no voice
#circuitpython-dev
wraith crow
lone axle
Thanks for hosting Liz! have a great week everyone 👋
thorny jay
Thank you @candid sun for hosting/reading.
main furnace
Where do I file Open Installer issues?
candid sun
Here is the notes document for the next CircuitPython Weekly Meeting in two weeks on August 25, 2025. 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/1YxMrgfSw5AcotMzRo8oWV4AOjsYhb9v2mWQmUSDRz9E/edit?tab=t.0
tiny peak
Does anybody know if CircuitPython already uses some kind of compression on QSTR strings (e.g. Hamming Huffman coding, LZW, or whatever)?
As I begin looking into fixing USB host glitches, I'm getting the sense I'll need to find ways of offsetting code size increases from any error handling code I might add.
It seems like most the stuff that can be turned off for the SAMD21 M0 boards has already been turned off.
slender iron
Does anybody know if CircuitPython already uses some kind of compression on QSTR...
qstr aren't but error messages are
tiny peak
thanks
does it seem plausible that error string compression is totally bypassed for non-ASCII strings? I'm looking at check_non_ascii().
by my casual reading of makecompresseddata.py, it looks like there might be a significant opportunity for improving the coding of non-english text.
lone axle
Github having problems with issues, actions, PRs again today https://www.githubstatus.com/
onyx hinge
@candid sun if the fruit jam schematic is up on Learn, can you point me at it .. and if not, can you send me a copy as png/pdf/whatever?
candid sun
<@347203168121126912> if the fruit jam schematic is up on Learn, can you point m...
guide isn't live yet but can you see this link? https://learn.adafruit.com/admin/guides/4298/editor/30676
onyx hinge
@candid sun no, I can't see unpublished guides anymore either
candid sun
<@347203168121126912> no, I can't see unpublished guides anymore either
here are PNG files
onyx hinge
here are PNG files
Thank you!
grizzled maple
Where's the best place to ask about the MatrixPortal S3 and its behavior with CircuitPython?
I get flickering updates in circuitpython, but not in C++
slender iron
Where's the best place to ask about the MatrixPortal S3 and its behavior with Ci...
Support forum to start.
tiny peak
For people who've worked on USB stuff:
Do you have any favorite books or other reference documentation for understanding the USB specs?
stuck elbow
the usb spec itself is a good starting point
tiny peak
Anybody have experience with "USB Complete" or "USB Embedded Hosts", both by Jan Axelson?
slender iron
Anybody have experience with "USB Complete" or "USB Embedded Hosts", both by Jan...
usb complete is my goto reference
tiny peak
okay. thanks. I'll probably get a copy of that then.
slender iron
did you get a sniffer?
tiny peak
yeah. talked to pt and ordered one today.
slender iron
k, great!
manic glacierBOT
lone sandalBOT
manic glacierBOT
manic glacierBOT
tulip sleet
@lone axle fruit jam volume discussed here: https://forums.adafruit.com/viewtopic.php?t=219717
I'll wait for more info.
lone axle
Thank you
manic glacierBOT
ionic elk
is there a requirement to keep use of float and int64_t to the port level? I don't see references to them in the shared-bindings layer.
slender iron
is there a requirement to keep use of `float` and `int64_t` to the port level? I...
micropython has typedefs for some of them. we use int64 for ticks so I think we do use it in the core
ionic elk
Ok, maybe should just throw a NotImplemented for Int64 message types like I do with Float64 then
I'd like to keep float, so I'll ask in my PR whether I need to reshuffle stuff down to the common-hal
slender iron
int64 should be fine. the vm supports them. they just aren't small ints
not sure what it does for float64
ionic elk
For context, I'm storing values from Circuitpython variables to internal objects that will be passed to micro-ROS. My impression was that this would be port-agnostic so I haven't put it in the common-hal so far, but I could be wrong. I don't think Circuitpython floats support double precision, right? If circuitpython has no vm variable for a double precision integer, there's no point in me supporting that either
Unless I offered something kind of janky like a manual byte array. Doing that would depend on whether there's a use case for it. Could always revisit later.
slender iron
right, no double support that I know of. MP might have it. I haven't looked. long ints should be on
tulip sleet
weren't long ints disabled on samd21?
only on 192kB builds and a few larger builds (like CPX displayio, I think)
devout jolt
Has anyone on MacOS been seeing CIRCUITPY drives not being mounted (or even visible from df) after a few unplug/replug cycles? I've been getting it a lot these past few days, with nothing I can think of having changed on my Macbook. A reboot of the Mac will fix it for a while. Killing the Finder doesn't help. It's not a TinyUSB issue, even the RP2 ROM UF2 bootloader won't show up.
Edit: The board is still there: shows up in USB Prober, can connect to via USB CDC if CircuitPython, just no USB MSC
ebon horizon
Has anyone on MacOS been seeing CIRCUITPY drives not being mounted (or even visi...
I didn't check with df, but I've sometimes needed to go into the Disk Utility GUI tool to manually re-mount the drive
I can't replicate when and why though
devout jolt
I didn't check with `df`, but I've sometimes needed to go into the `Disk Utility...
yeah not in Disk Utility or diskutil. Very weird. Well I just did the Apple Silicon versions of the "shake-a-chicken" incantations (SMC reset by powering down fully for 30+ seconds and NVRAM clear by booting w/ held power button to get into options to do nvram -c). Let's see if those make difference
ebon horizon
yeah not in Disk Utility or `diskutil`. Very weird. Well I just did the Apple ...
and this isn’t a case where the USB port’s power is disabled?
devout jolt
and this isn’t a case where the USB port’s power is disabled?
nope. If the board is CircuitPython, I can connect to it via USB CDC and it shows up normally in System Information or USB Prober. Just not USB MSC
ebon horizon
Ah, that's interesting.
devout jolt
Yeah super weird and frustrating
manic glacierBOT
tulip sleet
@lone axle re the TLV gain settings and the speaker: It appears that the power-up value for the speaker amp gain is 6dB (lowest), and for the attenuator, the power-up gain is -78dB. That is, it is as quiet as possible when first powered up. The CircuitPython driver does not mess with these values on instantiation. So I'm not sure there's anything to do except maybe add some warnings. If there is a volume setting API in a particular Fruit Jam library, it could do some range limiting (and there could be a boolean to override that, for instance). Not sure if that makes sense or not
lone axle
<@382939733107408897> re the TLV gain settings and the speaker: It appears that ...
