only mac and linux
#circuitpython-dev
slender iron
slender iron
ya, but usually simple to fix. got a link to the error?
slender iron
its being picky about white space
warm flame
lol
slender iron
mpconfigboard.h has an extra line at the end and the pins.c has some white space on the start of the line
you probably setup pre-commit in vscode and it'll autofix it
generally it's easiest to auto-run pre-commit before committing
warm flame
will try
slender iron
no need to close the PR, just push a new commit to the same branch
warm flame
ok. I am not used to github
slender iron
np
ornate breach
@slender iron paternity leave? congrats 🙂
slender iron
yup! thank you!
ornate breach
I missed that in taking my own paternity leave haha
slender iron
you'll need the empty board.c file in the folder too
(there may be other files as well)
warm flame
I'll check that later then
slender iron
and by empty, I mean it has functions that don't have contents
you should be able to copy it from another board
slender iron
good luck!
manic glacierBOT
orchid basinBOT
warm flame
is there a good tutorial to setup the workflow to add a board with VSCode ? I can't pass the pre-commit test
onyx hinge
@gilded cradle I think there's an issue with this entry in the 'features' of the qtpy esp32-s2 https://github.com/adafruit/circuitpython-org/blame/e402c8fd88a174e785186ea1703fbf5c34c8b325/_board/adafruit_qtpy_esp32s2.md#L15
@warm flame For the core, the documentation we have is https://github.com/adafruit/circuitpython/blob/main/BUILDING.md#code-quality-checks -- You didn't mention what OS you're using but Linux, Windows using WSL, or Mac should work as long as you follow the instructions in BUILDING.md. I am not aware of any vscode-specific instructions and I strongly suspect that Windows-without-WSL is going to have additional challenges. I always use the terminal on Debian Linux to do things like run pre-commit.
warm flame
even on windows ?
onyx hinge
@warm flame I use debian, the CI uses ubuntu.
warm flame
as it will be a Virtual machine, is there a debian fork that is more oriented toward DEV
warm flame
<@385537177838288898> I use debian, the CI uses ubuntu.
What do you think of Kali Linux ?
onyx hinge
I have not personally used it.
tulip sleet
@slender iron if you are eventually looking for smaller things to do, there are a couple of things for 7.3.0 that are in code familiar to you:
warm flame
I have not personally used it.
what desktop environement do you use with debian ?
onyx hinge
I use xfce, but that doesn't mean I necessarily recommend it. Gnome is probably a better suggestion for those new to Linux, it's more polished.
tulip sleet
I use cinnamon (with Ubuntu 22.04, but could be installed with Debian, I think).
Gnome takes up more space. I have a short taskbar at the bottom with lotsa icons
slender iron
@tulip sleet kk, I think I’ll try to get through all my email first. I made good progress on it yesterday
tulip sleet
<@329766224093249548> kk, I think I’ll try to get through all my email first. I ...
I agree on that!
slender iron
(On my phone now while holding the baby)
tulip sleet
your arm muscles will get stronger (speaking from experience)
i am not big into themes, etc. My desktop background is solid black 🙂
warm flame
@slender iron enjoy those moments, they are too short.
and nostalgia kicks in hard when they are 13 and are being Hades Bum Hole
ornate breach
and nostalgia kicks in hard when they are 13 and are being Hades Bum Hole
Sometimes it starts earlier, like 5 and 6 lol…
warm flame
Sometimes it starts earlier, like 5 and 6 lol…
Terrible 2, F**ing 4 and more
tulip sleet
pay no attention to these guys 😉
onyx hinge
vim but again this is not a recommendation. vscode is available for linux, so if you're comfortable with it choose it.
warm flame
vim is "dos" like right
tulip sleet
vim is "dos" like right
keyboard commands: it's a modal editor (command vs insert mode). I use emacs. If you are used to IDE's, atom, vscode, or PyCharm may be more up your alley
warm flame
pycharm is not only for python ?
orchid basinBOT
gilded cradle
<@478616919721771019> I think there's an issue with this entry in the 'features'...
Thanks for submitting a PR as well as coming up with a script that checks this.
onyx hinge
@gilded cradle you're welcome! after I found the 'frontmatter' pypi library it was not hard to write it
first I failed to find a way to do it in jekyll but that's not important since we can run other things in CI
gilded cradle
Yeah, I was thinking we had to do some pure actions thing. Writing a python script is much better.
orchid basinBOT
tulip sleet
pycharm is not only for python ?
it's part of the family of editors from JetBrains; I don't know how constricted it is
orchid basinBOT
idle owl
pycharm is not only for python ?
I use PyCharm for all kinds of things. It works with many more things than Python.
It doesn't do syntax highlighting on the ones it can't handle, if I remember correctly. But it does open them.
warm flame
thanks
idle owl
@lone axle Do you have a moment to review this? I'm blocked on it for a task I'm doing. I tested it thoroughly, so I don't feel like it needs testing. Plus, I'm not sure how many folks have more than one of these anyway. (Though I'm pretty sure no one tested the original and it didn't work.) https://github.com/adafruit/Adafruit_CircuitPython_VL53L1X/pull/6
lone axle
<@382939733107408897> Do you have a moment to review this? I'm blocked on it for...
yep I'll take a look.
idle owl
Thanks so much! I just pushed another change, but it should pass as well.
lone axle
Thanks so much! I just pushed another change, but it should pass as well.
Looks good to me, left review. Want me to merge while I'm there?
idle owl
Looks good to me, left review. Want me to merge while I'm there?
Yes please!
Thank you so much!
idle owl
@gilded cradle Is there some issue with platform detect? Getting the following error trying to build sphinx docs locally: Traceback (most recent call last): File "/Users/kattni/.virtualenvs/pylint/lib/python3.10/site-packages/sphinx/ext/autodoc/importer.py", line 70, in import_module return importlib.import_module(modname) File "/Users/kattni/.pyenv/versions/3.10.0/lib/python3.10/importlib/__init__.py", line 126, in import_module return _bootstrap._gcd_import(name[level:], package, level) File "<frozen importlib._bootstrap>", line 1050, in _gcd_import File "<frozen importlib._bootstrap>", line 1027, in _find_and_load File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked File "<frozen importlib._bootstrap>", line 688, in _load_unlocked File "<frozen importlib._bootstrap_external>", line 883, in exec_module File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed File "/Users/kattni/repos/Adafruit_CircuitPython_Simple_Text_Display/adafruit_simple_text_display.py", line 26, in <module> import board File "/Users/kattni/.virtualenvs/pylint/lib/python3.10/site-packages/board.py", line 166, in <module> elif board_id == ap_board.KHADAS_VIM3: AttributeError: module 'adafruit_platformdetect.constants.boards' has no attribute 'KHADAS_VIM3'
Or is that maybe a red herring.
Force reinstalled it and it fixed that error. Nevermind, apologies for the ping.
Ugh, ok, Sphinx issue. Docs are building properly locally, including the API page. Remotely, they're passing when they build, but the API page is blank. I have no idea where to start with this.
gilded cradle
Ugh, ok, Sphinx issue. Docs are building properly locally, including the API pag...
Where are you at with this? Different error or same?
idle owl
Where are you at with this? Different error or same?
No more errors involved, which is what's frustrating. I can build it locally, but it's building wrong remotely (even though it passes)
gilded cradle
No more errors involved, which is what's frustrating. I can build it locally, bu...
Oh ok. Did something change in the docs content-wise?
idle owl
Nothing changed.
gilded cradle
Hmm. Same version of sphinx as you have locally?
idle owl
Hmm, I didn't verify that. Checking.
gilded cradle
ok
idle owl
Looks like I was behind. Building again now locally.
