#circuitpython-dev
thorny jay
Won't make it for the meeting.
I have nothing to report except that we are having fun letting an LLM writing the code for us.
And also group hug.
tulip sleet
lone axle
Thanks for hosting Dan, have a great week everyone 👋
tulip sleet
@lone axle The echo-cancelling did not work well this time in my recording. Did you make a backup recording? If not, no big deal
lone axle
<@382939733107408897> The echo-cancelling did not work well this time in my reco...
Yes, I did. Will upload and share a link a moment.
tulip sleet
thank you!
lone axle
thank you!
tulip sleet
https://www.dropbox.com/scl/fi/y75mc5gteudgr8qwku5ui/CircuitPython_Weekly_meetin...
Thanks! downloaded -- you can delete
tulip sleet
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1AxFL2PBO0_5sNtGk03McAd9EYIKtvTdZTly28n_SL4U/edit?usp=sharing
orchid basinBOT
tulip sleet
@slender iron have you got some time for a video checkin about short-term PR discussions, etc.?
slender iron
yup!
surreal perch
I hope it was ok for me to close the issue. It is fixed in 10.1.1 and I opened it but I'm kind of new to the filing of bugs and closing them so let me know if I should have left it open for others to review.
tulip sleet
I hope it was ok for me to close the issue. It is fixed in 10.1.1 and I opened ...
It's fine to closed it. You retested and it was no longer a problem -- that's good to hear !
manic glacierBOT
manic glacierBOT
orchid basinBOT
tulip sleet
https://blog.adafruit.com/2026/02/18/circuitpython-10-2-0-alpha-1-released/
Highlights of this release
- Include 10.1.1 fix for performance regression.
- Better entropy for RP2350 os.urandom().
- Add
supervisor.get_setting(), which retrieves a string, boolean, or integer typed value fromsettings.toml.
orchid basinBOT
short tendon
Random question:
I've been doing a lot of work with dof sensors and have noticed that the Adafruit libraries are designed to be human readable vs performant (lots of lookups, multiple function calls, etc).
Trying to figure out if optimizations would be accepted, or if I should make my own forks
tulip sleet
Random question:
I've been doing a lot of work with dof sensors and have notice...
Do you mean with the Register library? It would be nice to figure out how to do that more compactly. If Python had macros...
short tendon
Do you mean with the Register library? It would be nice to figure out how to do ...
So both that and the sensor libraries themselves.
For example in the Adafruit_CircuitPython_LSM6DS library for the gyro it does:
@property
def gyro(self) -> Tuple[float, float, float]:
"""The x, y, z angular velocity values returned in a 3-tuple and are in radians / second"""
raw_gyro_data = self._raw_gyro_data
x, y, z = (radians(self._scale_gyro_data(i)) for i in raw_gyro_data)
return (x, y, z)
def _scale_gyro_data(self, raw_measurement: int) -> float:
return raw_measurement * GyroRange.lsb[self._cached_gyro_range] / 1000
We could, first store this on range change:
self._gyro_multiplier = radians(GyroRange.lsb[self._cached_gyro_range] / 1000.0)
And then do this in the property:
rx, ry, rz = self._raw_gyro_data
scale = self._gyro_multiplier
return (rx * scale, ry * scale, rz * scale)
And remove a ton of overhead (lookups, math, generators, etc)
manic glacierBOT
manic glacierBOT
faint wadi
Can Please Someone Build The Bootloader for the Sensorwatch Fork For Me?
https://github.com/joeycastillo/uf2-samdx1
I just want the build from make sensorwatch_pro, which #define LED_PIN PIN_PA12 changed to #define LED_PIN PIN_PB22 on this line: https://github.com/joeycastillo/uf2-samdx1/blob/master/boards/sensorwatch_pro/board_config.h#L13
I tried building uf2-samdx1 on both my laptop with Ubuntu 24.04.3 and via WSL on my Windows PC and cannot get it built.
I also tried running the adafruit/uf2-samdx1 repo, and it fails as well on both machines.
I get a overlaps section .relocate LMA error each time.
In-depth attempt steps can be found here: https://github.com/adafruit/uf2-samdx1/issues/229
**Edit: **Fixed by following this tutorial and using Ubuntu 18 in a virtualbox. https://forum.arduino.cc/t/creating-a-custom-samd-bootloader-and-arduino-board/1094588
Unless corrected, I think the root cause is development mistakes made in newer versions of Ubuntu.
https://bugs.launchpad.net/ubuntu/+source/newlib/+bug/2107222
tulip sleet
So both that and the sensor libraries themselves.
For example in the Adafruit_C...
That is a great example of what could be done. We would welcome such PR's. I don't think the way it was done was necessarily deliberate; it was just done quickly.
The same might be true of the corresponding Arduino library. In the past the Arduino library was written first and then served as the template for the CPy library
tulip sleet
Can Please Someone Build The Bootloader for the Sensorwatch Fork For Me?
<https...
Re your comments in the repos: nothing to apologize for! We have cooperated with Microsoft on this in the past. Their initial work was instrumental in getting MSC-based bootloading to be a common thing. The bootloaders work, but don't get a lot of maintenance attention, partly because we haven't shipped SAMD-based boards in a while. There's a recent problem with SAMD bootloaders on ChromeOS, which comes and goes -- it may be some kind of race condition.
faint wadi
Re your comments in the repos: nothing to apologize for! We have cooperated with...
That's interesting! I figured that SAMD boards were the go-to board for simple applications due to its versatility, price, and amount of support.
What has Microsoft moved to?
tulip sleet
not microsoft, us. The RP2040/RP2350 has been a good substitute. SAMD21 is really too small for CircuitPython except for small applications. SAMx5x is better but the cheaper versions are somewhat flash-flimited.
stuck elbow
the uf2 bootloader was someone's side project, they laid off all people who were working on it
tulip sleet
Microsoft Research moved to working on JacDac, but I haven't followed what they've been doing recently. The UF2 bootloader was just an enabling technology for MakeCode and the like
slender iron
Microsoft Research moved to working on JacDac, but I haven't followed what they'...