I was thinking to add warning print outs to all of the different volume/gain property setters inside of the TLV320 driver directly. That way any code downstream that sets the volume to anything higher than default will result in warning being printed.
that will include Fruit Jam library and OS, and any other project code.
tulip sleet
Are you saying there would be a new API that can set new ceilings on the various gains (volume control and speaker gain)? If you don't change the ceilings, it will limit the gain to the set ceilings and print a warning. I think that's what you mean?
the default ceilings would be safe for the included speaker
a ceiling on headphone volume would probably be good too, to save people's ears, though headphones have different characteristics.
lone axle
Are you saying there would be a new API that can set new ceilings on the various...
I was imagining a warning only, no enforcement of the ceiling. But am open to the idea of enforcing it too.
Maybe a loud_volume_protection boolean or similar as an init argument that defaults to True and disallows setting the level over a threshold.
tulip sleet
A printed warning is not going to be seen unless someone has the terminal open. And if they want to go above the ceiling (because they have a big speaker or whatever), then warnings you can't turn off would be annoying.
I was thinking of more quantitative limits instead of just a boolean. But the defaults would be appropriate for the supplied speaker. Could add .*_max setters/getters for each gain setting, or have a single call that sets them all (with optional arguments whose default values are the default ceilings)
There are a lot of volume and gain settings: https://docs.circuitpython.org/projects/tlv320/en/latest/api.html
lone axle
Okay, yeah re-reading your comment and thinking more I was coming to the same conclusion about numeric values instead of boolean.
We'll need to check how dac_volume interplayes with speaker_volume and headphone_volume.
tulip sleet
left/right, manual headphone, regular headpohone
lone axle
yeah that makes it trickier to effectively limit / warn
tulip sleet
does fruit jam os or the fruit jam library have a simplified volume/gain API?
The TLV library does not really need this API, I think.
It just needs warnings in the guides
lone axle
Not currently. It just exposes fruit_jam.dac currently and user code can do whatever with it.
tulip sleet
we are really talking about a particular speaker
lone axle
So, that is the place to start maybe. A helper volume function in adafruit_fruit_jam(and maybe helpers for setting headphone / speaker while at it) that simplifies the volume API down to only 1 thing, and then put the warning and enforcement there.
tulip sleet
a simple volume API for speaker and headphones (that does left/right together, etc.) might be welcome. And that API can do the limiting. This is off the top of my head. I haven't looked at the libraires at all
if the user wants finer control, it is up to them to use the .dac safely
what do all the non-Python ported games, etc., do?
did someone just choose appropriate volume ranges?
lone axle
I think so. For all of mine I believe I just copied the code from the TLV320 guide which included setting the volume levels. I assume Liz chose those values when writing the guide but havn't discussed with her.
Some of the existing projects use TLV320 library directly, they were written before the fruit_jam library. Once this change is in we ought to probably change them to use adafruit_fruit_jam so that all of our projects come with the warning / cieling mechanism that we add.
lone axle
@wraith crow maybe hold off on further audio changes in your PRs for the moment. We are going to refactor the volume handling in Fruit Jam OS and existing apps to use a new simplified API in adafruit_fruitjam (see above conversation for details). I'm working on this now and it would be good to just do any changes to the apps afterward ideally.
turbid radish
I think an API for handling the sound would be welcome and provide a common codebase. The tvl320 is kinda tricky in some of its settings.
tiny peak
foamyguy & dan, re: discussion on DAC volume... FWIW, in my experience with doing a portable synth project for Fruit Jam, it was pretty useful to be able to directly access dac.speaker_mute, dac.dac_volume, and dac.headphone_volume. In my case I only cared about line level or headphone level output, but I can see how other projects might want to make more use of the built in speaker. I really like the idea of having the FruitJamOS API provide a simplified volume control interface with guard rails, but also letting people use adafruit_tlv320 directly if they want finer-grained control for synth projects.
lone axle
foamyguy & dan, re: discussion on DAC volume... FWIW, in my experience with doin...
I have submitted with some proposed changes. https://github.com/adafruit/Adafruit_CircuitPython_FruitJam/pull/13 This adds a new unified and simpler 1-20 volume scale on the adafruit_fruitjam.peripherals object with a limit argument to prevent it from going too loud without deliberate action from user code.
The dac remains accessible directly via fruit_jam.dac so all control that is possible with it can still be done too, but it will bypass the new simpler audio API and thus be a bit more "use at your own risk" with regards to damaging speakers.
Once the API is settled and merged I intend to go over the existing projects from Fruit Jam OS / learn and change any that access the dac directly to use this new API instead.
tiny peak
@lone axle Nice. After reading some of your and others' comments, I'm starting to suspect that the current adafruit_tlv320 library may include some AI hallucinated stuff, or maybe things with swapped endianness. For example, the Functional Block Diagram on page 19 of the datasheet shows that the left and right DAC blocks go to the speaker outputs by way of a 0 dB to -78 dB attenuation stage followed by a 6 dB to 24 dB class-D amplifier stage. That does not match the API documentation for dac.speaker_volume
similar problem with dac.headphone_volume: API docs say
The current headphone volume in dB. :return: The volume in dB (0 = max, -78.3 = min)
but the Functional Block Diagram shows a 0 dB to -78 dB attenuation stage followed by a 0 dB to 9 dB Class A/B amplifier stage
lone axle
@tiny peak It's possible there are bugs in the driver. I don't have any real knowledge or experience with audio or DACs so all of the different ranges listed are a bit of a mystery to me. If you find parts that don't match the datasheet feel free to submit PRs if you end up finding changes that make it match the datasheet more closely.
tiny peak
I may look into this some more. The datasheet is a 111 page monster and this DAC has quite a lot of features. It's not exactly straightforward.
lone axle
yeah, it's definitely got a lot going on inside of it. Docstrings and implementation details may have come from the Arduino version of the driver as well which the CircuitPython ones are typically ports of/from.
tiny peak
reading some more of the datasheet... there may actually be a third gain stage... sounds like maybe the chain is digital volume control for each of the left/right DAC, then analog attenuation, then analog amplification.