Still building properly locally with the latest Sphinx. 😦
gilded cradle
Still building properly locally with the latest Sphinx. 😦
Ok, I don't have any other ideas at the moment, but will let you know if I do.
idle owl
Thanks. Appreciate the help.
slender iron
@tulip sleet all of my reload changes broke the ble reload. think I should PR a fix into 7.2?
slender iron
what do others think? should we fix ble reload in 7.2?
stuck elbow
@slender iron I know you are still catching up, but you did say yesterday to ping you if there is something we wait on. I'd love to be able to get the PewPew LCD merged before 7.3 is out, so if you have time, could you take a look at https://github.com/pidcodes/pidcodes.github.com/pull/733 ?
slender iron
ya, I can do a pid.codes pass now
idle owl
@tulip sleet Can you get me a build for Circuit Playground Bluefruit from that 7.2.x PR so I don't have to wait on CI to build it?
slender iron
I can
idle owl
Thanks!
slender iron
the transfer works but then the download button stops working
tulip sleet
<@329766224093249548> all of my reload changes broke the ble reload. think I sho...
Just got back from a walk. 7.2.x is fine, though there are not that many users, so could be main, since I think we could release 7.3.0 quite soon. If we can't fix the I2C issues quickly, for instance, I would just as soon release 7.3.0 to get all the recent changes into stable.
idle owl
the transfer works but then the download button stops working
There is no Transfer or Download anymore
I'll test it with the latest version
slender iron
well, its download after it is run it
slender iron
I'll merge 7.2.x into main after the PR is in
and we can decide what to actually release later
idle owl
@slender iron It works great with PyLeap! Fixed the issue with needing to click reset to get the code to run.
manic glacierBOT
tulip sleet
hurrah - thanks!
idle owl
Oh there's the new issue you reported.
slender iron
there had been a lot of churn with the reload code
idle owl
Where it "stops" during downloading. Even though it downloaded and ran the first program I tried
slender iron
yup
idle owl
Ok. I'll file the bug there, and close the bug related to the reload
idle owl
OK, so restarting the app made the button work again, but now it isn't loading anything new.
idle owl
the cp issue will be closed with a merge into main
There's a bug on PyLeap for the reload thing too. Not sure if that will be closed.
slender iron
ah, I don't think it will be
idle owl
So basically, it only works on the first try, then never again.
@onyx hinge Your PR failed on the same issue again?
idle owl
Heh fair enough.
@onyx hinge Merged!
@slender iron @tulip sleet So this will be a 7.2.6 release?
tulip sleet
we might not do a 7.2.6, because I think we will do 7.3.0 final pretty soon. And there are not a lot of people using PyLeap
tulip sleet
I could easily do a 7.2.6, though. I want to test a change I made in the release process anyway. I mean, is it true the audience for this is very small?
idle owl
I believe so, yes. But Trevor doesn't venture into beta builds, and at the moment, the other testers aren't either. So he's concerned about putting a warning into the app about hitting reset to get the code to run.
tulip sleet
then I can do a 7.2.6, if you think it would help that process.
idle owl
But this CP update causes a bunch of issues in PyLeap anyway, and Trev will need to experience them to fix them.
I think it might, yeah.
slender iron
trev should use the absolute latest
idle owl
I'll talk to him about it then. But if this is in 7.2.x, it won't be in latest, will it?
Also, if he doesn't use stable, then when PyLeap gets released, we'd have to ask users to use absolute latest as well.
tulip sleet
how imminent is PyLeap release?
I think 7.3.0 will get released within a week or two
idle owl
Ok.
tulip sleet
let's try to get Trev to use 7.3.0-beta.2 and absolute newest
idle owl
Ok I'll chat with him.
tulip sleet
it has to get tested anyway against PyLeap to make sure they work together
slender iron
@stuck elbow one nitpick is the hardware repo should link to the software branch
stuck elbow
<@351136009922019331> one nitpick is the hardware repo should link to the softwa...
you mean the pew library or the circuitpython repo?
slender iron
the pewpew-lcd repo should link to CP
stuck elbow
the library gets compiled in
slender iron
or your fork
stuck elbow
I link to circuitpython.org, but I guess I should link to the code directly, or even to the pull request?
slender iron
PR is fine
proven garnet
One thing I realized when I got a PR to my community bundle library is that it had not received the update to .pre-commit-config.yaml updating the black rev. I was able to fix it since I knew what the problem and fix was, but is there a way to do something like make updates available to libraries? Like something you could connect to via git and pull infrastructure changes like that from?
I could see it being particularly aggravating for someone that didn't realize what had happened with a repo that had been cloned from cookiecutter before the fix.
idle owl
I could see it being particularly aggravating for someone that didn't realize wh...
Jeff just ran into that with an Adafruit library. I'm not sure how we can automate that.
idle owl
@slender iron Can you let me know when that fix goes into main, so I can let Trev know? He's got the beta running on one CPB, I'm having him run latest on a second one, and test both.
stuck elbow
PR is fine
added, thanks
slender iron
@idle owl will do. probably tomorrow in another PR
idle owl
Ok, sounds good.
proven garnet
Jeff just ran into that with an Adafruit library. I'm not sure how we can automa...
I have an idea to try out, I can report back on how it goes 😁
idle owl
I have an idea to try out, I can report back on how it goes 😁
Fair enough! Sounds good
@tulip sleet Do you have a few minutes to maybe help me parse some guide feedback? I think it's an Arduino problem, but I'm uncertain. Has to do with PWM on QT Py SAMD21. Jeff helped me a bit, but we thought you might know better.
tulip sleet
<@329766224093249548> Do you have a few minutes to maybe help me parse some guid...
sure
idle owl
Vid might be fastest.
orchid basinBOT
idle owl
What is the exhaustive list of things that mpy-cross removes when a library is converted to a .mpy? I know essentially what it does, but I don't think I've ever asked for a canonical list in one question. That I can find anyway.
stuck elbow
comments, docstrings and type hints, I think
idle owl
Whitespace?
stuck elbow
no whitespace in bytecode
idle owl
Ok
stuck elbow
also any globals that begin with _ and have value set using an integer expression wrapped in const()
idle owl
Hmm
stuck elbow
they get replaced with the value of that expression
idle owl
Oof. Ok. Trying to explain what happens to a .py file when you run mpy-cross on it for a guide page.
stuck elbow
it gets compiled to bytecode, which is similar to machine code that normal program binaries use
https://opensource.com/article/18/4/introduction-python-bytecode <-- this might help?
idle owl
it gets compiled to bytecode, which is similar to machine code that normal progr...
And that compilation removes comments, docstrings, type hints, whitespace, and replaces any _ globals that have a value set using an int inside const() with the value. Yes?
stuck elbow
it's not like it removes those things, more like they don't get translated to the bytecode
idle owl
Ahhh. Ok
stuck elbow
there is no bytecode for a comment or for whitespace
the docstrings are removed, though
there is an option to not remove them, iirc
idle owl
Fair enough.
stuck elbow
the consts are simply replaced with their value, but only if they are "private" -- when they start with _
idle owl
Yeah ok
stuck elbow
if they are not private, they are left in there, because something might want to import them
idle owl
Ah ok, that makes sense.
stuck elbow
https://towardsdatascience.com/understanding-python-bytecode-e7edaae8734d <-- this is a better and more detailed explanation than the previous link
it's a bit like assembly
idle owl
The first one didn't seem all that helpful, so another more detailed one might help.
I only know Python. Heh.
I mean I can find my way around Arduino, and I've submitted to the CP core...
But I usually need help.
This link is much better.
stuck elbow
right, arduino compiles to machine code, which is a binary that can run directly on the microcontroller
stuck elbow
mpy-cross compiles to bytecode, which is a binary that can run directly on the python virtual machine
idle owl
And CP has the same concept of a virtual machine? I've heard Scott reference it a number of times, so I guess yes.
stuck elbow
python virtual machine is like an abstract computer that is the same no matter if you run python on an intel or arm or, say, risc-v
idle owl
Neat ok
stuck elbow
cp has a different virtual machine and slightly different bytecode than regular python
when you have to match the .mpy to the python version, that is because the virtual machine has changed and expects different bytecode
like, for example, new commands have been added
idle owl
Ohhh right.
I never really knew why we needed to change mpy version, I only knew that we needed to.
