You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
An async stand-up to provide status reports on what you've been up to the previous week, what your plan is for the upcoming week, and to request help for things you are working on.
Copy and paste the template below, and answer questions as needed!
**🙌 Thanks I'd like to give**- <name> helped me out with...
**✅ Updates from last week**- I worked on <link-to-pull-request>
**🔭 My plan for the current week**- I plan on working on...
**🙏 Challenges that I've faced and things I'd like help with**- I had a hard time figuring out...
**🏝️ My availability for next week**- I'll be off on <date>
**💬 Important items for discussion**- I have a question about <issue-or-pull-request>...
- Can @<github_handle> comment on issue #NN?
- I opened #NN for discussion.
- I need to discuss...
🎯 Task Status
Please make sure pull request labels and draft status reflect the current status of your work!
I worked on, edited, and published the METR blog post.
PR reviews.
Did the thing which one should NEVER do which is try to write a data visualization plotting library.
TL;DR: our plot/ctor implementation needs a lotttt of love, as the initial approach proved difficult to scale to other visualization types and difficult to generalize. That implementation needed to be scrapped.
After some R&D, the most viable candidate for replacement was authoring something which works on top of Vega/Vega-lite. Doing so would allow us to get out of the rendering game and simply provide more ergonomic wrappers. I decided to spend "a few minutes" to see what was possible. A few days later and I'm deep in refactoring mode on the refactor/plot branch.
Results are thus far promising. Thus far:
vendored vega, as they insist in being bleeding edge (particularly in using async imports). The vendored library is transpiled to ES5.
created various wrappers for Vega specification components (e.g., Axis, Scale, Title, etc).
initial PoCs for generating Vega specifications and rendering to SVG, with view output to Electron, stdout, and a browser window.
🔭 My plan for the current week
Code reviews.
Now that I am in data viz mode, I'd like to get an end-to-end line chart API working and fully polished.
BLAS extended signed zero summation clean-up (from last week).
Writing one or two "recipes" (from last week).
Scaffolding work (related to Gunj's QS internship).
🙏 Challenges that I've faced and things I'd like help with
Data handling in Vega and how we can avoid data copies. Turns out the dstructs/struct and array/struct-factory work is applicable here and should allow for zero-copy semantics.
I worked on stdlib-js/stdlib#7686, this is still WIP currently fighting with indexing issues in JS sigh
🔭 My plan for the current week
I don't think I should get too caught up in dlaqr5 and halt all further progress so i'll pick and work on a few routines side by side.
I plan to work on:
dladiv
dladiv1
dladiv2
cleanup previous PRs and make them ready for merge
🙏 Challenges that I've faced and things I'd like help with
struggling with indexing issues between so many loops in dlaqr5, i've discussed this with @kgryte over slack and we have an action plan, will ask for help if nothing else works out
🏝️ My availability for next week
I won't be able to make it to the standup tomorrow evening as I have some tests at that time, really sorryy 🙏
Thanks to @kgryte for the quick feedback and advice. Also, big thanks to @Planeshifter for reviewing my PRs, giving helpful suggestions, and merging some of the older ones too.
I also really appreciate the valuable feedback on the midterm evaluations. Definitely planning to explore areas outside math/base/special during GSoC, as suggested. Working on assertion utilities and ULP difference was a fun ride. Maybe next up: some manifest.json linting support?
I’m back to implementing trigonometric packages in math/base/special, currently focusing on sinpif and cospif.
🙏 Challenges that I've faced and things I'd like help with
I spent the entire day on sinpif. Porting the C implementation from FreeBSD to JS was harder than I thought, especially dealing with signed/unsigned bit manipulation. I also misunderstood how the fast floor macro worked, which costed me about 3 hours on a silly bug. But now I know better! Hoping to open a PR soon.
🏝️ My availability for next week
Available all week!
💬 Important items for discussion
Nothing at the moment. I'll ask if anything comes up while working on sinpif and cospif.
Anything else left to be implemented in this space?
@anandkaranubc The is-almost-same-value family of similar packages. E.g., isAlmostSameFloat64Array, hasAlmostSameValues, etc. With these, we should have all necessary assertion utilities for testing. These can be gone in parallel to the ongoing single-precision work in math/base/special.