It kinda sounds like the designers' intended way for applications to control volume is to mainly adjust the digital left/right volume registers because those include a "soft-stepping" feature to spread the changes out over many samples to avoid clicks and pops. So, like, maybe you're supposed to dial the analog stuff in for your hardware in a semi-set-and-forget kinda way, then use the digital gain at runtime? Not sure yet.
upper radish
It's early stages, but it builds
tiny peak
woah... what's that? do you have any plans for hooking it up to WebUSB, WebAudio, and such?
upper radish
woah... what's that? do you have any plans for hooking it up to WebUSB, WebAudio...
I've been working on a VS Code extension for over a year now and this is meant to be part of the 'shell'/backend of a custom terminal, rather like this, but CircuitPython: https://github.com/WebReflection/micro-repl
(The broad strokes are blocked in: the REPL terminal, serial port connection, data plotting, syntax highlighting, and workspace management. Now working on refining everything. I've playfully been calling it Mu 2)
tulip sleet
reading some more of the datasheet... there may actually be a third gain stage.....
I looked at the datasheet to try to find the power-on reset values for the volume/gain registers. It was indeed pretty confusing: the diagram (repeated twice, I think), shows "attenuation" but that's not the name in the register descriptions, I think.
tiny peak
I looked at the datasheet to try to find the power-on reset values for the volu...
I spent all evening playing with this stuff, including datasheet spelunking and setting up an audio scope to check amplitude changes. One result of playing with volume properties is that it seems like perhaps dac.headphone_volume has no effect at all on the output level?!? That's kinda weird. I plan to keep looking at this. Notes so far are at https://github.com/samblenny/tlv320dac-stuff
thorny jay
I am slowly catching up on CPday video, and in the "What is new in 10.0" there was discussion about /sd and another partitions that would be visible (at least with the FruitJam) when you connect to your computer.
So I was wondering on the (new) meaning of storage.disable_usb_drive() would that hide all the USB partition, or do we need more granularity (like I want SD card to be visible but not CIRCUITPY drive).
tulip sleet
I am slowly catching up on CPday video, and in the "What is new in 10.0" there w...
Good point that you might want more granularity. storage.disable_usb_drive() turns off USB MSC entirely. Fine to open an issue.
tulip sleet
I spent all evening playing with this stuff, including datasheet spelunking and...
See also https://github.com/adafruit/Adafruit_CircuitPython_TLV320/pull/8#pullrequestreview-3137579378 for some other weirdness.
tiny peak
huh... that's weird
tulip sleet
@slender iron did you want me to review this? https://github.com/hathach/tinyusb/pull/3217
you mentioned tinyusb when we talked
slender iron
Yes please
tiny peak
I think I may try hooking the Saleae up to the SDA and SCL pins on the GPIO header to see if I can watch what the driver is actually doing for the volume setters and getters
tulip sleet
I think I may try hooking the Saleae up to the SDA and SCL pins on the GPIO head...
have you instrumented the python getters and setters first?
tiny peak
no. probably won't get to looking at it until tomorrow
tulip sleet
no. probably won't get to looking at it until tomorrow
there might be some errors in the register numbers, etc.
also might try/look at the arduino library, which was probably the basis for the CPy code: https://github.com/adafruit/Adafruit_TLV320_I2S
tiny peak
@tulip sleet skimming through Adafruit_CircuitPython_TLV320
/adafruit_tlv320.py, I see that def speaker_volume() calls self._page1._set_spk_volume(), which includes the line:
self._write_register(_SPK_VOL, value)
Then, _SPK_VOL is defined as _SPK_VOL = const(0x26) (page 1, register 38). On the Functional Block Diagram in section 1.4, Page 1 / Register 38 is "Analog Attenuation" for the left speaker channel (range 0 dB to -78 dB). That path also goes through the "Mono Class-D Speaker Driver" stage (Page 1 / Register 42=0x2A, range +6 dB to +24 dB). That register gets set by _configure_spk_pga(), which gets called by speaker_gain().
It's not clear to me why the registers for the right channel attenuation and speaker gain seem to be ignored.
but... it does seem at first glance that the bitfield shifting and masking for Page 1 / Register 42=0x2A are implemented correctly for the amplifier gain bits and mute bit
The most suspicious thing I see is this:
@speaker_volume.setter
def speaker_volume(self, db: float) -> None:
"""
:param db: Volume in dB (0 = max, -63.5 = min)
"""
if db > 0:
db = 0
gain = int(55 + (db * 1.14))
gain = max(0, min(gain, 127))
self._page1._set_spk_volume(route_enabled=True, gain=gain)
The comment says 0 dB to -63.5 dB, but that's the range for the digital volume control (Page 0 / Register 65=0x41 and Page 0 / Register 66=0x42). Also, the gain scaling and clipping does not seem appropriate. The correct range for Page 1 / Register 38=0x26 is 0 dB to -78 dB (table 6-106 in the datasheet)
So, it looks like the driver is conflating the digital volume adjustment stage (Page 0, registers 64..66) with the analog attenuation stage (Page 1, registers 38 and 37 for speakers plus maybe 36 for headphones)
lone axle
@tulip sleet do you know if there is any support in CircuitPython for hmac? Ladyada mentioned it may have been added at some point in the crypto stuff for matter or something else perhaps.
tiny peak
The proper values for setting Page 1 / Register 38=0x26 are listed in Table 6-24
tulip sleet
<@329766224093249548> do you know if there is any support in CircuitPython for `...