So would you say the result is a file that is smaller and more optimised? And it takes up less.... RAM and flash? or only flash.
stuck elbow
it's smaller, but that's not the reason why it uses less memory
idle owl
Oh
stuck elbow
the thing is, to execute the python code, the python interpreter has to compile it to bytecode anyways, always
idle owl
I saw that in the link you sent
stuck elbow
what runs on the vm is always bytecode
so if you give it a python file, it has to compile it, and keep the compiled bytecode in memory while it is being executed
but if you give it an mpy file, it can just read it directly from the flash and execute it, without using additional memory, and without having to compile it
idle owl
Ohhhh
stuck elbow
so there are two things that use that extra memory: the compilation process itself, which needs to keep some extra variables while it's compiling, and the resulting bytecode, which needs to be stored in memory during and after the compilation
idle owl
This is clearing up a lot of things I've always kind of glossed over.
stuck elbow
this is why mpy files can have much more code in them
they are also slightly faster to start, since there is no compilation step, but usually you can't really tell the difference, with all the other delays that CP has on startup
stuck elbow
enjoy
onyx hinge
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1fc8UvMSBTq0x2Sh94_R6pSRm1v81sZCWA4L7TwsFzfg/edit?usp=sharing
manic glacierBOT
slender iron
mpy byte code is loaded into RAM to run. It isn't run from flash. So, it saves memory usually used in the compile step but not after. (Almost, the mpy process may do a little optimization (const) which could save a little RAM.)
@idle owl don't give too many specifics about mpy because it does change from time to time
orchid basinBOT
late cradle
@slender iron hey 👋 wasn't sure which channel to use (let me know if we should switch) but I'm coming from issue #5986 -- not sure what you meant by "iMX", but if you point me to the right direction I'm happy to start tinkering when I find free time!
jaunty juniper
<@252717193496756235> hey 👋 wasn't sure which channel to use (let me know if w...
it's one of the supported ports, mostly the Teensy boards as far as I know
https://circuitpython.org/downloads?q=imx <--- boards
https://github.com/adafruit/circuitpython/tree/main/ports/mimxrt10xx
https://github.com/adafruit/circuitpython/tree/main/ports/mimxrt10xx/common-hal/usb_host
late cradle
ah right thanks, I'll dig around a bit to try to get a better understanding of all the pieces involved
jaunty juniper
(Adafruit's M7 boards and development have been postponed because of chip shortage)
manic glacierBOT
manic glacierBOT
onyx hinge
github doesn't seem to have a way to note yourself as 'on vacation', either to reduce notifications or to let others know you won't be available to respond on issues/pull requests/review requests .. am I missing something?
I guess I can "set status", where is that shown...?
oh yes I can set myself 'busy'. yay.
blissful pollen
oh yes I can set myself 'busy'. yay.
Busy having fun I hope! Have a great vacation
onyx hinge
#ifndef MP_WARN_CAT
#define MP_WARN_CAT(x) #x
#endif
```oh good, micropython can warn the catI think this lets you write MP_WARN_CAT(stay off the table) though it may have no effect at runtime
jaunty juniper
oh interesting quirk of documentation:
_MCP23017_ADDRESS = const(0x20)
...
class MCP23017(MCP230XX):
...
def __init__(
self, i2c: I2C, address: int = _MCP23017_ADDRESS, reset: bool = True
) -> None:
leads to the docs:
class adafruit_mcp230xx.mcp23017.MCP23017(i2c: busio.I2C, address: int = micropython.const, reset: bool = True)
So the default address is "micropython.const"...
Worth not using a const for just that one for clearer documentation ?
tulip sleet
oh interesting quirk of documentation:
```py
_MCP23017_ADDRESS = const(0x20)
......
maybe we could add an explicit docstring which would give the value?
stuck elbow
or just put the value in there directly, it's not like it's used anywhere else
jaunty juniper
it is missing :param:(s) in the docstring
tulip sleet
it would be good to add :param:s anyway, and @stuck elbow 's point is good too
I think there are a number of libraries like this
jaunty juniper
I think it does that for micropython.const only, not just a normal UPPERCASE variable
tulip sleet
if you remove the underscore, does it change the signature?
stuck elbow
I wonder if the const could inherit from int in the stubs...
stuck elbow
if you remove the underscore, does it change the signature?
problem is, this is running in normal python, so it takes const from stubs, not the magic behavior of micropython's const
the docs compilation, I mean
I wonder if we could just set __repr__ on the const in the stub, to make it display the value passed
jaunty juniper
tulip sleet
because it's declared as int? That's an easy fix
jaunty juniper
a behaviour I fail to reproduce in the MCP23017
idle owl
<@330227457296957440> don't give too many specifics about mpy because it does ch...
I'll deal with that when it happens. This guide page is entirely meant to explain library file types, including mpys, and we get these questions a lot, and someone has to reiterate all of the info deshipu gave me every time it comes up. Ideally, we have it in a guide, and we can point folks to it instead.
tulip sleet
also no underscore. It doesn't have to be const(), except to make the variable not be in the compiled dictionary
slender iron
I do know mpy will change with MP 1.19
tulip sleet
a behaviour I fail to reproduce in the MCP23017
or maybe the tools versions you're using are not the same as the CI??
idle owl
I do know mpy will change with MP 1.19
I'll keep that in mind. Thanks for the heads up.
idle owl
Ok. What would change then? I'm not entirely sure what changed specific to the versions when we had to switch to a new version previous to this.
slender iron
the mpy version number has to do with the file structure largely. in the past it changed so that mpy could hold native code and more efficiently load strings
the next change is largely so that mpy code can be run directly from flash (with the huge caveat of having a specific filesystem to store it)
jaunty juniper
or maybe the tools versions you're using are not the same as the CI??
you'll never guess
I REMOVED micropython from autodoc_mock_imports
and poof it worked
(the AHT library didn't have it mocked)
no other modification was required
idle owl
the next change is largely so that mpy code can be run directly from flash (with...
Ahhh ok. Out of curiosity, is that specific filesystem something we will have? Or is that not something we can/will do.
slender iron
I don't think we will soon at least
jaunty juniper
oh that's a bummer
stuck elbow
as soon as Apple supports MTP
slender iron
the new format needs a filesystem that stores the file linearly
because it uses XIP to run the code
stuck elbow
right, but mtp would free us to use any filesystem we like
jaunty juniper
Eventually it will be possible to store such .mpy files in a linear, read-only, memory-mappable filesystem so they can be executed from flash/ROM.
oh they don't have one yet then ?
idle owl
Ok fair enough.
stuck elbow
@idle owl then I suppose I lied to you when I said mpy are executed from flash -- sorry
the frozen modules are, though
right?
jaunty juniper
yep
idle owl
<@330227457296957440> then I suppose I lied to you when I said mpy are executed ...
No worries! That's why having that conversation in this channel was helpful, because others can provide their knowledge as well.
slender iron
@tulip sleet let me know when you are around for a chat. I've got another issue to fix with ble reload
proven garnet
No worries! That's why having that conversation in this channel was helpful, bec...
For what it's worth it was also just fascinating to follow along
idle owl
For what it's worth it was also just fascinating to follow along
I should have considered that angle too! I feel the same a lot of the time, so I know what you mean.
idle owl
@tulip sleet Guide feedback saying that my explanation of the flash usage for CircuitPython is incorrect on ESP32-S3... 8MB of flash, how much is used for firmware, and how much for CIRCUITPY? I checked the S2, and it's 1MBish for CIRCUITPY, which is what I said in the guide. Am I missing something?
jaunty juniper
<@329766224093249548> Guide feedback saying that my explanation of the flash usa...
that was me, the S3 has 8MB, while the S2 has 4
jaunty juniper
yup
onyx hinge
@idle owl which board specifically? It looks like maybe any part with 8MB flash gets 2048KiB for firmware and 3776KiB for CIRCUITPY.
jaunty juniper
yup that's the 8MB partition scheme in tinyuf2
idle owl
Plugged in an S3. CIRCUITPY is 3.8MB.
Good catch. Thanks!
So this is now accurate? The 8 MB of flash is inside the module and is used for both program firmware and filesystem storage. For example, in CircuitPython, we have 4 MB set aside for program firmware (this includes two OTA option spots as well) and an approximately 3.8MB section for CircuitPython scripts and files.
