agenda for March 6th, 2024 **Present:** Melissa (@melissawm), Ralf (@rgommers) - [name=Lucas Colley] I can't make it, but all points from the previous agenda carry through, apart from the PyTorch tol issue. - also, I've added labels to the first-time contributor project: https://github.com/orgs/scipy/projects/2/views/2, so it's much easier to see at a glance what is waiting for review. For example, `optimize` and `integrate` have quite a few PRs which are waiting for a decision, while there is currently only 1 docs PR which isn't waiting for a decision. - [name=Melissa] Thank you! - [name=rgommers] We may want to touch on the next release, since creation of `maintenance/1.13.x` is imminent. Anything blocking or important to merge/resolve? - [name=Melissa] Use a SciPy hackmd space for all notes? - discussed: sounds like a good idea - [name=Melissa] Mailing list migration - Melissa to investigate steps needed to set up discourse and archive existing content (probably in a new github repo for more permanent storage) --- ## Agenda for February 21st, 2024 **Present:** _nobody_ - [name=Lucas Colley] I can't make it, but carrying over unresolved things from previous weeks: - @ralf there was another complaint about the `scipy.{numpy_function}` removals not hitting the release notes: https://github.com/scipy/scipy/issues/19974. - I tried https://github.com/scipy/scipy/pull/19935 with a PyTorch nightly but the problem is still there. Will probably go ahead with Albert's suggested loosening of the test, but not urgent. - The `scikit-umfpack` circular dependency issue is still a thing: https://github.com/scipy/scipy/issues/19783 - It would be good to get the docs design updates in for 1.13.0 (https://github.com/scipy/scipy/issues/19958). - Also, if we can get the version warning banner to point to the development version rather than stable on the contributor pages (https://github.com/scipy/scipy/issues/19699). - Could anyone give https://github.com/scipy/scipy/pull/20079 a look (>4GB WAV files in `scipy.io`)? A first time contribution, and there has been recent interest in the feature being added. --- ## Agenda for February 7th, 2024 **Present:** Lucas (@lucascolley), Yagiz (@yagizolmez) - [name=Yagiz] A few PRs moving forward, no blockers currently - Envelope extraction: https://github.com/scipy/scipy/pull/19814 - Multivariate Wasserstein Distance: https://github.com/scipy/scipy/pull/20024 - [name=Lucas] - @tyler I'm going to try to finish up the array API test infra PR for 1.13.0, but it may have to wait for 1.14 if I don't find the time - Similarly for https://github.com/scipy/scipy/pull/19935 - `linalg` array API will definitely have to wait until the summer now - [name=Lucas] Ralf said that we should "retroactively" fix that some expired deprecations were missing in the 1.12 release notes. What does that fix consist of? (https://github.com/scipy/scipy/issues/19974) - [name=Lucas] A reminder that there is still a circular-dependency issue with `scikit-umfpack` as an optional test dependency (https://github.com/scipy/scipy/issues/19783) --- ## Agenda for January 24th, 2024 **Present:** Melissa (@melissawm), Jake (@jbowhay), Tyler (@tylerjereddy), Lucas Colley (@lucascolley), Albert (@steppi) - [name=Lucas] updates from me :) - thanks Axel for checking through old milestones for things that have been missed! - PyData-sphinx-theme update is finally in! Good time to think about any docs improvements now that we have current and (🤞) stable docs rendering. I can give the version-warning banners a look when I have time but I'm more than happy for someone else to have a look at that. - [name=Melissa] for the version warning banner, we may want to check if we want to backport or find a solution similar to what numpy did before the pydata sphinx theme, for example https://numpy.org/doc/1.18/index.html - We also may want to check plausible and see if the older versions of the docs are being visited. If not, we don't need to touch them. - array API is slowly pushing forwards - gh-19682 should help reduce all future diffs and make testing more complete (just waiting for confirmation from Ralf w.r.t how we want to test array-likes going forwards) - Special failure with `torch` backend: https://github.com/scipy/scipy/issues/19928. Still not sure what's causing it, maybe has been