We have some hmac code for CircuitMatter. I might have gotten it to run on CPy, but I don’t remember. Scott was making CircuitMatter work in CPython, and the I started working on porting that to CPy. I’ll look more tomorrow.
tiny peak
I made a scatter plot of TLV320 datasheet table 6-24 for the speaker volume register value vs. analog gain dB. It turns out to be a non-linear piecewise function, so the linear formula in speaker_volume() can't be right.
tiny peak
@tulip sleet since it will be hard to keep track of the above dac volume research notes here in the chat log, I made an issue for it and copied all the stuff over: https://github.com/adafruit/Adafruit_CircuitPython_TLV320/issues/9
manic glacierBOT
fiery pecan
Does CircuitPython make use of RTT (https://www.raspberrypi.com/documentation/pico-sdk/runtime.html#group_pico_stdio_rtt)? I think it would be a more effective alternative to https://adafruit-playground.com/u/SamBlenny/pages/circuitpython-core-dev-debug-tricks#fruit-jam-led-debugging-3205599. You don't even need the official Raspberry Pi Debug Probe. See the section labeled "Debug with a second Pico or Pico 2" under Appendix A of Getting started with Raspberry Pi
Pico-series and the comments on https://github.com/raspberrypi/debugprobe/issues/175.
tiny peak
Does CircuitPython make use of RTT (<https://www.raspberrypi.com/documentation/p...
woah... I didn't know about that! It sounds potentially really useful. Have you used that before with Pico-SDK on an RP2350? Do you have an impression of how much delay it adds? Does it work inside interrupt handlers? Do you need some special software to get the RTT messages? I actually have a pico debug probe, and the Fruit Jam board I'm using has a debug port. So, if that all works, it might be an interesting solution for me.
My tentative plan is to do some profiling in the USB stack using several GPIO pins and Saleae's Logic 2 software. If I set it up right, I'm expecting the resulting logic analyzer trace pulse waves will act kind of like a flame graph showing how much time gets spent in which function. Can RTT be used to do that sort of thing?
fiery pecan
woah... I didn't know about that! It sounds potentially really useful. Have you ...
I haven't tried it with my Pico 2, but according to my understanding of the "Target implementation" section of https://www.segger.com/products/debug-probes/j-link/technology/about-real-time-transfer/, it's just an in-MCU-memory ring buffer of log messages that's shared between the target and the debug software. Thus, I'd assume it'd work inside interrupts, but I don't know for sure.
fiery pecan
I haven't tried it with my Pico 2, but according to my understanding of the "Tar...
You'd probably want to configure it to only use non-blocking mode, which according to that linked webpage:
In non-blocking mode only data which fits into the buffer, or none at all, will be written and the rest will be discarded.
tiny peak
Do you have software you like for viewing RTT messages?
fiery pecan
Do you have software you like for viewing RTT messages?
This is specific to Rust, but I like the ecosystem around the defmt crate: https://defmt.ferrous-systems.com/. I don't have much experience coding in C/C++.
tiny peak
ah... okay. I just came across a mention of that in connection with probe-rs as I've been reading up on RTT tools. Sounds like Segger has a tool called J-Link RTT Viewer for Linux that might possibly work, but it's not clear to me if that is tied down to only support their own debug probes.
fiery pecan
ah... okay. I just came across a mention of that in connection with probe-rs as ...
It looks like OpenOCD has support: https://openocd.org/doc/html/General-Commands.html#Real-Time-Transfer-_0028RTT_0029
tiny peak
hmm... interesting. I used openocd with the Raspberry Pi debug probe and a Feather RP2350 when I was messing with Zephyr stuff a while back. That worked pretty well. Thanks for the tips!
tiny peak
It looks like OpenOCD has support: https://openocd.org/doc/html/General-Commands...
I just checked, and TinyUSB has an option to use RTT for its debugging output. That could potentially be really useful.
tidal meteor
Hello. I found a potential bug in the HX711 library of CircuitPython. I opened an issue 7 days ago : https://github.com/adafruit/Adafruit_CircuitPython_HX711/issues/7. Do I need to do something more ?
manic glacierBOT
Overheating problem is also noticed in my case. With a basic LVGL application with only a spinner, the part of screen near the USB connector is getting really hot. In fact, screen in this area is getting black, making the gadget unusable.
tiny peak
@lone axle are you actively working on volume setting fixes in the Adafruit_CircuitPython_TLV320 driver? (asking to avoid duplicating work)
lone axle
<@382939733107408897> are you actively working on volume setting fixes in the Ad...
I am not. My primary aim is to get the simplified version in the Fruit Jam library in place and then update the rest of our existing code to use it. It seems to work as is for moving the volume up and down. And we had some folks get blown out speakers so having the limiter in place will be good I think.
Improvements to the base TLV driver are welcomed but not something I'm actively working on at the moment.
tiny peak
okay. I'll continue with what I'm doing. There appears to be at least one significant bug in calculating speaker_volume register values. Maybe other things.
lone axle
Thank you for looking into it!
tiny peak
Yeah. This is fun stuff. Excited to see it all up and working and boards in stock in the shop.
Is there a guide somewhere that says how to do test builds for libraries to see if the .mpy and documentation builds will run without errors?
I found this: https://docs.circuitpython.org/en/latest/docs/design_guide.html, which looks helpful for making edits. But, I didn't see obvious mention there of how to do builds. Perhaps I overlooked something.
lone axle
Is there a guide somewhere that says how to do test builds for libraries to see ...