I had finished all my guide feedback, refreshed and there was a new one. Had to squash it! Must reach 0! 😄
jaunty juniper
the original poster on the forum who mentioned the other issue with the Feather S3 page also noticed that the schematics uses a bloc for the S3 that has two "DAC" pins, it's a really minor thing, but if Limor feels like updating it 🤷
jaunty juniper
it's deep in the schematic, I don't think anyone is gonna be confused by it
tulip sleet
you'll never guess
great sleuthing
tulip sleet
<@329766224093249548> let me know when you are around for a chat. I've got anoth...
i was out for a walk and am back now
idle owl
Still fighting (as I do other things) with this Sphinx issue. Docs passing and building locally and remotely. However, the API page is blank remotely, and shows up locally. No changes were made, the docs build under the GitHub Actions for the latest change to the remote repo shows exactly the same Sphinx build output as my local setup does. Tried rebuilding through RTD, no change remotely. So I have no explanation or idea where to start on this one. No errors to work with, no obvious direction. https://github.com/adafruit/Adafruit_CircuitPython_Simple_Text_Display
onyx hinge
@idle owl by viewing the raw log something interesting becomes visible but I'm not sure what it means yet. ```
WARNING: autodoc: failed to import module 'adafruit_simple_text_display'; the following exception was raised:
Traceback (most recent call last):
File "/home/docs/checkouts/readthedocs.org/user_builds/adafruit-circuitpython-simple-text-display/envs/latest/lib/python3.10/site-packages/sphinx/ext/autodoc/importer.py", line 62, in import_module
return importlib.import_module(modname)
File "/home/docs/.asdf/installs/python/3.10.0/lib/python3.10/importlib/init.py", line 126, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File "<frozen importlib._bootstrap>", line 1050, in _gcd_import
File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 688, in _load_unlocked
File "<frozen importlib._bootstrap_external>", line 883, in exec_module
File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
File "/home/docs/checkouts/readthedocs.org/user_builds/adafruit-circuitpython-simple-text-display/checkouts/latest/adafruit_simple_text_display.py", line 28, in <module>
import terminalio
File "/home/docs/checkouts/readthedocs.org/user_builds/adafruit-circuitpython-simple-text-display/envs/latest/lib/python3.10/site-packages/terminalio.py", line 26, in <module>
FONT = fontio.BuiltinFont()
File "/home/docs/checkouts/readthedocs.org/user_builds/adafruit-circuitpython-simple-text-display/envs/latest/lib/python3.10/site-packages/fontio.py", line 31, in init
self._font = ImageFont.load_default()
File "/home/docs/checkouts/readthedocs.org/user_builds/adafruit-circuitpython-simple-text-display/envs/latest/lib/python3.10/site-packages/PIL/ImageFont.py", line 361, in load_default
f._load_pilfont_data(
File "/home/docs/checkouts/readthedocs.org/user_builds/adafruit-circuitpython-simple-text-display/envs/latest/lib/python3.10/site-packages/PIL/ImageFont.py", line 107, in _load_pilfont_data
image.load()
File "/home/docs/checkouts/readthedocs.org/user_builds/adafruit-circuitpython-simple-text-display/envs/latest/lib/python3.10/site-packages/PIL/ImageFile.py", line 241, in load
n, err_code = decoder.decode(b)
SystemError: PY_SSIZE_T_CLEAN macro must be defined for '#' formats
looking for now-outdated files... none found
maybe see what happens if you switch it back to python3.9 for building on rtd https://docs.python.org/3/c-api/arg.html#strings-and-buffers
I'd also just consider asking readthedocs, they may either know an answer or view it as something they need to fix
idle owl
I don't see where to switch the version of Python used. I know it's a thing, because we dealt with it recently, but I'm not sure where it is.
I'm wondering if it was something on Sphinx's end.
I guess I can ask RTD.
onyx hinge
.readthedocs.yaml
onyx hinge
/home/docs/checkouts/readthedocs.org/user_builds/adafruit-circuitpython-simple-text-display/envs/latest/bin/python -m pip install --upgrade --no-cache-dir mock==1.0.1 pillow==5.4.1 alabaster>=0.7,<0.8,!=0.7.5 commonmark==0.8.1 recommonmark==0.5.0 sphinx sphinx-rtd-theme readthedocs-sphinx-ext<2.2
it's installing a specific version of pillow. This problem was fixed in pillow 6.0.0.
idle owl
Yeah ok
onyx hinge
I thought maybe it was the blinka version of displayio but it's not.. https://github.com/adafruit/Adafruit_Blinka_Displayio/blob/main/setup.py#L39=
onyx hinge
np
onyx hinge
what did you try?
idle owl
Looking deeper.
Tried setting pillow version in the docs/requirements.txt.
Which it does install from after grabbing pillow-5.x
but apparently that does nothing
onyx hinge
maybe there are two problems, because the specific error I saw earlier isn't in the output now .. but it still made no difference to having the API page appear
you're just looking in the sidebar for it right?
idle owl
I clicked "simpletest" and then clicked "next" on the bottom of that page
which should go to API
goes to "no title" now.
onyx hinge
https://docs.circuitpython.org/projects/simple-text-display/en/latest/api.html is there for me now
idle owl
oh!
Hard refresh and it's there for me too!
Hurrah. I guess it did install a later version of Pillow.
onyx hinge
it probably installed one and then the other
idle owl
I reverted the change to the version of Python used
so it's only me adding a specific version of pillow.
onyx hinge
Attempting uninstall: pillow
Found existing installation: Pillow 5.4.1
Uninstalling Pillow-5.4.1:
Successfully uninstalled Pillow-5.4.1
Successfully installed pillow-9.1.0
yeah it installs the one then replaces it with the other.
onyx hinge
@gilded cradle WDYT of adding a >= 6.0.0 requirement to pillow here, would it negatively affect platforms we want to support or would they have the necessary version available?
We may be able to write 'any pillow version, but at least 6 if python 3.10 is installed', I'd have to check how requirements.txt deals with a situation like this pillow pillow >= 6.0.0; python_version>=3.10 (any version of pillow, unless python 3.10 or above is installed, in which case at least version 6.. is the intent)
onyx hinge
raspbian oldstable has 5.4 (and python 3.7)
gilded cradle
<@478616919721771019> WDYT of adding a >= 6.0.0 requirement to pillow here, woul...
What does adding pillow as a requirement do? It's an image manipulation library.
onyx hinge
It's already a requirement of https://github.com/adafruit/Adafruit_Blinka_Displayio/blob/main/requirements.txt I just want to get a version that's compatible with python 3.10
sorry I was unclear on that point (where I wanted to add it)
gilded cradle
Sure, we can change that to >= 6.0.0.
onyx hinge
my syntax for applying it based on Python version is wrong
pillow>=6.0.0; python_version>'3.9'
I'm uneasy because that LOOKS like a string comparison but maybe it's not
gilded cradle
I'm not sure what the pip syntax is
onyx hinge
so I just tested this in docker python:3.9 and docker python:3.10: ```pip install 'pillow==5.4.1'; pip install 'pillow' "pillow>=6.0.0; python_version>'3.9'"
the install of pillow==5.4.1 first is to copy what RTD is doing, that causes us problems.
so I feel like this is relatively safe
onyx hinge
oh yes, I guess so!
proven garnet
Oh, good I just came here to talk/ask about this 😅
idle owl
There are others that aren't rendering properly. No idea if it's the same issue.
onyx hinge
now you can look in the raw log for that error though!
I'd hold off fixing any more if melissa is going to change blinka displayio though
I mean, fixing them by changing the python version
idle owl
I'd hold off fixing any more if melissa is going to change blinka displayio thou...
Right. I agree.
idle owl
Ok, so.... a .py file vs a .mpy file. A .py file on a smaller micro will result in memory allocation failures? As well as taking up more flash, but that's typically not the issue. I understand why .py requires more memory, but I want to make sure I'm explaining this properly.