fixed in a more recent PyTorch version. - [name=Melissa, Lucas] New PyData Sphinx Theme and dark mode/colors - Theme does not look very good on dark mode, Melissa will try to fix. - Toolchain roadmap update: https://github.com/scipy/scipy/issues/19815 - Potentially related: https://github.com/scipy/scipy/issues/19832 - [name=Tyler] NumPy 2.0 delayed - We'll put out a minor release to compensate for that - 1.13.0 will go out then NumPy 2.0 is on PyPI - [name=Albert] Interactive documentation - It would be nice to merge interactive documentation soon. Pending is deciding on a workflow to make sure the version of SciPy in Pyodide is synced with the releases. See https://github.com/scipy/scipy/issues/19729 - There is a draft deployment you can check here: https://steppi.github.io/scipy --- ## Agenda for January 10th, 2024 **Present:** Melissa (@melissawm), Lucas Roberts (@rlucas7), Ralf Gommers (@rgommers), Evgeni (@ev-br) - [name=Lucas Colley] I can't make it, but here are some updates: - I archived the community notes for 2023 on the archive repo. Update this notebook for 2024 (or start a new notebook)? - [name=Melissa] Thanks! I cleared up this one. - I triaged all of the PRs in the First time contributor project: 43 waiting for author, 61 need review. - I went through most open PRs that are > 1 month old and added `needs-work` or `need-decision` labels. The ones without either label are generally PRs that seem uncontroversial and just need a review. - I wasn't able to host the newcomers meeting last week as the zoom meeting wasn't open. - [name=Melissa] I will investigate, this should not happen :( - [name=Lucas Roberts] scipy.special, Stirling numbers of the first kind - Might reach out to Albert to ask questions - [name=Evgeni] Changes incoming for interpolate - Might use a review: https://github.com/scipy/scipy/pull/19633 --- ### Notes for December 27th, 2023 **Present:** Melissa (@melissawm), Jake (@j-bowhay), Lucas (@lucascolley) - [name=Lucas] M1 Mac + `venv` build issues - I reproduced issues that someone was running into during the PyLadies sprint by following the `venv` build instructions. - [name=Lucas] `-pie being ignored` build warnings - Any idea where they're coming from? - https://github.com/scipy/scipy/issues/19736 - [name=Lucas] Directing to `development` version of the contributor docs from other versions - +1 from Pamphile and Ralf, but how to go about it? - https://github.com/scipy/scipy/issues/19699 - [name=Lucas] Incorrect test in `cluster` - Looks like a typo, although errors are introduced on fixing it. - https://github.com/scipy/scipy/issues/19683 - [name=Lucas] DX PR: `requirements/*.txt` files - Ralf gave initial review, ready for another look. - Based on the script used in skimage and numpydoc. - https://github.com/scipy/scipy/pull/19750 - [name=Jake] https://github.com/scipy/scipy/issues/19223 --- Please remember to archive this file and commit it to https://github.com/scipy/archive ### Useful Resources * [Our contributor Guide](http://scipy.github.io/devdocs/dev/dev_quickstart.html) * This is where we keep past meeting notes from the new contributors meeting https://github.com/scipy/archive/ * You will soon find today's meeting notes there * You can also have a look at topics and links that were shared before 🧐 ### Communication channels How will we communicate asynchronously while working on the project? * [Developer mailing list](https://mail.python.org/mailman/listinfo/scipy-dev) * We publish information which is important for contributors on this list. People from other projects might also share useful information or questions here. * Slack [(Invitation link)](https://join.slack.com/t/scipy-community/shared_invite/zt-1a76bomjr-fuS1ZTnmP7b32kIhLb6QMg)

Syntax	Example	Reference
# Header	Header	基本排版
- Unordered List	Unordered List
1. Ordered List	Ordered List
- [ ] Todo List	Todo List
> Blockquote	Blockquote
Bold font	Bold font
Italics font	Italics font
~~Strikethrough~~	~~Strikethrough~~
19^th^	19^th
H~2~O	H₂O
++Inserted text++	Inserted text
==Marked text==	Marked text
[link text](https:// "title")	Link
![image alt](https:// "title")	Image
`Code`	`Code`	在筆記中貼入程式碼
```javascript var i = 0; ```	`var i = 0;`
:smile:		Emoji list
{%youtube youtube_id %}	Externals
$L^aT_eX$	L^aT_eX
:::info This is a alert area. :::	This is a alert area.