I don't think there is a guide. Whenever I need to build .mpy I end up following breadcrumbs from the actions to the build tool. I can get you the command that builds mpy versions for a library in a moment..
lone axle
@tiny peak The code that does it inside of the actions task is here: https://github.com/adafruit/workflows-circuitpython-libs/blob/3413c80af1b610fba8d653deec5bd9014a16895d/build/action.yml#L106-L110. I'm not 100% certain what all of the arguments get used for in that context but I've not found them to be terribly important for local testing. I usually do something like:
circuitpython-build-bundles --filename_prefix adafruit_fruitjam --library_location .
with filename_prefix filled in for whatever library I'm working in. This command gets run in the root of the library repo and builds the 9.x and 10.x mpy bundles. It requires https://github.com/adafruit/circuitpython-build-tools
tiny peak
<@1220648290941669377> The code that does it inside of the actions task is here:...
awesome. Thanks! I'll give that a try.
manic glacierBOT
tiny peak
@tulip sleet what's the proper procedure to fix stuff like this after pulling newly merged upstream PR's into a local fork feature branch:
$ git status
On branch usb-find-unplug
Changes not staged for commit:
(use "git add <file>..." to update what will be committed)
(use "git restore <file>..." to discard changes in working directory)
modified: frozen/Adafruit_CircuitPython_DotStar (new commits)
There were initially a bunch of git status lines about modified: frozen/... stuff. When I ran make update-frozen-libraries, it fixed most of them. But, the dotstar one is still left over.
tulip sleet
<@329766224093249548> what's the proper procedure to fix stuff like this after p...
I have this alias for getting the submodules at the right commits:
alias gitsubupdate='git submodule sync --quiet --recursive && git submodule update --init --filter=blob:none'
does what I saw mean there needs to be a PR to update the DotStar version in the CircuitPython repo?
kinda looks like the submodule commit ref is one tag behind the latest in the DotStar repo?
When I did this, it got git status to stop complaining:
$ cd frozen/Adafruit_CircuitPython_DotStar/
$ git checkout 163f2f166aee11d82303492bb1e5af4937e57b62
latest Adafruit_CircuitPython_DotStar tag points to 4b0ba64, so make update-frozen-libraries was one tag ahead of the submodule ref.
looks like that was something about fixing SPI locking to fix a funhouse crash
crimson ferry
busio.UART code shows
//| class Parity
(https://github.com/adafruit/circuitpython/blob/6ed41a0d2f7422c89410d53a3ef3156613384ae5/shared-bindings/busio/UART.c#L371)
readthedocs shows
class busio.Parity
(https://docs.circuitpython.org/en/latest/shared-bindings/busio/index.html#busio.Parity)
but it should be busio.UART.Parity?
tulip sleet
looks like that was something about fixing SPI locking to fix a funhouse crash
Yes, I just made that change. We'll update all the frozen modules before the next 10.0.0 beta release. We always do.
tulip sleet
`busio.UART` code shows
> //| class Parity
(<https://github.com/adafruit/circuit...
Yes, it should be, I just checked in the REPL.
crimson ferry
I checked it too, thought it would be a quick PR, but wasn't sure how to make it right
//| class UART.Parity and busio gets automagically pre-ended?
tulip sleet
I checked it too, thought it would be a quick PR, but wasn't sure how to make it...
I think the indentation for Parity might be wrong
crimson ferry
I'm not set up to build right now, but I could try that somewhat blindly... I don't suppose there's a way to test docs non-locally
tiny peak
danh, thanks for all the work wrangling releases. makes my head hurt just watching from the sidelines.
so. many. dependencies.
exciting to get all the new synthio stuff though
tulip sleet
danh, thanks for all the work wrangling releases. makes my head hurt just watchi...
a lot is in my head after 8 years of working on this 🙂
lone sandalBOT
wraith crow
I'm working on a new board (https://www.olimex.com/Products/RaspberryPi/PICO/RP2350pc/) and I've got an initial CP firmware working with the DVI display, but when I try to enable PS RAM using #define CIRCUITPY_PSRAM_CHIP_SELECT (&pin_GPIO8) the only thing I get on the serial console is "Serial console setup", nothing comes up on the USB console. Assuming their documentation is correct and the CS line is indeed GPIO8 is there anything else needed to get SPI PSRAM working on an RP2350 board?
ah ha 😁 I was just cleaning up and not expecting any change but the serial console displayed "tlsf_add_pool: Memory size must be between 20 and 9 bytes." A lead to follow......
and with CIRCUITPY_DEGUG_TINYUSB 1.... This prints continually
Serial console setup
*** PANIC ***
HardwSerial console setup
I believe I'm crashing on this instruction:
// Try and read the PSRAM ID via direct_csr.
qmi_hw->direct_csr = 30 << QMI_DIRECT_CSR_CLKDIV_LSB |
QMI_DIRECT_CSR_EN_BITS;
in ports/raspberrypi/supervisor/port.c
This is deeper into the hardware than I was hoping to have to look :/
tiny peak
I believe I'm crashing on this instruction:
```
// Try and read the PSRAM ID...
Have you confirmed that you actually need to do anything to the chip select pin? Comparing the Fruit Jam and Olimex schematics, they both have CS pullups. The Fruit Jam doesn't have anything else wired up to the CS pin.
perhaps I'm misreading it. not sure
oh... I did miss some other stuff on the PSRAM_CS net. Looks like the Fruit Jam has that going to GPIO47 on the RP2350. hmm
What are you basing the new board off of? Is your source build defining some serial console stuff for TinyUSB debug or ESP32 AirLift serial out?
maybe those are trying to use GPIO8? I say that because the Fruit Jam ESP32-C6 module uses that pin for serial debug
wraith crow
Looking at the two schematics it did look like psram should work fine, I think the Metro RP2350 even used GPIO8
for the PSRAM chip select
My first message was probably misleading, the Olimex board is the new board (new for me 😁)
tiny peak
yeah, I got that part, I'm just wondering if you copied an existing board def to create the olimex board that you're working on
like, maybe there's some lingering use of GPIO8 defined in a spot that you overlooked?
wraith crow
I think I started from the Metro RP2350 but it could have been the Fruit Jam, I definitely borrowed from both
I looked pretty carefully for any GPIO8 conflicts, but I'll give it another look
tiny peak
what serial pins is tinyusb set to use?
wraith crow
I get a bit confused about the differences between tinyusb/uart/usb resources. The serial UART is using 0 & 1
tiny peak
I see in ports/raspberrypi/boards/adafruit_fruit_jam/mpconfigboard.h that there's a #define DEFAULT_UART_BUS_TX (&pin_GPIO8)
wraith crow
#define DEFAULT_UART_BUS_RX (&pin_GPIO1)
#define DEFAULT_UART_BUS_TX (&pin_GPIO0)
I'm trying to remember where I found the uart pin config stuff for TinyUSB. I was just looking at that the other day. Might be off in lib/tinyusb somewhere.
wraith crow
I think there's a way to turn off USB serial, maybe I'll dig up the setting and give it a try just to make sure there's no conflict from there....
err, I'm thinking that was on ESP and you could use the esptools to flash the bin. I don't know if there's an easy option on the RP2350 to do the same....
tiny peak
When you saw the error messages after doing the CIRCUITPY_DEGUG_TINYUSB 1 thing that you mentioned, are you sure that was the same error as was happening without enabling tinyusb logging?