Or "can" result in anyway.
jaunty juniper
a .py file requires more memory when imported, so if there's not much memory remaining the import is more likely to fail
jaunty juniper
because the process of turning the py text into the internal representation (bytecode) uses memory
stuck elbow
an mpy just needs to be loaded into memory, a py needs to also share that memory with the running compiler
this is why sometimes the order of importing files matters
you want to import the big ones that require more memory during compilation first, while there is still memory free
idle owl
Yeah I had a few sentences on why .py requires more memory from yesterday. But I was struggling to connect "what is a .py file" with "uses extra memory". These things are the connection.
jaunty juniper
oh @idle owl, I believe there was a conversation about a font repository a while ago. I mentioned this to @lone axle during a stream recently, I think it would be really interesting to mention U8g2 as a repository of fonts, (under some type of open license), in the fonts guide.
there's a bunch: https://github.com/olikraus/u8g2/wiki/fntgrp
https://github.com/olikraus/u8g2/tree/master/tools/font/bdf
Maybe even somebody would have the courage to go through them and pick a few at different sizes that look good and can be showcased for people to select.
idle owl
Hmm. How often is it updated? The issue with linking to other folks' repos is if it gets deleted etc in the future. If it's not updated that often, we could fork it, and link our fork... But it might be fine? If @lone axle is up for it, I think that might be a good idea. Is that the repo you always link when asked that question?
Need to jet for dinner. Be back in a bit.
I would like to know where the sys.path() for libs is defined when I get back through. 🙂
jaunty juniper
I only discovered it recently, it seems active, it's a "monochrome graphics library for embedded devices", the fonts are only part of it, some under MIT, others under creative commons, etc. We could totally mirror them as long as that's ok with the license
idle owl
I only discovered it recently, it seems active, it's a "monochrome graphics libr...
Fair enough. We've linked to other folks' repos before, so it's not unprecedented, but opting to fork it means we have our own copy if they disappear for any reason.
@tulip sleet Where is sys.path() in terms of the library file order? As in where is the code telling CP where to look? I thought I knew, but apparently I don't.
tulip sleet
<@329766224093249548> Where is `sys.path()` in terms of the library file order? ...
>>> import sys
>>> sys.path
['', '/', '.frozen', '/lib']
Ohhhhhh. Right.
tulip sleet
if nothing is frozen, .frozen is not in the path
tulip sleet
>>> sys.path
['', '/', '/lib']
Nifty.
tulip sleet
e.g. on Qt Py M0
idle owl
Ah ok
tulip sleet
i only know that because I happened to try it on the board I currently had plugged in
idle owl
Hahaha. Nice
stuck elbow
nobody knows how this works anymore, we are doomed
idle owl
It has become sentient.
jaunty juniper
by the way there's like 25 occurences of "micropython" being in "autodoc_mock_imports" in the bundle (excluding comments)
idle owl
@tulip sleet What is the first '' in the sys.path? I thought it started in root /, then checked .frozen if it's there, and then /lib... So what is the ''?
tulip sleet
current directory, I think
https://docs.python.org/3/library/sys.html#sys.path
As initialized upon program startup, the first item of this list, path[0], is the directory containing the script that was used to invoke the Python interpreter. If the script directory is not available (e.g. if the interpreter is invoked interactively or if the script is read from standard input), ***path[0] is the empty string, which directs Python to search modules in the current directory first. ***Notice that the script directory is inserted before the entries inserted as a result of PYTHONPATH.
idle owl
Hmm. So '' for us is CIRCUITPY as well?
idle owl
Ok got it. Thanks!
orchid basinBOT
manic glacierBOT
For LC709203F, I am seeing bad handling of a clock stretch. The top trace is ESP-S3, the bottom trace is ESP-S2. The first four I2C transactions are the same. Then there is a clock stretch, which works on the S2, but not the S3. Note that SDA goes high after a relatively short time in the top (S3, bad) trace.
warm flame
building my debian workframe... virtualbox is being a Hades Pit with me
orchid basinBOT
manic glacierBOT
orchid basinBOT
orchid basinBOT
onyx hinge
I tried out pre-commit.ci on two of my personal projects .. https://results.pre-commit.ci/run/github/77753953/1652368838.qvt34fwsRSiYze0oigbXMQ https://results.pre-commit.ci/repo/github/293359557
It's too bad their URLs are opaque. It does run fast, but in one of the two cases I ran into trouble where pre-commit behaved different than locally OR on github actions.
Here's an example of it pushing to a pull request: https://github.com/jepler/wwvbpy/pull/31 https://results.pre-commit.ci/run/github/77753953/1652369245._qJ40VmDTM2jffJpA_bWNQ
orchid basinBOT
warm flame
I hope it will be accepted this time
jaunty juniper
I hope it will be accepted this time
you know you can push to the same branch to update a PR
jaunty juniper
well now you know !
stuck elbow
for a moment ;-)
warm flame
lol
pre-commit failed again
but in vscode it told me all was ok in my pre-commit
go figure
jaunty juniper
I can do suggestions in the PR that you can then accept in the github UI
that would then fix it
stuck elbow
a leftover tab
jaunty juniper
there should be a button to "accept changes" or something like that in the github UI
warm flame
ok
jaunty juniper
what software did you use to create the PR ?
warm flame
github directly
jaunty juniper
ah, if you use the github app for example you can modify your code and push more changes to the PR to fix pre-commit issues and stuff
warm flame
cool cool
idle owl
@tulip sleet For sys.path, is it weird to do print(sys.path) in the REPL? vs simply sys.path which does the same thing. I'm asking because in the guide I suggested that you could run the two lines of code in code.py or the REPL, but in code.py it would need to be print(sys.path). So I figured include the REPL example, but use print() so it would work in code as well.
warm flame
My prog workflow is dating from before 2005 and the other one I know is for PLC (automation) which is also outdated
tulip sleet
<@329766224093249548> For `sys.path`, is it weird to do `print(sys.path)` in the...
it's fine, if you want a consolidated example. If you don't say print() then anything that's None will not print anything, but that's not going to happen here.
idle owl
it's fine, if you want a consolidated example. If you don't say `print()` then a...
I don't quite understand what you mean.
jaunty juniper
there's a small difference between print(x) and x in the REPL, relative to the automatic use of repr() for strings, but that's fine I think
jaunty juniper
>>> "1234"
'1234'
>>> print("1234")
1234
>>> None
>>> print(None)
None
So basically for this example, it's fine, but don't assume it is always that way?
jaunty juniper
for a list, there would be no difference
idle owl
Ok thanks
warm flame
I'm actually nervous about that pull request
idle owl
I'm actually nervous about that pull request
Why?
idle owl
We all started somewhere, and we all had our first PRs, and our first complicated PRs.
We're all here to help out with it to make sure you're able to contribute.
I understand being nervous, I was too. But there's no need. We understand.
warm flame
I also have general anxiety disorder ...
idle owl
That I also understand.
warm flame
so yeah pretimuch anything makes me nervous
jaunty juniper
it passed pre-commit, it's building
idle owl
Well, once this is sorted, we can be excited for you. 🙂
warm flame
---> Cannot find file ../ports/espressif/build-Lilygo-T-oi-Plus-ESP32C3/firmware.uf2
this is making me pannic
lol
onyx hinge
I can empathize with nervousness about waiting for the computer to say "no". It does it without a hint of empathy. But it's not capable of any. We are, though, or aspire to be.
idle owl
this is making me pannic
This is the Update ports/espressif etc PR, yes? You're Simon?
idle owl
The build hasn't failed yet. Sometimes things that look like failure warnings are standard parts of the process.
onyx hinge
@warm flame so you're right that the line about not finding .uf2 is important
esp32-c3 can't use uf2 (as far as I know), so it is necessary to add the board to a list of exceptions
jaunty juniper
it would be nice to have defaults per variant or whatever it's called
onyx hinge
look in tools/build_board_info.py and find the lines like ```python
# esp32c3
"adafruit_qtpy_esp32c3": BIN,
And yeah it would be better if this were more automatic like neradoc suggests! but as of now it's not
that's around line 72 in my copy
jaunty juniper
And yeah it would be better if this were more automatic like neradoc suggests! b...
warm flame
ok
idle owl
Hey. @tulip sleet ... I just realised I was running sys.path on the Circuit Playground Bluefruit, and still got this: ```Adafruit CircuitPython 7.2.5-1-g8d7390992 on 2022-05-10; Adafruit Circuit Playground Bluefruit with nRF52840
import sys
print(sys.path)
['', '/', '.frozen', '/lib']```
I don't see any frozen libraries in that build.