Activity
kgryte commentedon Jul 21, 2025
🙌 Thanks I'd like to give
✅ Updates from last week
TL;DR: our
plot/ctorimplementation needs a lotttt of love, as the initial approach proved difficult to scale to other visualization types and difficult to generalize. That implementation needed to be scrapped.After some R&D, the most viable candidate for replacement was authoring something which works on top of Vega/Vega-lite. Doing so would allow us to get out of the rendering game and simply provide more ergonomic wrappers. I decided to spend "a few minutes" to see what was possible. A few days later and I'm deep in refactoring mode on the
refactor/plotbranch.Results are thus far promising. Thus far:
vega, as they insist in being bleeding edge (particularly in using async imports). The vendored library is transpiled to ES5.Axis,Scale,Title, etc).🔭 My plan for the current week
🙏 Challenges that I've faced and things I'd like help with
dstructs/structandarray/struct-factorywork is applicable here and should allow for zero-copy semantics.🏝️ My availability for next week
💬 Important items for discussion
aayush0325 commentedon Jul 21, 2025
🙌 Thanks I'd like to give
@kgryte for being there when i needed help :)
✅ Updates from last week
I worked on stdlib-js/stdlib#7686, this is still WIP currently fighting with indexing issues in JS sigh
🔭 My plan for the current week
I don't think I should get too caught up in
dlaqr5and halt all further progress so i'll pick and work on a few routines side by side.I plan to work on:
🙏 Challenges that I've faced and things I'd like help with
struggling with indexing issues between so many loops in
dlaqr5, i've discussed this with @kgryte over slack and we have an action plan, will ask for help if nothing else works out🏝️ My availability for next week
I won't be able to make it to the standup tomorrow evening as I have some tests at that time, really sorryy 🙏
💬 Important items for discussion
none atm
ShabiShett07 commentedon Jul 21, 2025
🙌 Thanks I'd like to give
✅ Updates from last week
blas/base/ggemvstdlib#7697blas/base/dsyrstdlib#7689🔭 My plan for the current week
blas/base/ggemvstdlib#7697BLASblas/base/gsyr🙏 Challenges that I've faced and things I'd like help with
🏝️ My availability for next week
💬 Important items for discussion
headlessNode commentedon Jul 21, 2025
🙌 Thanks I'd like to give
plotAPI and as always Thankyou for constant support and feedback.✅ Updates from last week
ndarray/base/somestdlib#7653ndarray/base/any-bystdlib#7664ndarray/base/anystdlib#7640blas/ext/last-index-ofstdlib#7665🔭 My plan for the current week
findIndexi.e. BLAS implementations, ndarray wrappers and top-level implementation.ndarray/base/findstdlib#7426 is merged, I'll work on its top-level implementation and tryndarray/find-last🙏 Challenges that I've faced and things I'd like help with
🏝️ My availability for next week
💬 Important items for discussion
anandkaranubc commentedon Jul 21, 2025
🙌 Thanks I'd like to give
Thanks to @kgryte for the quick feedback and advice. Also, big thanks to @Planeshifter for reviewing my PRs, giving helpful suggestions, and merging some of the older ones too.
I also really appreciate the valuable feedback on the midterm evaluations. Definitely planning to explore areas outside
math/base/specialduring GSoC, as suggested. Working on assertion utilities and ULP difference was a fun ride. Maybe next up: somemanifest.jsonlinting support?✅ Updates from last week
assert/is-almost-equal-float32arraystdlib#7682assert/is-almost-equal-complex64arraystdlib#7688assert/is-almost-equal-complex128arraystdlib#7687assert/is-almost-equal-arraystdlib#7690Anything else left to be implemented in this space, @kgryte?
Also, thanks to Philipp for merging:
math/base/special/cpolarfstdlib#7342math/base/special/kernel-sincosstdlib#5935🔭 My plan for the current week
math/base/special, currently focusing onsinpifandcospif.🙏 Challenges that I've faced and things I'd like help with
sinpif. Porting the C implementation from FreeBSD to JS was harder than I thought, especially dealing with signed/unsigned bit manipulation. I also misunderstood how the fast floor macro worked, which costed me about 3 hours on a silly bug. But now I know better! Hoping to open a PR soon.🏝️ My availability for next week
💬 Important items for discussion
Nothing at the moment. I'll ask if anything comes up while working on
sinpifandcospif.Thanks for reading!
kgryte commentedon Jul 22, 2025
@anandkaranubc The
is-almost-same-valuefamily of similar packages. E.g.,isAlmostSameFloat64Array,hasAlmostSameValues, etc. With these, we should have all necessary assertion utilities for testing. These can be gone in parallel to the ongoing single-precision work inmath/base/special.gururaj1512 commentedon Jul 22, 2025
🙌 Thanks I'd like to give
✅ Updates from last week
dztest2,sztest2&ztest2packages instats/base/ndarray/*dcovarmtk,scovarmtk&covarmtkpackages instats/base/ndarray/*🔭 My plan for the current week
🙏 Challenges that I've faced and things I'd like help with
🏝️ My availability for next week
💬 Important items for discussion