I'm wondering if there's some lurking thing that defaults to sending TinyUSB uart logging data to GPIO8.
wraith crow
I decided the tinyusb PANIC message was a red herring and didn't really think about it much more. GPIO8 is used on the Fruit Jam to communicate with the ESP network coprocessor though so I suspect that would have problems if there was a conflict with tinyusb.
tiny peak
how long after the board boots are you hitting these errors? As I mentally review things I've heard people mention in relation to weird low-level bugs with bringing up different boards, I remember sometimes there's talk of adjusting oscillator startup delays.
that's about the extent of my knowledge on such things. curious to hear what you come up with if you figure it out.
wraith crow
Maybe a second after power up, I am using the PICO_XOSC_STARTUP_DELAY_MULTIPLIER 64 setting from the Fruit Jam and the PSRAM is the same part number.
Thanks, I appreciate the brainstorming, if I get it figured out there will probably be a PR to add the board 😁
tiny peak
yeah. good luck. I'm very curious to learn more about weird ways things can go wrong as I'm presently trying to track down some odd behavior in the USB stack.
wraith crow
yeah. good luck. I'm very curious to learn more about weird ways things can go w...
Breakthrough! Don't have the answer yet, but after talking I noticed that the feather_rp2350 uses the same psram chip and uses GPIO8 for the CS line and I just loaded up that firmware from circuitpython.org and it loads and sees the memory 😁
tiny peak
nice!
I suppose you could just do a recursive diff on the board directories for the feather and your new-board-in-progress
maybe something will jump out
rogue ifdef... who knows
wraith crow
Yep, but unfortunately one of the first things that jumps out is that the feather is an rp2350a and this board is the rp2350b so there's going to be differences in how the port does things even if the board files are exactly the same. But at least I know the chip works and can be seen by CP
wraith crow
Well that was a lot of wasted time and noise for nothing, it had something to do with my build environment. I now have the dvi and the psram working, on to the next peripheral 😁 but that's enough for tonight!
crimson ferry
Yep, but unfortunately one of the first things that jumps out is that the feathe...
does the 2350B differ in other ways besides just more pins exposed?
wraith crow
does the 2350B differ in other ways besides just more pins exposed?
Not sure, I was afraid it did before I discovered, my problem was in my build environment
tiny peak
For people interested in improved TLV320DAC volume controls
( 🔥 🔊 😱 🚒 ), I've got a draft PR in progress at https://github.com/adafruit/Adafruit_CircuitPython_TLV320/pull/10
It's pretty far along. I found a few issues with the headphone signal chain, which my PR has fixes for. That includes a lower default volume for safe use with earbuds. I've tested that part, and I'm reasonably confident it's good. I also revised the doc-comments so the html docs built by Sphinx will be more complete (include some constants that got left out, etc).
Speaker signal chain still needs testing.
tiny peak
I tested the speaker, fixed a bug, and now it's ready for review and testing. Could probably use several sets of eyes and ears reviewing the proposed changes. From looking at the driver code, it seems like the default speaker volume was actually increased at one point, so there may be a project guide somewhere that uses the TLV320DAC with a higher wattage speaker. These changes are aimed at the 8Ω 1W speaker for Fruit Jam, so I'm not sure what the effect on other projects might be.
lone axle
@tulip sleet when you have a moment I'm curious about hmac from back on the Matter work. Is/was it ever usable from CircuitPython? or how big of a barrier would there be to enabling it? The interest arises now because it seems to be necessary for modern AWS auth. The AWS_IOT library uses an older one that is no longer supported I think. I am looking into using amazon polly TTS service for a project idea from Limor.
Oooh there is an implementation from micropython here: https://github.com/micropython/micropython-lib/blob/master/python-stdlib/hmac/hmac.py I'll see if this can be used under CircuitPython
lone axle
Yeah, following up on this. Between the micropython hmac implementation and adafruit_hashlib I was able to successfully make a request to AWS 🎉 
tiny peak
that might also be useful for generating TOTP codes?
lone axle
There are a few TOTP projects on learn I know: https://learn.adafruit.com/search?q=TOTP but I am unfamiliar with the crypto that underpins it.
tiny peak
can it do hmacsha1 and hmacsha256? IIRC, that should cover what most auth services need.
looks like this one has a bring your own HMAC function in the code: https://learn.adafruit.com/circuitpython-totp-otp-2fa-authy-authenticator-friend/software
lone axle
can it do hmacsha1 and hmacsha256? IIRC, that should cover what most auth servic...
I think it must be doing hmacsha256. This is the code I have https://gist.github.com/FoamyGuy/6375c15b8e5d67753d23017e63023e51 in _sign() it does hmac and passes sha256 function.
tulip sleet
<@329766224093249548> when you have a moment I'm curious about `hmac` from back ...
