# scikit-image Community Call

- **Time:** Tue, 2024-11-20, 17:00 – 18:00 UTC
- **[Join video call via Jitsi](https://meet.evolix.org/skimage-meeting)**
- ~~**[Join video call via Zoom](https://us06web.zoom.us/j/88060567580?pwd=THRpaWFnSFNwK0Fycy9FVk5RYnV5UT09)**~~
- **Notes:** https://hackmd.io/@scientific-python/HJWUBDVNi
- **[scikit-image Community Calendar](https://scientific-python.org/calendars/skimage.ics)**
- **[Archive — Meeting Notes](https://github.com/scikit-image/meeting-notes)**
- **[Code of Conduct](https://scikit-image.org/docs/stable/conduct/code_of_conduct.html)**

**Present:** Lars, Marianne, Stéfan

## Agenda

- EOSS5 interim report due on Jan 1, 2025
- Need to ask NumFOCUS for financial report
Reporting period: end date of previous report to submission date of current report
- Extension for Marianne's ICA already went through, Lars's is still pending
- Should get a few items from the skimage2 list in (prioritize these items and request reviews)
- Lars will focus on releasing docstub v0.3 (minimal viable prototype) in time
- Then update [Generate typing stubs with docstub #7546](https://github.com/scikit-image/scikit-image/pull/7546) to include first infrastructure in skimage

- skimage 0.25 release progress
- [milestone](https://github.com/scikit-image/scikit-image/milestones/0.25)
- S suggested to Lars that we also see if we can close the loop on [7566](https://github.com/scikit-image/scikit-image/pull/7566), [7292](https://github.com/scikit-image/scikit-image/pull/7292), [6892](https://github.com/scikit-image/scikit-image/pull/6892)
- 7566: just needs a test
- 7292: ready to merge, need to check CI failure on Windows
- 6892: lazy loading is interfering with `lookfor`, Lars is debugging it; not super urgent
- Will try to get all this merged today
- [7504](https://github.com/scikit-image/scikit-image/issues/7504) can be closed
- Double-checked that watershed fixes ([7071](https://github.com/scikit-image/scikit-image/pull/7071)) made it in ✓
- Also review [path to skimage2](https://github.com/scikit-image/scikit-image/issues?q=is%3Aopen+is%3Aissue+label%3A%22%3Ahiking_boot%3A+Path+to+skimage2%22)
- Timeline: end-of-week rc2

- Dispatching
- Aditi received the NF Small Developer Grant
- First PR to get over the line: [7520](https://github.com/scikit-image/scikit-image/pull/7520)

- General
- [Marco Gorelli: Narwhals](https://www.youtube.com/watch?v=kPtUPe5Egak)
- As close as we're likely to get to DataFrame API
- [spin 0.13rc0](https://pypi.org/project/spin/0.13rc0/)
- [extend built-in commands](https://github.com/scientific-python/spin/pull/248)
- [pre-import for IPython](https://github.com/scientific-python/spin/pull/251)
- [diff for NumPy](https://github.com/numpy/numpy/pull/27794/files)
- Lars will reply to Juan about typing (going for the Annotated route)
- Close the loop on [the discourse topic](https://discuss.scientific-python.org/t/typing-scikit-image/877/32)
- For now, we can't easily get what we want without a mypy plugin:
- We want to check, if input is annotated as, say, Mask, it cannot be passed into a parameter that expects an Image
- At the same time, a Mask *can* be specified as an Array, otherwise users will *always* need to annotate their types
- We cannot simply make parameter requirement "Mask | ndarray", because then Image will also pass (being an narray)
- Logic we need: "if annotated as Mask, Image, etc., this parameter only accepts *Mask*"
- As we add the type annotations, this will force us to define precisely what we mean by *Mask*

## Previous agenda (2024-11-05)

**Present:** Stéfan van der Walt, Lars Grüter, Aditi Juneja

- Remove docstring manipulation in `RegionProperties` during runtime (see `_install_properties_docs`)?
- It's brittle, see [#7448](https://github.com/scikit-image/scikit-image/pull/7448)
- Doesn't work with our planned use of docstub
- Property descriptions in `regionprops` tend to be short and lacking literature references
- Suggested fix: expose `RegionProperties` and each `@property` in HTML docs, and document the properties there

- [0.25.0 release?](https://github.com/scikit-image/scikit-image/milestone/29)
- [Use scipy.sparse arrays #7576](https://github.com/scikit-image/scikit-image/pull/7576)
- Make a TODO item on the pytest warning handling
- Should we [check before downcasting](https://github.com/scikit-image/scikit-image/pull/7576/files#r1824544991)?
- Anything else holding it up?