Also, where is .frozen anyway? It's not a visible directory, so I should probably explain the concept...
tulip sleet
it's a fake directory; the frozen stuff is not really in the filesystem. I'm trying to remember which board didn't show frozen stuff at all.
jaunty juniper
you can't os.listdir(".frozen") for example
idle owl
right
jaunty juniper
the pico has no .frozen, is it a RP2040 thing ?
idle owl
I read up on sys.path and for the first time, wondered how it actually worked with .frozen. It simply made sense before, .frozen means the frozen modules.
But now knowing what sys.path does, I questioned it for the first time.
tulip sleet
I was looking on a QT Py SAMD21, hmm
idle owl
That's what I thought.
tulip sleet
the "original" QT Py
idle owl
Yeah.
Do you have a Trinket or GEMMA handy?
Maybe it's a lack-of-flash thing?
Stretching here.
tulip sleet
yes, I'm going to check that
onyx hinge
I think that in 7.2.5 it is listed on all boards, but in 7.3.0 it's only listed if used
@tulip sleet @idle owl
idle owl
Ohhhhhh.
Hmph.
I guess I'll pretend folks are using 7.3.0 and ignore any feedback between now and then about the page being wrong.
tulip sleet
i have no memory of such a PR, hmm
onyx hinge
It came with a micropython merge is my guess
idle owl
I think that in 7.2.5 it is listed on all boards, but in 7.3.0 it's only listed ...
Good call. Thank you!
tulip sleet
@idle owl I cannot reproduce LC709203F problems on ESP32-S2, but it's completely reliable to reproduce on S3: https://github.com/adafruit/circuitpython/issues/6311
idle owl
<@330227457296957440> I cannot reproduce LC709203F problems on ESP32-S2, but it'...
Oof. Yeah you mentioned that.
I guess at least I'm not completely off?
tulip sleet
there is definitely a problem on S3, it's obscure, I can't make it happen when using the Arduino library. It is clock-stretching-related, it appears
idle owl
It's always clock stuff.
tulip sleet
hmm, you actually reproduced on S2 with bitbangio
idle owl
Yeah.
tulip sleet
so anyway I am going to concentrate on S3 with regular busio
tulip sleet
i am amazed this sensor works at all. The datasheet implies it's hard to wake up, and that it may prevent other I2C devices from working when it's asleep. But that does not happen
idle owl
Hah! No I haven't run into any of that nonsense.
tulip sleet
I may push this past 7.3.0, this may be another job for Espressif to fix, but I have to come up with a non-CIrcuitPython failure. And the fellow who fixed the previous I2C bug is stuck at home Shanghai so his supply of I2C devices is very limited
anyway, just FYI
idle owl
Oi. Ok thanks for letting me know.
stuck elbow
I may push this past 7.3.0, this may be another job for Espressif to fix, but I ...
spritetm?
tulip sleet
o-marshmallow: Omar
idle owl
Oh that's where the username came from. That's pretty funny.
warm flame
how do I add file to a pullrequest
tulip sleet
how do I add file to a pullrequest
git add or the equivalent in the GUI or the web interface.
idle owl
I'm not sure how to do that through GitHub. I have run into this previously and usually end up switching to working on it locally on my machine. I would google whether there's a way to do it through the GitHub UI though.
tulip sleet
@warm flame in your branch in GitHub, use this:
idle owl
Oh nice!
warm flame
did it.
I forgot a space between ":" and "BIN" failed pre-commit
lol
should be ok now
idle owl
This guide page is a massive wall of text. I have no idea what to use as images. Have to come up with something.
Probably screenshots of CIRCUITPY with various libraries or some such. Kind of boring, but better than nothing.
warm flame
build worked
idle owl
build worked
Great job sticking through this!
warm flame
what are the 2 yellow check ?
idle owl
Yellow dots? They're queued up, or haven't finished running yet.
Some CI things require other things to complete before they show up and run.
So sometimes, you'll think it finished because one of them completed successfully, but then another one shows up.
warm flame
so it takes a lot of time ?
idle owl
Oh yes. Much less than it used to, but still can be quite a bit.
jaunty juniper
it's gonna build every board because of the change to build_board_info.py, so it takes a long time
warm flame
ohhhhhhhhh
jaunty juniper
nothing we can do about it for now
warm flame
it should pass then
because I didn't touch to anything more than the ESP32-C3 board
and it already passed
wth
my build failed
again : Cannot find file ../ports/espressif/build-Lilygo-T-oi-Plus-ESP32C3/firmware.uf2
"lilygo_ttgo_t-oi_plus": BIN,
jaunty juniper
yeah you need to use the exact directory name "Lilygo-T-oi-Plus-ESP32C3" it's the board identifier
also I would prefer personally that it uses all lower case
warm flame
ok will change that then
jaunty juniper
that's just me I guess, but I think we should have a rule about that
I'm just bothered by Seeed_XIAO_nRF52840_Sense and the inconsistency with other xiao boards
tulip sleet
in general we use underscores in the directory names
and I agree, they should be all lowercase
manic glacierBOT
warm flame
anyways be back later
proven garnet
@idle owl am I still good to run the patch tomorrow for the pre-commit file?
idle owl
<@330227457296957440> am I still good to run the patch tomorrow for the pre-comm...
Should be! Do it slowly simply to make sure things are still working properly, obviously. But yep!
proven garnet
Sweeto! Will do!
proven garnet
Whoops, I meant Saturday, but same same!
idle owl
Whoops, I meant Saturday, but same same!
Right, that's what I thought, but thought maybe something had changed.
proven garnet
The core adafruit_bus_device module is only present when frozen in, right? I'm gonna mention that and link to the external one repo in the core docs but wanted to check
idle owl
The core `adafruit_bus_device` module is only present when frozen in, right? I'...
It's built into the core, it's not frozen in... I thought?
And if it doesn't fit in a given build, you have to load it into the lib folder manually.
tidal kiln
and to determine if its in any given boards firmware build:
https://docs.circuitpython.org/en/latest/shared-bindings/support_matrix.html
idle owl
Turns out NeoPixel is a great library to show the difference in memory usage between mpy, py, and frozen imports (on CPX). Tried CircuitPlayground library first, but I knew what the result would be when I got to .py file, failed. I got screenshots anyway, might have a place somewhere on this page.
marble hornet
Hi all!
Cookiecutter is still the preferred way to make new libraries?
(A quick glance at the learn guide says yes but thought it'd be worth a doublecheck)
idle owl
Hi all!
Cookiecutter is still the preferred way to make new libraries?
(A quick...
Yes! Unfortunately there are a few things still missing from cookiecutter, but we're doing our best to keep it updated. That is the preferred way to start a new lib for sure, though.
Also, good to see you! It's been ages 🙂
marble hornet
Yes! Unfortunately there are a few things still missing from cookiecutter, but w...
Awesome! I see it's changed
idle owl
Quite a bit!
marble hornet
And you too! It's good to be back 😊
idle owl
The README should be up to date with what prompts mean what, and I think the guide is also up to date, but I'm not certain.
marble hornet
So far it looks like it is both update date and mirrored
idle owl
Excellent
marble hornet
Could a run a specific to this question by please?
I've been working on a thing and am not sure which if any bundle it would be best for.
idle owl
Sure! I might have an answer. If I don't, someone will, I'm sure.
marble hornet
Thanks, I ask b/c I'm guessing you will 🤷♂️😄 (or pass it to the weekly meeting)
(typing...)
marble hornet
A) Would there be interest in a minimal typing library? (with ~one alteration to the Core)
B) Would it be something that should be on the adafruit, circuitpy org, or community track?
C) Has there been previous discussion of this? I did some searching around but didn't find PRs or other documents (I have not checked recording)
My bad, It's three. Maybe this should be a issue or PR
Optional explanation:
I've made a branch of CP that supports __class_getitem__ after making a typed codebase and struggling to hack it into working on stock CP. In short, i think the only things needed to make it "hack-less" is bing able to do SomeType[...] and a typing module with TYPE_CHECKING, Generic, Protocol, TypeVar, and
marble hornet
Sure! I might have an answer. If I don't, someone will, I'm sure.
Sorry for the delay 😅
idle owl
No worries, thanks for tagging me, I was off elsewhere.
Reading now.