My version is here: https://github.com/adafruit/CircuitMatter/blob/circuitpython/cm_hmac.py
lone axle
@tiny peak is dac_volume intended to control output from the speaker? In testing it seems like raising and lowering it does not change how loud the audio coming from the small JST speaker is.
tiny peak
speaker chain is dac_volume -> speaker_volume -> speaker_gain, and 1 dB of change to any of those should sound approximately the same as 1 dB of change to another (note that speaker_gain step size is 6 dB though)
What code are you using to test with?
if you're using my test code, the "q" for +1 dB and "z" for -1dB dac_volume adjustments are pretty subtle. You should be able to hear a definite change by the time you move ±6 dB though
Also, when you say "small JST speaker", do you mean the 8Ω 1W mini speaker from the Fruit Jam kit? That's what I calibrated to.
lone axle
What code are you using to test with?
from adafruit_fruitjam import FruitJam
fj = FruitJam()
print(f"dac_vol: ", fj.peripherals.dac.dac_volume)
fj.peripherals.dac.dac_volume = -40
print(f"dac_vol: ", fj.peripherals.dac.dac_volume)
fj.peripherals.dac.headphone_output = False
fj.peripherals.dac.speaker_output = True
print(fj.peripherals.dac.speaker_volume)
# fj.peripherals.dac.speaker_volume = -1
print(fj.peripherals.dac.speaker_volume)
fj.peripherals.play_mp3_file("saves/awspollyoutput.mp3")
But play_mp3_file() is new and not in the released library. It is the mp3 version of play_file() which does wav files.
lone axle
Also, when you say "small JST speaker", do you mean the 8Ω 1W mini speaker from ...
Yes I think it's the same one, though mine is older I think from PyGamer days but not certain.
tiny peak
dac.speaker_output is a magic change-lots-of-stuff-at-once property that will overwrite previous changes to dac_volume
I'd recommend having a go with the test code that I posted in the PR, or at least using the same order of property setting calls
the tester is set up with a continuous tone, then you can use keys above the QWERTY left hand home row to increase various volume / gain settings and keys below the home row to decrease them.
spacebar will toggle the speaker amp power, so you can compare headphone jack noise level with and without the speaker turned on
lone axle
Okay, I commented over on the PR, I'd like to add that as an example in the repo if possible.
tiny peak
Okay, I commented over on the PR, I'd like to add that as an example in the repo...
Sounds goood. I'll add that file to the examples folder and put a brief "Usage Examples" section at the top of the API reference page. That will explain how the easy way to use the API is to start with speaker_output = True or headphone_output = True, then read dac.dac_volume to get your initial volume setting, then provide a user control so people can changedac_volume
tulip sleet
Thanks @lone axle !
lone axle
Thanks everyone. Have a great week 👋
uncut nexus
thanks @lone axle
tiny peak
@lone axle I'm working on example code for you for the TLV320 PR. Ruff absolutely despised my coding style for the existing example code, so I'm having to rework it a bit.
lone axle
<@382939733107408897> I'm working on example code for you for the TLV320 PR. Ruf...
Ok sounds good. Feel free to use noqa comments if it is being overly difficult.
tiny peak
I tried that, but it didn't seem to work. do you have examples of known working noqa comments?
lone axle
I tried that, but it didn't seem to work. do you have examples of known working ...
There is a line that disables one of the rules: https://github.com/adafruit/Adafruit_CircuitPython_Display_Text/blob/7d1f187aac8e899e791324cc78633bf4f32c984b/adafruit_display_text/text_box.py#L267 you can specify different or multiple rules by changing F841 to whatever rule you want it to ignore. Then by custom we put a "human readable" comma seperated list of the disabled rules at the end of the comment as well.
tiny peak
okay, thanks. I'll have another go at that. Might be I put the comment on the wrong line
lone axle
Here is the notes document for the next CircuitPython Weekly Meeting. It is in two weeks on September 8, 2025 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/10SvxnkxPP0HhgSv0_x4WXYV0ySvnUcm6KjmpRXEPlUQ/edit?usp=sharing
tiny peak
There is a line that disables one of the rules: https://github.com/adafruit/Adaf...
that worked. pushed the commit to the PR branch.
tiny peak
fruit jam back in stock 🎉 https://www.adafruit.com/product/6200
manic glacierBOT
manic glacierBOT
slender iron
I'd love help testing https://github.com/adafruit/circuitpython/pull/10584 before I un-draft it
manic glacierBOT
lone axle
Would we want to bring the micropython implementation of hmac into the bundle as adafruit_hmac.py? Or if I am going to use it within a project should I just include hmac.py in with my project files?
tulip sleet
@slender iron fyi I am working on updating pico-sdk and cyw43-driver because we are behind on these (your inlining of the critical section routines is now in pico-sdk)
thought it might fix some RP2 wifi issues but I am not optimistic
slender iron
kk, always good to stay up to date on them. I don't really like debugging when we're behind
thach merged in the c5 and c61 defs so I can switch back to upstream tinyusb too
empty salmon
Status LED Blink Codes
I am sure I asked this before but can't find the prior answer and internet searches failed me.
What are the 'blink codes' for the non-neopixel LED?
I am working on a new M0 board (SAMD21G18).
- I put it in bootloader
- I drag/drop my CircuitPython build
- The upload finishes
- the board boots
- the status LED blinks 5 times
UPDATE: I just tested a similar board from last year - my COIN_M0 - and I get the same behavior.
Tested with 9.2.4, 9.2.8, and 10.0.0.b2
empty salmon
https://learn.adafruit.com/welcome-to-circuitpython/troubleshooting#circuitpytho...
I don't see what "5 blinks" means. (also, I never get the mass storage device)
slender iron
5 blinks at the start is probably the 1 second of safe mode blinking
empty salmon
5 blinks at the start is probably the 1 second of safe mode blinking
OK. Any ideas for how to debug without the usual CircuitPython console (no serial device and no mass storage device)?
empty salmon
The plot thickens. I found my original test from the prior release. I didn't load new code.
I connect to a Windows 11 machine and I see CIRCUITPY. I use the same cable and PCB and connect it to MacOS 15.6.1 and no mass storage device.
When I run the same test with my new PCB version I get the mass storage under Windows and no mass storage with MacOS.
manic glacierBOT
empty salmon
I'd use SWD
More testing points to a MacOS issue (I am running 15.6.1 on Apple Silicon).
I have tested my new board, my old board, and an Itsy Bitsy M0.
They all exhibit the same issues: no mass storage device and no serial device.