Peli is working on the github actions AI stuff. Michal is at OpenAI according to LinkedIn
short tendon
That is a great example of what could be done. We would welcome such PR's. I don...
Is there a list somewhere of all the DOF sensors Adafruit has (other them me combing through all the products)? I'd love to see how most have been written and see what standards I can come up with
also adafruit_circuitpython_register is pretty small. has there ever been thought of putting it into CP itself so it can be faster?
short tendon
ahhh, it has a motion sensor section. That will be a good starting point. and about half I've visited 😉
I'm building out profiling now. Most of what I build on that will be using a random generator for getting the raw values, so there isn't any i2c issues that skew results
I have hours or recorded readings I'm using to optimize some code I'm writing for am agnostic fusion library
slender iron
@tulip sleet the zephyr bleio PR is ready for review. Will have smaller follow ups after it
I was dreading fixing the last four build failures... and pi did it all by itself with just a short prompt.
short tendon
Here's some interesting stuff I've started with
100000 iterations
a:
import math
def test(value):
return math.radians(value)
b:
from math import radians
def test(value):
return radians(value)
c:
from math import radians
local_radians = radians
def test(value):
return local_radians(value)
a) 341.86 ms
b) 257.54 ms
c) 238.25 ms
so things that happen in tight loops like computing sensor data can benefit a lot from some optimizations...
orchid basinBOT
orchid basinBOT
slender iron
Get dan's comments from the PR and address them
and it did most of the things
I did have to stop it before it pushed
tulip sleet
> Get dan's comments from the PR and address them
as long as it did what you wanted to do 🙂
tulip sleet
I would have been more explicit. ... yeah, I expected you could easily ask it to do that.
slender iron
next time I'll tell it to not push the changes
faint wadi
not microsoft, us. The RP2040/RP2350 has been a good substitute. SAMD21 is reall...
Is there a suggested way to install arm-none-eabi on Ubuntu 24.04?
I wasn't able to find any official documentation and forums I found warn of many pitfalls in symlinking.
I found this bash script, which uses a version from Summer of 2025, and used it (after reviewing it for any neferious commands, of course).
Is that the method you'd use?
https://github.com/kmhallen/gcc-arm-none-eabi/blob/main/gcc-arm-none-eabi-install.bash
stuck elbow
Is there a suggested way to install arm-none-eabi on Ubuntu 24.04?
I wasn't able...
just unpack the tarball and point the PATH to it?
tulip sleet
Here is where to download it: https://learn.adafruit.com/building-circuitpython/linux#cortex-m-builds-3049646. Then you can just do as deshipu says: unpack it and add its bin directory to the front of your PATH.
faint wadi
just unpack the tarball and point the PATH to it?
Thanks! Is there a suggested folder it should be in? Does usr/bin make sense?
If usr/bin is in $PATH, does that mean that I just need to unpack it there and run apt update?
Thanks for helping me with my novice questions btw
serene token
Is there a suggested way to install arm-none-eabi on Ubuntu 24.04?
I wasn't able...
If I'm not mistaken, it's in the Ubuntu apt-get repos. That said, if you want the newest version, that probably isn't it.
(If you install it by download and unpacking a zip or tar file, apt will not keep track of it or even be aware of it. So apt update won't do anything for it. If you install it using apt or synaptic, then apt will know it is there and update it when it is updated in the repo by running apt update.)
I might be wrong. I'm finding stuff about ARM not hosting a PPA for it anymore, but that doesn't necessarily Ubuntu isn't hosting in their own repos.
Try this to check: apt-cache search arm-none-eabi If there's a package in the repos for your Ubuntu version, it should come up. If it doesn't, then I'm wrong and it isn't in the Ubuntu repos afterall.
serene token
Thanks! Is there a suggested folder it should be in? Does `usr/bin` make sense?
...
Please don't put it in /usr/bin. dpkg assumes that it has control of everything in there, and if you put something in there that overwrites something installed with apt, it can break other applications. It is generally recommended, if you are going to install something outside of apt, to install it in ~/.local for your normal user. That way you don't risk damaging something installed by apt. From there, find the bin directory in the ARM toolchain directory, and add that bin directory to your PATH variable.
faint wadi
Please don't put it in `/usr/bin`. `dpkg` assumes that it has control of everyt...
Thanks. I've given up on anything related to PPA. All sources show that arm-none-eabi isn't maintained there anymore, and I'd rather learn how to install the tar myself than try to find another host.
serene token
Not a bad plan, but yeah, apt update won't do anything for it if you install it manually.
What I would do is upzip the tarball, either in my home directory or in .local under my home directory (if I don't want to clutter my home). Then if you go down into the directory created when unzipping, you'll find a bin directory in there. Get the full path of that directory (you can cd into it and then do pwd to print the full path), and add that to the PATH environment variable. That should do it.
If you need to upgrade to a newer version, you can unzip it in your user .local (or home, if you put it there), and change the PATH entry to use the directory for the new version (usually just changing the version number). Once that is done, you can delete the whole directory tree for the old version, unless you want to keep it. Note that if you are setting and changing PATH in .bashrc or .profile, you'll need to restart your terminal to use the new one. (Or you can run source ~/.bashrc or source ~/.profile, depending in which you are using.)
Personally, I usually drop that one in my home directory, so that if I have a reason to start digging into it, I can find it easily. That said, pick whichever you prefer and then be consistent about it, and you'll be fine!
tulip sleet
Thanks. I've given up on anything related to PPA. All sources show that arm-none...
I have a ~/bin directory which has some executables in it directly, and I also put other local software there and add to PATH as appropriate. It's something I've done for decades, before .local was a convention.
serene token
I have a `~/bin` directory which has some executables in it directly, and I also...