We have a typing library?... hold on
@marble hornet How close is this to what you mean?
warm flame
didn't work
marble hornet
Oh i mean the opposite. Bringing some cpython typing to CP.
warm flame
didn't name the directory right
marble hornet
<@368190919083556865> How close is this to what you mean?
Only the parts that cannot go in type-hints
idle owl
Oh i mean the opposite. Bringing some cpython typing to CP.
Ah ok. I think this is best place for this question is In the Weeds on Monday, or in an issue on the core, or both, for that matter. I think we'll come to a more thorough solution with more input.
marble hornet
Ah ok. I think this is best place for this question is In the Weeds on Monday, o...
Thank you and will do!
Good to know it sounds like it's work bringing-up (?)
idle owl
I'd say so yeah, for sure. No idea what will come of it, but it's definitely worth bringing up.
blissful pollen
The core `adafruit_bus_device` module is only present when frozen in, right? I'...
Dunno if you had this answered fully but adafruit_bus_device is part of the core. The python module can be frozen in if for some reason you didn't build the core module
jaunty juniper
I would think the frozen python module would take more space than the native implementation ?
idle owl
If I'm comparing import memory free between frozen, mpy and py, what's the best comparison to use? Difference between frozen and py, and py and mpy? Or difference between frozen and both py and mpy, or difference between baseline and frozen, mpy, and py? Now that I'm thinking about it, probably baseline.
Otherwise it's not really a good comparison.
blissful pollen
Baseline may make it easy to see, rather then relative
idle owl
Yeah that's what I realised after typing it. Basically, once I had it in front of me typed out, it was obvious.
warm flame
38 boards to check left
warm flame
@jaunty juniper What's the next step ?
jaunty juniper
somebody reviews it, approves it and then somebody merges it, meanwhile you can add it to circuitpython.org
https://learn.adafruit.com/how-to-add-a-new-board-to-the-circuitpython-org-website/overview
proven garnet
Dunno if you had this answered fully but `adafruit_bus_device` is part of the co...
So is that a configuration setting for some builds? Like some just have it built in by default, or they all do?
jaunty juniper
So is that a configuration setting for some builds? Like some just have it buil...
like all (optional) builtin modules
it's CIRCUITPY_BUSDEVICE, which is on by default in full builds
full builds (CIRCUITPY_FULL_BUILD) is turned off on some small boards, typically M0
boards that require the external adafruit_bus_device if using a library that requires it:
https://docs.circuitpython.org/en/latest/shared-bindings/support_matrix.html?filter=-bus_device
that's more than I expected
but I guess it's all M0s
warm flame
somebody reviews it, approves it and then somebody merges it, meanwhile you can ...
This process is overwelming
and I never used ruby
too many new stuff
blissful pollen
So is that a configuration setting for some builds? Like some just have it buil...
What Neradoc said. Most have it built-in. I could search if you need a list but I think it is only turned off in the smaller SAMD21 builds.
proven garnet
like all (optional) builtin modules
Got it thanks!
tulip sleet
@onyx hinge what are you thinking about https://github.com/adafruit/circuitpython/pull/6338 ? I talked to Scott about this a couple of days ago, but he wasn't sure either. I think we don't know of an example where SH1107 is used for non-one-bit displays
onyx hinge
@tulip sleet Unless someone thinks there are >1 bit SH1107 displays then we either want to ignore the positional argument, or give an error. Ignoring it had the lower flash size cost.
giving an error would be better, but I didn't want to add messages to the core
tulip sleet
Scott was interested in an error. I will see if there's a message we can reuse, or maybe use the range checking arg checker
I will take it over
onyx hinge
thank you, feel free.
tulip sleet
are you quitting work early today? You will have a marvelous trip, bon voyage!
onyx hinge
if we were taking my 'solution' then at a minimum a note in the docs should be added that the flag is ignored unless it's a 1bpp display
thank you! I will be 'around' today but I am not necessarily 'working'.
jaunty juniper
say, I'm not sure what that implies by the way:
typing.TYPE_CHECKING
A special constant that is assumed to be True by 3rd party static type checkers. It is False at runtime.
onyx hinge
and then tomorrow is travel day, sunday is the first day in paris
tulip sleet
say, I'm not sure what that implies by the way:
> typing.TYPE_CHECKING
>
> ...
it is only True when processed by mypy, etc. Is that what you mean?
the actual module defines it as False
jaunty juniper
but what does it to test it then ? wouldn't it cause a runtime error when annotating with the types guarded by it?
onyx hinge
not when used in conjunction with from __future__ import annotations
or using string-annotations
from __future__ import annotations
from typing import TYPE_CHECKING
if TYPE_CHECKING:
from typing import Optional
def f() -> Optional[str]: pass
```so for instance this is acceptable to Python3 even though 'at runtime' `Optional` isn't actually importedas is this variant, which uses a string annotation: ```python
from typing import TYPE_CHECKING
if TYPE_CHECKING:
from typing import Optional
def f() -> "Optional[str]": pass
and in CircuitPython, python try: from typing import TYPE_CHECKING except ImportError: TYPE_CHECKING = const(0) # can't be const(False) means everything in the if TYPE_CHECKING: block is not even placed in the mpy file, which is pretty keen
tulip sleet
could do?
try:
from typing import TYPE_CHECKING
# put conditional stuff here
from typing import Optional
optional_str = str | None
#etc.
so it's telling the static analyzer where to look for types that are otherwise hidden from the runtime
onyx hinge
could do?
```py
try:
from typing import TYPE_CHECKING
# put conditional ...
There's a small chance it won't be doing what you expect (say you had from typing import Optionall, the wrong import error will be suppressed) and I'm not sure mpy-cross will successfully eliminate the dead code the way if TYPE_CHECKING + const will
tulip sleet
you're right about the dead-code, yes, I forgot that
onyx hinge
actually adding typing to jepler_udecimal enough for it to pass mypy --strict is feeling like a challenge!
onyx hinge
Found 436 errors in 3 files (checked 3 source files)
Yup, typing jepler_udecimal is a slog.
idle owl
Seriously arbitrary style question: If I'm referring to a ".mpy file", should I expect people to read the . as "dot", meaning I put "a" before it, or should I expect people to skip over the . and read only "mpy", in which case it should be "an" before it?
proven garnet
Seriously arbitrary style question: If I'm referring to a ".mpy file", should I ...
I skip the dot in my head, so to me "an" makes more sense
idle owl
I keep going back and forth, I've been leaning towards "an" but haven't been consistent, and I need to be.
proven garnet
Looking at it too long, I almost feel it works the other way - I'll read it differently depending on whether you use "a" or "an" haha
proven garnet
idle owl
@proven garnet I was thinking. Equally as important as the pre-commit-config patch, is getting cookiecutter in line with the current updates. Otherwise, new libraries are being created that aren't up to date and we end up missing things. Is that something you could also work on this weekend? I left one suggestion here: https://github.com/adafruit/Adafruit_CircuitPython_TSC2007/pull/2 but if that's everything, please add these changes to cookiecutter if you have the cycles.
proven garnet
Yeah, I can definitely do that. Is everything ready to go? If so I can 
idle owl
Yeah, I can definitely do that. Is everything ready to go? If so I can <a:ship...
If you caught everything that needed updating when you made that PR, then yes, it's ready to go. That's the only reference I was aware of as to what needed to be updated in cookiecutter.
proven garnet
Sweeto! Yeah, I can get it done over the weekend, possibly today
idle owl
Perfect, thank you. I can't help folks who started a library 2 months ago, but I can help folks creating libraries moving forward. 🙂
onyx hinge
Perfect, thank you. I can't help folks who started a library 2 months ago, but I...
you helped me 🙂
proven garnet
I'm playing around with automatically building a template library with cookiecutter and seeing if hooking it up as an upstream in git has any success
idle owl
you helped me 🙂
Hah! Valid. You're exactly who I was thinking of when I said that 😄
proven garnet
So every cookiecutter change would build a library, but I'm not sure how having a repository setup like that will behave since it isn't really an upstream (I think?)
idle owl
So every cookiecutter change would build a library, but I'm not sure how having ...
Hmm. No idea.
