Talking Iodide yesterday
@mdboom how is scipy india?
@mdboom Sadly we had a technical difficulty thus we couldn't manage to record the event. But we recorded all the events prior to this: https://www.youtube.com/channel/UCyakf4xgcgarY2tnmlkg-xA
I hope to do an event(s) that is (are) solely focused on Iodide and I will make sure to get them recorded!
Reminder that we'll be having our usual weekly iodide meeting today at 9am PST / 12 pm EST / 4 pm UTC (that's about 25 minutes from now) on Zoom -- [Agenda], [Zoom Link] -- the formal agenda is short, feel free to come and ask questions
as an interesting aside, I wrote up a small iodide notebooks to do the gruntwork of generating the meeting template for these things -- https://definitely-staging.iodide.io/notebooks/225/?
@mdboom fiddling around with iodide-project/pyodide#504
what am i doing wrong in this NB?-- https://alpha.iodide.io/notebooks/3809/ . I expected it to return several matches including the variables defined further up, but the matches array has only two entries, the fist of which is "undefined" (i don't know what that means coming from Python...), and the second of which is just builtins.range
we appear to have a test failing on master:
FAIL src/editor/actions/__tests__/server-save-actions.test.js
● saveNotebookToServer › works as expected when there is no notebook on the server yet
expect(received).toEqual(expected) // deep equality
- Expected
+ Received
- Array [
- Object {
- "notebookInfo": Object {
- "notebook_id": 1,
- "revision_id": 1,
- "revision_is_latest": true,
- "serverSaveStatus": "OK",
- "tryItMode": false,
- "user_can_save": true,
- "username": "this-user",
- },
- "type": "UPDATE_NOTEBOOK_INFO",
- },
- ]
+ Array []
69 | undefined
70 | );
> 71 | expect(store.getActions()).toEqual([
| ^
72 | {
73 | notebookInfo: {
74 | notebook_id: 1,
at Object.<anonymous> (src/editor/actions/__tests__/server-save-actions.test.js:71:32)
anyone else see this failure, or do i have a bad master somehow? anyone work on related files recently?
@mdboom i have autocomplete working for js -- if you can add a method to the pyodide object such that i can call
window['pyodide']['completeRequest'], i think we can have Jupyter quality completions for Pyodide within a day
NM: It's related to my own hack I think...
And the iodide-side changes to make autocomplete language-agnostic: iodide-project/iodide#2539
I am working on the issue - iodide-project/pyodide#437 (instead of
I updated all the patches by running
1.16.2, I am upgrading it to 1.17.4 (latest release)).I updated all the patches by running
git diff between the versions (1.51.1 and 1.17.4) but I keep getting this error when I run make :(error: Linking globals named 'npy_sinl': symbol multiply defined!
shared:ERROR: Failed to run llvm optimizations:
Can anyone help me how can I go about this issue?
Not sure. I will have to check if it appears more than once.
Also, can you please explain why are we updating patches by taking
git diff? What is its use?
Looking at the issue again, I wonder if the contents of
config.h (in the packages/numpy/files) might need to be updated to support the new version of Numpy...
If you have any links I can read from, please let me know 😌
You may want to consider removing all of the patches to start, trying to build the new numpy and then adding the patches back in one-by-one as you hit what seem to be related errors.
Numpy generates these files at build time based on what it finds on the system. When compiling to Wasm, it can't actually do this -- it can't compile little test executables and run them to see what the system is capable of -- so we have to hard code it. It is mostly the output of what happens on x86_64, with a few changes as necessary to get it to compile.
@bcolloran : A new pyodide release is cut, so iodide-project/iodide#2539 should work now.
That's correct. What I meant was in a normal invocation of Numpy for native hardware, it generates those files. When building for webassembly, it can't do that so we have to provide them pre-made and hardcoded. They get included in the Numpy emscripten build because we pass
-I../../config as cflags from the meta.yaml file.