I've also seen people do a ~/.bin as well. As long as you don't step on the toes of the package manager, any of that works.
tulip sleet
@slender iron I'm not sure why we are suddenly having this build problem with click. I guess some other pypi module is now requiring an older click. Click 8.3.1 was released in November, so this is not because of some new release. Maybe some cache expired?
the logs don't say who is requiring click<8.2;>=7.0
tulip sleet
Found it, it's in ~/.espressif/espidf.constraints.v5.5.txt. That seems to be generated during install.sh.
I looked at an older working build and it was using some CI caches. The failing new ones are not.
I think maybe this is because requirements-dev.txt reqs are installed after ESP-IDF reqs
maybe we talked about fixing this before?
tulip sleet
... Trying a change to actions.yml to do that
tulip sleet
now missing dependencies for espressif boards
Run python3 -u build_release_files.py
Traceback (most recent call last):
File "/home/runner/work/circuitpython/circuitpython/tools/build_release_files.py", line 12, in <module>
import build_board_info as build_info
File "/home/runner/work/circuitpython/circuitpython/tools/build_board_info.py", line 13, in <module>
import sh
ModuleNotFoundError: No module named 'sh'
Error: Process completed with exit code 1.
sh is in requirements-dev.txt, so it's not just that it was missing
slender iron
maybe we talked about fixing this before?
We definitely have had this issue before where we feel the impact of a new release only after our cache key changes. Could have been a python version bump
tulip sleet
We definitely have had this issue before where we feel the impact of a new relea...
there was 3.14.2 to 3.14.3. I'm looking at .github/actions/deps/external/action.yml and trying to see what to do. I think always do pip install -r requirements-dev.txt, even for espressif. Why is it turned off for zephyr-cp? Does that have a completely separate list of dependencies?
slender iron
I'm not sure why it is off. Zephyr does have its own list of python deps
tulip sleet
trying again...
trying to have the port-specific install re-do any installations that the general requirements-dev chose
maybe I will take away that check for zephyr too, if the zephyr builds fail too
slender iron
I compared the failed run 22229143457 with the last successful Build CI that included espressif (22190907999).
What changed
- Successful run (22190907999) restored IDF python cache with key:
- Linux-/opt/hostedtoolcache/Python/3.14.3/x64-tools-idf-...
- Log shows: “Cache hit for …3.14.3…tools-idf…”
- Failed run (22229143457) looked for:
- Linux-/opt/hostedtoolcache/Python/3.14.2/x64-tools-idf-...
- Log shows: “Cache not found for input keys …3.14.2…tools-idf…”
So it did not reuse cache because the cache key includes env.pythonLocation, and that changed from 3.14.3 to 3.14.2.
Why that caused the build break
After cache miss, it rebuilt/install python deps in the IDF env, which pulled newer click:
- Failed run error:
- Requirement 'click<8.2,>=7.0' was not met. Installed version: 8.3.1
In the successful cached run, the env had click 8.1.8, so dependency check passed.
────────────────────────────────────────────────────────────────────────────────
If you want, I can propose a workflow fix (e.g. pin Python minor/patch for stable cache keys, and/or prevent requirements-dev.txt from
upgrading IDF-constrained packages like click).
looks like the python update caused the cache miss
the simplest fix is to change our -dev.txt to limit click
though there is a minor idf update we can do too
gonna see if codex can do that
tulip sleet
> I compared the failed run 22229143457 with the last successful Build CI that i...
I was thinking of limiting click in that way, but hold on for the build I am doing as part of a PR to fix this
I would like to do a 10.1.2 with the MagTag fix and this fix.
slender iron
that is a better fix
I wonder what else we can unpin then
this is definitely not the first time its happened
tulip sleet
I wonder what else we can unpin then
no, it didn't work, problem with nested virtual environments. I will pin click for now, and we can work on a better solution in main.
slender iron
lemme check what version of click is used my idf 5.5.3
tulip sleet
this PR is for 10.1.x
slender iron
I can't tell where the <8.2 requirement comes from
it is in a constraints file in the idf tools
tulip sleet
it is in a constraints file in the idf tools
it is DOWNLOADED from somewhere, it's not in the repo
orchid basinBOT
orchid basinBOT
tulip sleet
https://blog.adafruit.com/2026/02/20/circuitpython-10-1-3-released/
Highlights of this release
- Adafruit MagTag: auto-detect different versions of e-ink display on 2025 MagTag.
- Espressif: fix board I2C object deinit(). Fixes crash with Adafruit Qualia library.
manic glacierBOT
empty salmon
I just read the code change for the MagTag.
Bitbang 4-wire SPI with a bidirectional data line to read the first word of register 0x2e
All I can say is that is some serious code-fu! Congratulations 🫡
tulip sleet
I just read the code change for the MagTag.
```Bitbang 4-wire SPI with a bidirec...
That code was already there, written by Scott, to try to read a different register to distinguish first-generation MagTags from 2025 MagTags
empty salmon
That code was already there, written by Scott, to try to read a different regist...
It’s impressive sleuthing! It’s a trick I will note “for future use” 👍🏻
Our Explorer Badge needed to differentiate various ePaper displays. We hand no idea about reading registers. Fortunately, we caught the scenario before final hardware and added a voltage divider to a pin. (Changing the resistors to yield different voltages based on which display was used in manufacturing.)
tulip sleet
It’s impressive sleuthing! It’s a trick I will note “for future use” 👍🏻
Our Ex...
Well,apparently that our initial test is not fine-grained enough. Someone has come up with a "new" display whose markings suggest one setting, but it seems to need another setting: https://forums.adafruit.com/viewtopic.php?p=1082017#p1082017 and following. I await the results. It appears the supplier is providing work-different variants for the same part number.
empty salmon
Well,apparently that our initial test is not fine-grained enough. Someone has co...
We are encountering similar “variants”. It has made supply chain even more challenging than it was (which for a tiny shop like ours, is saying something).
Anyway, cool trick and if I ever get into this kind of bind, I will remember to do this level of digging.
tulip sleet
We are encountering similar “variants”. It has made supply chain even more chall...
Many displays have no way of reading anything back: they are write-only. We were just lucky here.