Not even sure I'm following 😄
proven garnet
It's a python script that could run via CI that would submit a pull request to a library (say, "CircuitPython_AutoBuild") with infrastructure changes, and anyone with that as an upstream could easily pull changes from it. Or that's the hope anyway.
idle owl
Oh oof. I'd be totally lost as to how to do that. But I now understand what you mean.
proven garnet
cookiecutter has a pretty sleek Python interface it seems, so I'm starting there
idle owl
Nice
proven garnet
But it can't catch everything, if for some reason we need to modify __version__ in library files or something then womp womp
but maybe at least the other stuff
idle owl
@proven garnet Bonus points if you do a pass on .pylintrc to up min-similarity-lines= to 12 while you're at it. 😆 This is not as high a priority though. So don't feel obligated. I think Adabot might be able to do that, but then again, I think that regularly and am regularly incorrect. So...
slender iron
is jerry out of town?
I'm revisiting https://github.com/adafruit/Adafruit_CircuitPython_ESP32SPI/issues/72 and wondering why get_time doesn't just block until the time is valid
deft jay
Is there any hope for support for XMP (not NEC) IR protocol within the adafruit_irremote library? Is it already there and I just don't know how to read a help manual?
guptaxpn — Today at 2:28 PM
https://github.com/crankyoldgit/IRremoteESP8266/issues/1414 << It would appear that it's been done elsewhere, but I just can't figure out how to port this.
Probably better suited for this channel.
slender iron
@deft jay I don't think it should go in irremote but could be in a new library
I don't know why we'd add support for it though
deft jay
Comcast remotes specifically utilize XMP
slender iron
is making a non-blocking function blocking an api break?
deft jay
which services a huge portion of the US
idle owl
Tannewt is asking whether you changing a non-blocking function into a blocking function is considered breaking the API. The wording was a bit odd.
deft jay
Would adafruit accept pull requests?
idle owl
Well, there's two options here. If it's a separate library, which it sounds like it sort of needs to be, you could submit it to the Community Bundle. If it were to be included in the existing library, then yes, you would make a PR to the lib. However, IRRemote is one of those libs we can't really extend too much because it's included in CircuitPython builds for memory-constrained microcontroller boards. So we have to be careful in considering what to add.
Basically, you would create a library under your own GitHub, submit it to the Community Bundle, and then it's available within the Community bundle for download on circuitpython.org/libraries.
deft jay
Aah
so that's why it's NEC only
an odd omission considering the arduino ones are so big
but I guess they also get cut back at compile time
idle owl
I think we implemented NEC, and that was enough for most things, so we went with it for CircuitPython, and then when we ended up needing it in memory-constrained situations, we decided to keep it as such.
proven garnet
<@135415127879974912> Bonus points if you do a pass on `.pylintrc` to up `min-si...
Is this for the .pre-commit-config.yaml patch to the libraries, or the cooiecutter patch?
deft jay
@idle owl was there inspiration from this project in the dev for the adafruit library? : https://github.com/cyborg5/IRLibC ?
It looks several years since last commit, I wonder if there's a way to pull some stuff around
idle owl
Is this for the `.pre-commit-config.yaml` patch to the libraries, or the cooiecu...
That would be to the libraries. Cookiecutter is still 12 I think.
proven garnet
Got it, I'll add that in for the infrastructure patch then
Is that also ready to roll out?
idle owl
<@330227457296957440> was there inspiration from this project in the dev for the...
You know, I'm not sure. There might have been. Chris is a close friend of Adafruit.
idle owl
Is that also ready to roll out?
The min-similarity-lines bit?
proven garnet
The min-similarity-lines bit?
I added that in for the cookie cutter patch
deft jay
He's the reason I felt like I could take this project on. Or at least his personal remote project.
I'm thinking his library might actually work, I think GICABLE might be comcast?
idle owl
I'm thinking his library might actually work, I think GICABLE might be comcast?
No idea, but worth trying for sure.
proven garnet
Wait I think I unconfused myself haha. I'm rolling out the .pre-commit-config.yaml patch this weekend but didn't have a general patch put together for the changes proposed for libraries I listed here: https://github.com/adafruit/Adafruit_CircuitPython_BME680/pull/48
idle owl
Wait I think I unconfused myself haha. I'm rolling out the `.pre-commit-config....
Ahhh yeah ok that's separate.
Probably better to do in pieces anyway.
idle owl
The .pylintrc change from 4 to 12 came out of the Sprints. The reason it's no rush, is it hasn't come up in other libs yet, but it almost certainly will.
So I'm getting ahead of things.
For once 😄
proven garnet
No worries! I'll add it to that "PR" I just mentioned and after this round of patches start working on something for those. Unless Adabot is able to do them, but otherwise I'm happy to
idle owl
Well, we can try Adabot in teaching you how to do it, to find out if you can.
proven garnet
Oh yeah, I 'member
idle owl
The more complicated the patch, the more likely it is to fail across more libraries.
So it might be worth doing multiple patches for that one.
I say might be, I mean is.
proven garnet
Sounds good! If next week is good, I can make myself available from 12-1 any day!
Or another week, no rush for me
Wednesdays are a little dicey since I have a 1pm, but doable!
idle owl
Mondays are always busy with meetings. Tuesday is typically free. We can start then and move it into further days if necessary.
proven garnet
Sounds good!
idle owl
I added it to my calendar. Hopefully you're better at remembering things than I am.
Feel free to ping me before that so it's on my radar, heh.
idle owl
Hah 😄 Well, maybe between the two of us, lol.
idle owl
Hmph. CircuitPlayground library was missing from /frozen for some reason. I was on main branch. Obviously pull and make fetch-submodules fixed it. But I'm still curious how it wasn't there to begin with. 🤷🏻♀️
devout jolt
Just curious, does anyone know of any reason why compiling with#define CIRCUITPY_DISPLAY_LIMIT 4 and creating four displayio.I2CDisplay objects would cause a HardFault?
I don't see anything obvious in "shared-module/displayio/init.c" (I'm trying to use four SSD1306 I2C OLEDs on a ESP32s3. Any three work)
tulip sleet
Just curious, does anyone know of any reason why compiling with`#define CIRCUITP...
I can only think of checking whether you're running out of some resource (RAM, etc.)
devout jolt
I can only think of checking whether you're running out of some resource (RAM, e...
100kB free (this is on ESP32S3)
warm flame
@tulip sleet yeah T-01 and T-oi are confusing!!!
tulip sleet
quote: _T-OI As the name suggests, T is still our T series. TOI is from the right to the left, meaning IOT products. _
devout jolt
lol
jaunty juniper
! tcefreP
tulip sleet
@jaunty juniper so RTD broke even though there were no errors in the CI build?
jaunty juniper
yup, it runs its own thing from .readthedocs.yml which does not use the same rules as the CI
but apart from this file, I don't know what makes it do what where why
I wonder if RTD offers a way to test build the docs with their system in the CI so it's run with the actual configuration file it uses (and fails if the configuration causes an error)
tulip sleet
I wonder if RTD offers a way to test build the docs with their system in the CI ...
that would be nice! not sure either
jaunty juniper
I kind of want to add a comment in ci_fetch_deps.py:
elif target == "docs":
# NOTE: must match .readthedocs.yml as this script is not run by readthedocs
submodules = ["extmod/ulab/", "frozen/"]
feel free, push a new commit in the fixup PR
it will restart (it hasn't started yet anyway)
onyx hinge
https://docs.readthedocs.io/en/stable/pull-requests.html yes, readthedocs has something for this -- we have to enable it on rtd, I think
I've experimentally enabled this setting on circuitpython
(just now)
tulip sleet
You could also set up your own readthedocs site just for testing (maybe you can hide it from search engines), and do it on your own pushes, or trigger the webhook manually only
onyx hinge
why yes, it's my travel day and here I am making CP patches
tulip sleet
well, if it's a distraction from impatience or worry, sounds fine
onyx hinge
impatience, mostly
onyx hinge
you couldn't make all of these in CP, but you could make some (and think about what core changes would enable some of the other ones at a minimum of fuss) .. https://blinry.org/50-tic80-carts/
like maybe the heart particle system https://blinry.org/50-tic80-carts/24/
tulip sleet
@onyx hinge willing to approve https://github.com/adafruit/circuitpython/pull/6398 ? (frozen updates)