- [feat: add mask parameter to graycomatrix function for selective computation #7544](https://github.com/scikit-image/scikit-image/pull/7544)
- New feature would introduce slow-down or duplication, do we have a general stance on this?

- Speed-up CI
- [Build docs in parallel instead of sequentially on Azure #7593](https://github.com/scikit-image/scikit-image/pull/7593), ready to review
- Principle: test suite has to run, anything we add for "convenience" is OK but should not slow test suite down

- Progress on [pre-skimage2 list](https://github.com/scikit-image/scikit-image/wiki/API-changes-for-skimage2)
- We have a NCE on the CZI grant until end January 2026
- [Replace square, rectangle, cube with footprint_rectangular #7566](https://github.com/scikit-image/scikit-image/pull/7566)
- Part of incremental changing selems to `footprint_*`
- https://github.com/scikit-image/scikit-image/wiki/API-changes-for-skimage2

- Update the wiki

- NIH grant: postponing submission until next year, 4 June 2025
- [Mechanism](https://grants.nih.gov/grants/guide/rfa-files/RFA-OD-24-010.html)
- Think about setting up survey

### Dispatching Topics

- Narrative dispatching documentation - 3 parts
- scikit-image users
- How to use it (in general - `SKIMAGE_NO_DISPATCHING`, etc.)
- Introspection/logging/warnings : DispatchNotification, etc.
- Backend developers(on the [development page](https://scikit-image.org/docs/dev/development/index.html))
- What makes up a valid backend?
- entry_points : “skimage_backends”, ”skimage_backend_infos” (“implementation” and “info”)
- How does things like `can_has` work?
- How does scikit-image offer testing facilities for backends?
- Implementation details
- entry_point mechanism
- Working and usage of the `dispatchable` decorator - how are function signatures handled?
- Testing details
- How does the backends’ metadata gets handled, etc.
- Should some of it go into a SKIP-5? - https://scikit-image.org/docs/dev/skips/index.html
- TODO[Aditi]: continue this(or other developments) in [PR#7513](https://github.com/scikit-image/scikit-image/pull/7513)

- Created [#dispatching](https://discord.com/channels/786703927705862175/1303422714987679754) channel on SP discord for day-to-day discussions

- Some inconsistencies between scikit-image’s and networkx’s dispatching mechanisms
- Which ones are necessary and why do they exist?
- Eventually we should ensure ecosystem-wide consistency
- `get_implementation` vs having the algorithms as the attributes of the entry_point object
- Having some consistency in setup/private functions’ names : like steps(or functions involved) in fetching the backends(`all_backends()` vs `_get_backends()`) and then fetching the implementations from the backends, and loading the backend metadata(or info), etc. (adding these functions/setup in `spatch`?); Meaning and usage of the term “backend info” in scikit-image Vs networkx
- Naming:
- `can_has` vs `can_run`
- `dispatchable` vs _dispatchable`
- `_backends.py` vs `backends.py` (maybe ok to have this difference, because both projects are organised differently)
- Should NetworkX’s logging and BackendInfo be more modular like scikit-image’s?
- [Lars] It would be best to address these once we have a good enough prototype in place.

- [TODO] Letting users set the order of backends instead of hard coding it.

- potential backend libraries, for reference, other than `cuCIM`?

- NumFOCUS’s SDG R3 : might get notified tomorrow
- If positive, next step would be - ICA form (draft): https://hackmd.io/@Schefflera-Arboricola/SynEV9vZ1x (contract will finalise within 1-3 weeks (based on the responsiveness of PI and contractor) of submitting the above form)

-------------------------------------------------

**After the meeting**, please commit the notes to our [scikit-image/skimage-archive](https://github.com/scikit-image/skimage-archive) repository.