I originally did not try to read a register, and just guessed on a value, but gave the user the ability to override in settings.toml: https://github.com/adafruit/circuitpython/pull/10831
empty salmon
I originally did not try to read a register, and just guessed on a value, but ga...
Good to know: I will revisit the use and timing of settings in toml file. (another good detail to remember for future use)
tulip sleet
Good to know: I will revisit the use and timing of settings in `toml` file. _(an...
Turns out the apparent fourth display version was not true; I am relieved 🙂
empty salmon
Turns out the apparent fourth display version was not true; I am relieved 🙂
We had to switch vendors so this topic may become relevant for us sooner than anticipated.
manic glacierBOT
To add further information, the colors are also wrong when this happens. This is running 10.1.3 and the only difference is the correct display I ran off the battery because it doesn't happen on battery and then the abnormal display I plugged it into USB where it does show the problem.
surreal perch
Let me know if you want me to test but I guess you found you have a board that will exhbit the error. Have a good evening.
manic glacierBOT
lone sandalBOT
manic glacierBOT
tulip sleet
@slender iron do you remember how to get ESP_LOGx() output on a different UART TX line on ESP32-S3, instead of TXD0? I tried:
CONFIG_ESP_CONSOLE_UART_CUSTOM=y
CONFIG_ESP_CONSOLE_UART_TX_GPIO=9
to get it to go to GPIO9. The sdkconfig has those settings in it, but I am still seeing the output on TXD0. This is so I can see logging output on a QT Py, and I have very limited pins and no TXD0 without a tiny bodge wire. (Can't reproduce the problem on a more convenient board.)
slender iron
that seems right to me. did you clean and rebuild?
tulip sleet
oh, yeah, multiple times, trying to get the config right (the CONFIG_ESP_CONSOLE_UART_CUSTOM was not obvious) I was testing on a Feather where I can see output on both TXD0 and 9
and the pin is marked as in use by CPy.
tulip sleet
I'll keep trying. the setup is fine. interestingly Espressif's IDF LLM help gave me the CONFIG_ESP_CONSOLE_UART_TX_GPIO, but didn't mention the _CUSTOM, and it was just pointing to a human-written thing that was also inadequate. I had hoped there was some commented-out something in our board files, but nothing there
slender iron
I'd probably look in menuconfig
and also verify the sdkconfig in the build dir has the right setting
tulip sleet
i've looked through the Kconfig's (used by menuconfig)... yup verified in sdkconfig as well. There may be another setting I am missing, but the settings in the generated sdkconfig look fine.
slender iron
did you try claude code or something in your working directory?
tulip sleet
just tried that now, it suggests using a different UART, instead of UART 0 which I was also thinking of. I was just hoping you remembered off the bat, since we've had to do this before.
i trust you more than an llm
slender iron
it's been a while. I normally use uart0
tulip sleet
i think the last time I did this was for a C3 or C6, which has different considerations (JTAG SERIAL)
anyway, I will try UART1, thanks
tulip sleet
yes, might be logging directly via the ROM routines or something
candid sun
<@&356864093652516868> We'll have our weekly meeting in about 60 minutes from now in this text channel and in the circuitpython voice channel. Please take the time to add your notes in advance to the document: https://docs.google.com/document/d/1AxFL2PBO0_5sNtGk03McAd9EYIKtvTdZTly28n_SL4U/edit?tab=t.0 -- I look forward to everyone's updates!
wraith crow
👋
tidal kiln
👋
slender iron
thanks for hosting!
lone axle
Thanks for hosting Liz, have a great week everyone.
candid sun
thanks folks!
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1O7WLFJtB-30PoMKpC718f7xujL1zuBMFYD059GWIdZk/edit?usp=sharing
manic glacierBOT
slender iron
zephyr bleio PR is finally ready
lone axle
@slender iron core lvfontio is already mostly supporting higher bpp fonts I think. However the palette gets defined at compile time inside gen_display_resources.py ultimately. I made a test build with 4bpp and it mostly works.
Do you have thoughts on how the palette could be made more dynamic or influenced by something end user accessible?
slender iron
looks at gen_resources.py
@lone axle supervisor/shared/display.c needs to be updated to override the palette when setting an lvfont. Is the palette always N bits in the normal pattern or does each index have a custom color?
lone axle
<@382939733107408897> supervisor/shared/display.c needs to be updated to overrid...
It looks to me like the palette is always N bits with background at index 0, and foreground at the last index. I've been putting a gradient evently stepped from one to the other in between and it seems to look correct.
but I have not tried to find the converter source/spec to verify
lone axle
Ahh, the pixel_shader on the terminals TileGrid is writable at runtime, so this is possible in REPL/code.
CIRCUITPYTHON_TERMINAL[0].pixel_shader = gradient.make_gradient_palette(0x0, 0x00ff00, 16)
yup, it is
lone axle
Do you know what the original font you converted for FJOS was? https://github.com/adafruit/Fruit-Jam-OS/tree/main/fonts I tried a unifoundry/unifont OTF file but I am coming up with considerably different filesizes. I also seem to have some kind of clipping going on for decenders, I'm not sure if that is a result of the font I'm using or something else in the code.
slender iron
I'm so good at git comments.....
slender iron
in my command history I see python generate_font.py --output fonts/unifont-16.0.02-ja.lvfontbin ja
lone axle
I got a different version. Will try that one.
slender iron
looks like the script is in circuitpython-font-generator
lone axle
Ahhhh, nice that has the ranges also. Thank you!
slender iron
python generate_font.py --output fonts/unifont-16.0.02-en_US.lvfontbin en_US too
lone axle
okay I will try some higher bpp of that specific font tomorrow and compare to current default.
slender iron
👍
lone sandalBOT
manic glacierBOT
slender iron
I'm fixing the tinycircuits board defs
tulip sleet
@slender iron I've got some questions about displayio display refreshing and bus competition, if you have time. This is to fix crashes when SD card and displayio are competing for the SPI bus, but could come up in other ways as well.
slender iron
<@252717193496756235> I've got some questions about displayio display refreshing...
I'm around
modern sundial
Update on the async busio: I have the DMA working for rp2040 I2C. Strangely it can read upwards of 2400 bytes, but can only write 54. I'm going to verify SPI and UART behavior and then submit a pull request, at which point we can talk about structuring and perhaps do some refactoring.
lone sandalBOT
candid sun
is there anyway to turn off ble and ble workflow when building for esp32-c3? when i try CIRCUITPY_BLEIO = 0 i get errors
tulip sleet
is there anyway to turn off ble and ble workflow when building for esp32-c3? whe...
do you want to turn it off to save time, or because it's broken in some way?
candid sun
do you want to turn it off to save time, or because it's broken in some way?
it's with the xteink- i keep intermittently hard faulting after setting up web workflow so i was trying to think of ways to make space to see if that was why
tulip sleet
by coincidence, do you see the bug just filed above?
I don't think it is necessarily space.
is there a 10.0.x build available?
candid sun
yeah i was just reading that.. i'll try to rebuild on 10.0.x
also can test with a c3 qt py since that isn't a hack
slender iron
@tulip sleet should we always rebase our changes onto zephy main or do merges from zephyr/main?
tulip sleet
<@329766224093249548> should we always rebase our changes onto zephy main or do ...
you mean in our fork of zephyr?
slender iron
yup, that we reference from cp
tulip sleet
As long as we can find our changes, rebase is fine and makes the commits list a lot shorter
slender iron
I was thinking we'd push it as our main branch and force pushing to it seems a bit weird
it does make it clear what is different though
tulip sleet
I was thinking we'd push it as our main branch and force pushing to it seems a b...
you could make a circuitpython branch that we merge to and use as the submodule, and keep rebasing that from their upstream main.
unlike esp-idf, we wouldn't need to do circuitpython-v1.2.3, etc.
slender iron
right, ok
modern sundial
I suppose six years was too long to wait
orchid basinBOT
slender iron
<@&356864093652516868> We'll have our weekly meeting in about 50 minutes from now in this text channel and in the circuitpython voice channel. Please take the time to add your notes in advance to the document: https://docs.google.com/document/d/1O7WLFJtB-30PoMKpC718f7xujL1zuBMFYD059GWIdZk/edit?tab=t.0 Talk to you soon!
turbid radish
Good
lone axle
turbid radish
I cannot hear anyone
lone axle
I hear Scott.
tidal kiln
i can hear
turbid radish
That worked, thanks
modern sundial
I can't hear either, but I suspect my workplace is blocking it.
turbid radish
You can try the web vs. the app. I had to refresh the window to hear
tidal kiln
spring forward!
crimson ferry
Spring ahead, Fall back.
lone axle
Thanks for hosting Scott, have a great week everyone.
slender iron
@lone axle I wish I had the zephyr display stuff checked in for your font testing
lone axle
Basic test can be done without any special code for the font. Just need to set a var in settings.toml to point to the custom font file, and then set the palette on the terminal tilegrid to use the right number and value for the colors. Bottom pixels get clipped without more changes but it's enough to check if it works generally.
slender iron
Basic test can be done without any special code for the font. Just need to set a...
Are you going to work on the changes needed to fix the clipping? What do you need my help with?
lone axle
Are you going to work on the changes needed to fix the clipping? What do you nee...
Yeah, I have local changes that resolve it. The main issue IMO is that these changes while fixing the clipping of custom fonts would also change the size that is calculated for the current default font which means it would no longer divide evenly into the Fruit Jam's display sizes.
Ideally I'm hoping to come up with a way that can fix it for custom fonts, but keep the default font at the current 16px size. That way everything default on Fruit Jam would look the same as it does today but other custom fonts would use the new measuring logic to fix the clipping and as a consequence also end up with larger line spacing (bigger vertical gap between rows).
slender iron
hrm, I wouldn't special case the current font
lone axle
Okay, I can have the current font use the new logic too but it will result in a bit less room for the serial terminal with the default configuration. Mostly bigger gaps between rows since the majority of glyphs don't actually use the fully allotted space.
slender iron
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Check your local time if you are outside the US. The US starts daylight savings time before the meeting. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! https://docs.google.com/document/d/1fbZVEge03Sl0UyAzeAO9SpFfF3urYhs_GNY5pC2Wbec/edit?usp=sharing<@&356864093652516868>
slender iron
Okay, I can have the current font use the new logic too but it will result in a ...
Are you taking the bounds of all glyphs into account or some global size instead?
lone axle
Are you taking the bounds of all glyphs into account or some global size instead...
I have local code with both ways actually. One of them just reads ascent and descent from the header and adds them together (adjusted for negative) to get the full size.
And another approach checks the size of each glyph as it loops over them during init to track the maximum space used by any of the glyphs that get iterated.
I haven't checked yet, but if they arrive at the same answer I'm inclined to go with reading data from the header because I think it's more similar to how BitmapFont does it.
slender iron
I'm not sure how the converter computes the two
slender iron
zephyr main PR is green
slender iron
thank you!
proven garnet
@lone axle catching up on the weekly today - I can expand the patch to include the changes to .pre-commit-config.yaml - good idea!
tulip sleet
@slender iron in https://github.com/adafruit/circuitpython/pull/10607#discussion_r2320342842 you mentioned avoiding MP_WEAK by using a #define instead. I was looking for an existing example of that. I am thinking of adding a new common_hal_busio_spi_wait_for_lock() function that would be the same on all ports except espressif. Was thinking of a flag like CIRCUITPY_BUSIO_SPI_WAIT_FOR_LOCK_PORT_SPECIFIC but I don't know if that's the kind of thing or name you were thinking of.
slender iron
<@252717193496756235> in <https://github.com/adafruit/circuitpython/pull/10607#d...
I think that'd be fine. The goal is to only have one to link against. Otherwise you risk accidentally linking against the weak version
tulip sleet
@slender iron if you have time, I have a fairly brief thing to discuss with you re user-created SPI buses used for displays.
modern sundial
Has anyone tried pinning tasks on esp32 or rp2x in a circuitpython build?
It might be useful for things like jpeg encoding, ulab, bitmapfilter, or audiofilters
I guess Zephyr will solve this for ESP32 but not rp2x
Do we have a Zephyr ESP32 build??
serene token
If I understand correctly, CircuitPython uses the second core for USB. USB is pretty timing sensitive, so while it might be possible to throw something else on the second core with USB, doing in a way that keeps the USB stable is not a trivial task.
If you want to use the other core for your own purposes, I've heard that MicroPython has support for that. That might be better suited to your needs, if you want to toss any of the tasks in your list over the second core. (I'd love to be able to use the second core for rendering, but all of my use cases also require USB, so that's not really an option. Maybe the next one in the RPx line will have four cores...)
tulip sleet
If I understand correctly, CircuitPython uses the second core for USB. USB is p...
Do you mean the second core on Espressif chips? It's used for WiFi. The USB task runs on the same core as CircuitPython.
manic glacierBOT
modern sundial
WiFi is already pinned to the second core? That’s fine then
manic glacierBOT
tulip sleet
<@&356864093652516868> weekly public meeting is in about 1 hr 45 min, at 2 pm US ET / 11 am US ET. Note that US went to Daylight Saving Time early yesterday, so those outside the US, recheck the time.
slender iron
Do we have a Zephyr ESP32 build??
No, but you should be able to do west -b in the zephyr port with any zephyr supported board and it'll do its best.
slender iron
@tulip sleet we can hear you
lone axle
@tulip sleet a few of us spoke but seems like you do not hear.
tulip sleet
i am going to reboot
lone axle
thanks for hosing Dan, have a great week everyone 👋
slender iron
Thanks all!
tulip sleet
@lone axle I could use your backup recording; there was an input problem until I changed some settings partway through. Thanks - I'm very grateful.
tulip sleet
Uploading now. will share a link a moment
all set - you can delete now - thanks!
manic glacierBOT
orchid basinBOT
tulip sleet
CircuitPython 10.1.4 Released!
https://blog.adafruit.com/2026/03/09/circuitpython-10-1-4-released/
Highlights of this release:
- Fix pin-reset regression which caused display problems on some boards.
- Fix object creation and deinitialization problems, most notably when SPI was used for displays on SAMD.
short tendon
Are there any optimizations that circuitpython makes when accessing variables? For example if self.foo is accessed multiple times in a method or loop, will it create a temporary record on the stack, or will it look it up each time?
stuck elbow
Are there any optimizations that circuitpython makes when accessing variables? F...
circuitpython code is not executed line-by-line, it's first compiled to bytecode, which looks a bit like assembly language, and that is executed in a vm
the compiler might do all kinds of optimizations, but I'm not aware of it doing the particular optimization you mention
you can read about the passes the compiler does here https://docs.micropython.org/en/latest/develop/compiler.html#compiler-passes
modern sundial
Are there any optimizations that circuitpython makes when accessing variables? F...
That optimization is normally performed by the CPU cache, not the compiler
tulip sleet
Are there any optimizations that circuitpython makes when accessing variables? F...
PYthon allows you to change the object dict at runtime, so theoretically you have to do a lookup each time. __slots__ avoids this in CPython, but is not implemented in CircuitPython/MicroPython: https://github.com/adafruit/circuitpython/issues/10517
short tendon
So, it seams that no, this isn't. And yes that's what I'm doing now, but wanted to make sure it was needed.
Thanks for all those links. Super helpful
tulip sleet
it does help, but it's worth doing a little benchmarking to see how much it helps. In a loop of many iterations, it could make a significant difference
short tendon
yeah, I'm specifically looking at loops (either in the class, or that are called in code)
manic glacierBOT
tulip sleet
I think something upstream there is broken temporarily. Also githubstatus.com is showing degradation for CI due to download failures
devout jolt
Is it expected behavior that the following actions will result in "locale/circuitpython.pot" having changes?
git checkout main
git pull
make translate
...
git diff
diff --git a/locale/circuitpython.pot b/locale/circuitpython.pot
index bf4c5d110f..09371d2a6a 100644
--- a/locale/circuitpython.pot
+++ b/locale/circuitpython.pot
@@ -113,6 +113,7 @@ msgstr ""
#: ports/espressif/common-hal/espulp/ULP.c
#: ports/espressif/common-hal/mipidsi/Bus.c
+#: ports/espressif/common-hal/qspibus/QSPIBus.c
#: ports/mimxrt10xx/common-hal/audiobusio/__init__.c
#: ports/mimxrt10xx/common-hal/usb_host/Port.c
#: ports/raspberrypi/common-hal/picodvi/Framebuffer_RP2040.c
@@ -856,8 +857,7 @@ msgstr ""
msgid "Coordinate arrays types have different sizes"
msgstr ""
-#: shared-module/usb/core/Device.c
-#: ports/espressif/common-hal/qspibus/QSPIBus.c
+#: ports/espressif/common-hal/qspibus/QSPIBus.c shared-module/usb/core/Device.c
msgid "Could not allocate DMA capable buffer"
msgstr ""
[and a few more changes]
oh! Scott's livestreaming sorry. I'm not able to watch. I'll ask later
dense glacier
tulip sleet
Here is the notes document for Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1GeBRhF1Mp4S-fEACLCzU_XVjj7yNxlFZuYesaJ1g9Rw/edit?usp=sharing
lone axle
<@&356864093652516868> the weekly meeting will occur here in #circuitpython-dev at normal time 2pm US Eastern / 11am Pacific. About 90 minutes from now. Add your hug reports and status updates to the notes doc if you'd like to participate https://docs.google.com/document/d/1GeBRhF1Mp4S-fEACLCzU_XVjj7yNxlFZuYesaJ1g9Rw/edit?usp=sharing. We look forward to hearing from any who can attend.
lone axle
Thanks everyone, have a great week 👋
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/10W0TOyUCAml9Tc6jSNZa5mgK05YGfCPuexfbpSFLteE/edit?usp=sharing
I don't seen to be able to pin/unpin messages currently. When I try to unpin previous meeting message it gives this fairly unhelpful error message. When I open the context menu on the new meeting message I don't see the "pin message" action in there like I would normally. Not sure if Discord made some change to permission behavior under the hood or if this is just a temporary issue, will try again in a while.
slender iron
I'll try
slender iron
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is...
worked for me. not sure if it is a permission thing or just a client issue
orchid basinBOT
willow totem
Can I grab someone to re-review/merge this as it's been mentioned as released in this weeks python newsletter. https://github.com/adafruit/circuitpython/pull/10778
slender iron
Can I grab someone to re-review/merge this as it's been mentioned as released in...
I'll look now
willow totem
I'll look now
Thanks Scott 
slender iron
np, feel free to ping me whenever something isn't getting a response but you are waiting for one
anything longer than 24 hours (assuming it is over a business day)
proven garnet
Is anyone else having their GitHub Actions running crawling and not starting jobs? It says I'm at my job concurrency limit but that doesn't appear to be true as far as I can tell.
slender iron
check you actions tab for other stuff running
proven garnet
check you actions tab for other stuff running
I didn't see anything in that repository, so maybe it was something else under my account? Seems to have resolved now.
But I do finally have a prototype for the imported memory analysis: https://github.com/adafruit/Adafruit_CircuitPython_ImageLoad/pull/104
manic glacierBOT
slender iron
I didn't see anything in that repository, so maybe it was something else under m...
Nice! I think it is across all of your repos
orchid basinBOT
orchid basinBOT
orchid basinBOT
manic glacierBOT
manic glacierBOT
orchid basinBOT
orchid basinBOT
proven garnet
Howdy! There's a couple PRs that would be good to merge today before the next bundle cronjobs, otherwise I am pretty sure it will think each bundle was entirely updated:
https://github.com/adafruit/Adafruit_CircuitPython_Bundle/pull/543
https://github.com/adafruit/CircuitPython_Community_Bundle/pull/274
proven garnet
Thank ya kindly 
candid sun
<@&356864093652516868> We'll have our weekly meeting in about 1 hour and 15 minutes from now in this text channel and in the circuitpython voice channel. Please take the time to add your notes in advance to the document: https://docs.google.com/document/d/10W0TOyUCAml9Tc6jSNZa5mgK05YGfCPuexfbpSFLteE/edit?tab=t.0#heading=h.jb0k8k2z2vqn -- I look forward to everyone's updates!
candid sun
going to need a minute or 2 to setup, previous meeting ran a little over
tidal kiln
👋
slender iron
thank you!
lone axle
Thanks for hosting Liz, have a great week everyone.
candid sun
thank you!
candid sun
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/17MhtQLKxqw0RLkAXOwIhFo6tgUB7oy7ZhBopKIDmMEU/edit?usp=sharing
orchid basinBOT
slender iron
@tulip sleet I'm having claude work on the idf6 update
tulip sleet
<@329766224093249548> I'm having claude work on the idf6 update
I saw that in notes 👍
slender iron
I'm poking the asyncio stuff too
orchid basinBOT
orchid basinBOT
tulip sleet
@devout jolt Scott is off until next week so you won't get PR feedback for a bit.
devout jolt
<@352910176736772096> Scott is off until next week so you won't get PR feedback ...
ah thanks. It's not a big deal. I think I found the problem. A bunch of new zephyr boards were added in the time between when I rebased and ran update_board_info.py.
devout jolt
Hey @tulip sleet or anyone else with merge privs on the "circuitpython" repo: Are squash merges done? I've got a PR with a mess of commits, some caused by rebasing while chasing zephyr updates, and I'd hate for all that to end up in the git history. Should I close that PR and offer up a cleaner one or will the mess get squashed?
tulip sleet
Hey <@329766224093249548> or anyone else with merge privs on the "circuitpython"...
No, we don't squash. We don't really care about all the commits. If you want to rebase and squash and then do a git push -f, that's OK, but we won't hold it against you for multiple commits.
we decided years ago not to squash
devout jolt
we decided years ago not to squash
good to know, thanks!
blissful pollen
I have had to rebase and force push more times then I want to admit, because me and git are not always friends
lone axle
Looks like the reports for https://circuitpython.org/contributing failed to generate
orchid basinBOT
kindred wren
@tulip sleet I have a couple of quick PRs up for circup that should be easy hammer ins.
pre-commit fix:
https://github.com/adafruit/circup/pull/272
error message adjustment:
https://github.com/adafruit/circup/pull/271
Thanks!
manic glacierBOT
random junco
I'm working on my first library (https://github.com/prcutler/CircuitPython_bambulabs/) that queries a BambuLabs printer over MQTT, which returns a JSON blob. My library handles the MQTT connection, processes the JSON and has a method to return each key in the JSON, like this:
def bed_temperature(self):
"""Current bed temperature in degrees Celsius."""
return self._print.get("bed_temper")
But ruff is complaining: bambulabs.py:60:1: PLR0904 Too many public methods (26 > 20) and so GitHub actions errors out. Is there a different way I should be doing this that to reduce the number of methods? Or is there a way around this?
lone axle
I'm working on my first library (<https://github.com/prcutler/CircuitPython_bamb...
near the bottom of ruff.toml there is an ignore list. You can add PLR0904 to it to get the tool to skip that rule. https://github.com/adafruit/Adafruit_CircuitPython_VCNL4030/blob/main/ruff.toml#L91
the file could be split into two separate ones. But I don't think it would really improve the readability or organization in this case. Not necessarily worth it just to appease the checker IMO. For community libraries you can always feel free to ignore any of the rules that you don't care about in your own code.
random junco
the file could be split into two separate ones. But I don't think it would reall...
thank you!
random junco
If anyone has a chance, I'm stuck on GitHub actions for creating my library and not sure how to troubleshoot: https://github.com/prcutler/CircuitPython_bambulabs/actions/runs/23710945555/job/69070471870 - it appears to be failing because it can't import my library bambulabs and wifi: WARNING: autodoc: failed to import 'bambulabs'; the following exception was raised: and ModuleNotFoundError: No module named 'wifi' - but isn't wifi a native module? And I've looked at other community libraries, and they seem to import themselves, so I thought I was doing that right
blissful pollen
If anyone has a chance, I'm stuck on GitHub actions for creating my library and ...
Since this seems to be about autodoc (I think?) is it just lines 1187/1188 about the TODO entries, the warnings are probably treated as errors?
As a total side note your copyright line in the your .py file is not updated either, just happened to see that.
random junco
As a total side note your copyright line in the your .py file is not updated eit...
Thanks
Removing the TODO entries from the README didn't seem to fix it, it's still giving me autodoc errors trying to import the two classes in bambulabs.py: https://github.com/prcutler/CircuitPython_bambulabs/actions/runs/23713090916/job/69075817995 - sorry for the repeat question, but I'm not getting it yet
random junco
As a total side note your copyright line in the your .py file is not updated eit...
Which file did you see that in? In bambulabs.py I have # SPDX-FileCopyrightText: Copyright (c) 2026 Paul Cutler - is that incorrect?
blissful pollen
Which file did you see that in? In `bambulabs.py` I have `# SPDX-FileCopyrightTe...
Okay maybe I need more caffeine, I saw the first line with Scott's name and missed yours below, oops
random junco
yeah, I wasn't sure why that was there either, but I checked some random community libraries and they all had it. 🙂
lone axle
If anyone has a chance, I'm stuck on GitHub actions for creating my library and ...
for CircuitPython core modules that aren't part of CPython we have to add them to this list of "mock imports" in docs/conf.py https://github.com/prcutler/CircuitPython_bambulabs/blob/main/docs/conf.py#L28
it looks like wifi is the only one that needs to be mocked for your library I think.
Neat utility by the way. Thanks for publishing it!
random junco
Thanks, I appreciate the help. And it's my pleasure - todbot found an Arduino library that inspired it
Ok, so autodoc doesn't complain about wifi anymore, but it still complains about the bambulabs library itself: WARNING: autodoc: failed to import 'bambulabs'; the following exception was raised: on line 1123, and then fails on the 2 classes within the library. Right above that on line 1122 it says reading sources... [ 67%] examples - is it reading the simpletest I made that imports bambulabs and failing there? I feel like I'm so close and missing something obvious
blissful pollen
I see this now I didn't before ValueError: Unsupported radio class: radio does that also need to be included in that autodoc line?
random junco
trying now
Nope, didn't seem to matter, same error: https://github.com/prcutler/CircuitPython_bambulabs/actions/runs/23714395801/job/69079062794 - line 1124: WARNING: autodoc: failed to import 'bambulabs'; the following exception was raised:
blissful pollen
Not sure if I know much more, if you look down line 1205/1207 it tries to show the root cause. I think
random junco
ah, missed that
The only time radio is called is as wifi.radio when setting up wifi so it can use MQTT in bambulabs.py in lines 39 & 40. I have no idea why it's failing as ImportError: (ValueError('Unsupported radio class: radio'), as it's not a class (I think)
blissful pollen
I thought Radio in radio is a class but I haven't got into autodoc so kinda at my limit of knowledge
random junco
no problem, I appreciate the help - thank you!
lone axle
I think this may be a bug or issue with connection manager and/or the docs build, not on your end. I'm not positive though. I'll try to look further into it when I can.
random junco
Thank would be great, thanks in advance
random junco
I think this may be a bug or issue with connection manager and/or the docs build...
Hi Tim - todbot helped me out and suggested I move the pool, ssl_context, and MQTT setup into the BambuPrinter class and it passed! So I'm good, unless there's a different way I should be doing it. Thanks again.
lone axle
Nice, glad it was easy to work around.
random junco
I'm trying to publish this library to pypi and getting a 403 forbidden error. The Learn Guide just says Note: for a release to be deployed to PyPi, it requires the PyPi credentials to be added to the repository secrets. - is that as simple as creating two different secrets, pypi_username and pypi_password in Settings / Secrets & Variables? The pypi docs reference using tokens, so I'm a bit confused: https://packaging.python.org/en/latest/guides/publishing-package-distribution-releases-using-github-actions-ci-cd-workflows/
random junco
Also new accounts require 2FA, which makes me wonder if that might be the culprit
lone axle
I'm trying to publish this library to pypi and getting a 403 forbidden error. T...
IIRC you need to use an "API token" in the password field. And I think they are upper-case but I'm not sure if it actually matters. like PYPI_USERNAME, PYPI_PASSWORD. The password gets set to the API token which can be generated when you are logged in to a pypi account. It's down some ways on the account settings page.
random junco
aha, tokens! I see it now, trying again, thank you
lone axle
and I do think setting up 2FA is required, though I'm not sure of that either.
random junco
it worked! Thanks again for all the help
lone axle
🎉 You're welcome
slender iron
<@&356864093652516868> We'll have our weekly meeting in about 80 minutes from now in this text channel and in the circuitpython voice channel. Please take the time to add your notes in advance to the document: https://docs.google.com/document/d/17MhtQLKxqw0RLkAXOwIhFo6tgUB7oy7ZhBopKIDmMEU/edit
random junco
Next week's episode of The CircuitPython Show , I interview Esha Patel who is in that article, along with her classmate Abby Bergman
tidal kiln
yah - a pico build seems like a good idea
whatever version/flavor makes sense
^^ of pico. 1 or 2. wifi or non-wifi. etc.
👋
lone axle
Thanks for hosting Scott, have a great week everyone.
slender iron
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1rLHPbsMA2tOnUd3Ok-1xXsTQ9ts8WD4ZetJRcW7pmIE/edit?usp=sharing