Using system_profiler SPUSBDataType does show the board.
tulip sleet
More testing points to a MacOS issue (I am running 15.6.1 on Apple Silicon).
I ...
did you recently upgrade macOS? I have macOS 26 beta on a mini and haven't yet seen trouble
empty salmon
did you recently upgrade macOS? I have macOS 26 beta on a mini and haven't yet s...
Sadly, "yes". Just before starting this project, I updated various software including MacOS 15.6.1
tulip sleet
Sadly, "yes". Just before starting this project, I updated various software incl...
do you have any third-party antivirus or backup products running?
empty salmon
do you have any third-party antivirus or backup products running?
No 3rd party antivirus or backup. I do use Time Machine on the Mac.
tulip sleet
and you did reboot already?
empty salmon
Yes. Once. I can do it again.
tulip sleet
i am at a loss otherwise. I'll upgrade my macOS 26 and see if it's having troubles too. Other mac i have is stuck on Ventura (Intel)
empty salmon
There was a post about a week ago ...
#circuitpython-dev message
... that I just found. It sounds similar (but could be a red herring)
empty salmon
Has anyone on MacOS been seeing CIRCUITPY drives not being mounted (or even visi...
Are your issues on Apply Silicon and are you on 15.6.1?
I am having similar issues of "no mass storage device" and "no serial device" while still being able to locate the device using system_profiler.
devout jolt
Are your issues on Apply Silicon and are you on 15.6.1?
I am having similar iss...
Apple Silicon: Yep; 15.6.1: Yep
tulip sleet
@empty salmon @devout jolt I just upgraded to the latest macOS 26 beta, beta 8. I tried an Itsy M0 running 8.2.10 and then upgraded it to 10.0.0-beta.2. I unplugged it and replugged it several times, with and without ejecting properly, and also then wrote some files to it and repeated unplugging and replugging it. It did not disappear, and I was able to see the BOOT drive to, to update it.
This is a Mac Mini m1
empty salmon
<@401870246157484033> <@352910176736772096> I just upgraded to the latest macOS ...
Thanks. I was occasionally able to see the BOOT drive and never able to see the CIRCUITPY drive.
tulip sleet
This would be the fourth or fifth time Apple has broken something about FAT filesystem access.
You could try doing a backup on some other machine and then doing storage.erase_filesystem() and see if that helps.
empty salmon
This would be the fourth or fifth time Apple has broken something about FAT file...
Yeah. Sadly.
empty salmon
You could try doing a backup on some other machine and then doing `storage.erase...
I am assuming you mean to do the erase on the Itsy Bitsy ?
tulip sleet
yes
empty salmon
yes
OK. I will give that a try on my Windows machine. I will also do some testing from Linux.
tulip sleet
if you and @devout jolt can file bugs via Feedback Assistant, that would help. You could cross-reference each other's
empty salmon
Apple Silicon: Yep; 15.6.1: Yep
Thanks. I have the same configuration and the same problem(s).
empty salmon
if you and <@352910176736772096> can file bugs via Feedback Assistant, that woul...
I can try but it would need to be on a day when someone has slipped me happy pills. I have had so many bad experiences with Apple support mechanisms that I loath even the filing step.
tulip sleet
I looked at the release notes for 15.6 and 15.6.1 and as usual they are useless. I think 15.6.1 was a hurry-up release for that zero-day re image files
empty salmon
I looked at the release notes for 15.6 and 15.6.1 and as usual they are useless....
Those were my conclusions.
devout jolt
if you and <@352910176736772096> can file bugs via Feedback Assistant, that woul...
Yep, I filed a Feedback several days ago on this. @empty salmon Feedback #FB19813868. I discovered when it gets in this state, it also prevents .DMG files from mounting. Hopefully Dan's experience with the beta means this will be addressed in a few months
empty salmon
Yep, I filed a Feedback several days ago on this. <@401870246157484033> Feedback...
Thanks. I included you feedback number in my submission.
empty salmon
As an entry for the record, here is the output from system_profiler which shows my M0 board is recognized as a removable media even though Finder does not show the mass storage device ...
ECCN M0:
Product ID: 0x5687
Vendor ID: 0x1209
Version: 42.01
Serial Number: F159672950304A46462E3120FF143327
Speed: Up to 12 Mb/s
Manufacturer: Bradán Lane STUDIO
Location ID: 0x02100000 / 1
Current Available (mA): 500
Current Required (mA): 100
Extra Operating Current (mA): 0
Media:
Capacity: 8.2 MB (8,192,000 bytes)
Removable Media: Yes
BSD Name: disk4
Logical Unit: 0
Partition Map Type: Unknown
S.M.A.R.T. status: Verified
USB Interface: 2
Important: There is an error in the report. The flash chip on the ECCN M0 is only 4MB. I do not know why it is reporting 8.2MB when in 'bootloader' mode. It does show 4.2MB is 'run' mode.
Coin M0:
Product ID: 0x5687
Vendor ID: 0x1209
Version: 1.00
Speed: Up to 12 Mb/s
Manufacturer: Bradán Lane STUDIO
Location ID: 0x02100000 / 1
Current Available (mA): 500
Current Required (mA): 100
Extra Operating Current (mA): 0
Media:
Coin M0:
Capacity: 4.2 MB (4,190,720 bytes)
Removable Media: Yes
BSD Name: disk4
Logical Unit: 0
Partition Map Type: MBR (Master Boot Record)
S.M.A.R.T. status: Verified
USB Interface: 2
Volumes:
disk4s1:
Capacity: 4.2 MB (4,190,208 bytes)
File System: MS-DOS FAT16
BSD Name: disk4s1
Content: DOS_FAT_16
The 8MB vs 4MB when in bootloader mode may be because UF2 files are 2x the size of the flash they program, so the UF2 bootloader lies to the OS so it will allow the copy to happen
slender iron
@lone axle there is an open method on mp3decoder that takes in a new filename
lone axle
<@382939733107408897> there is an `open` method on mp3decoder that takes in a ne...
Thank you, the fruit jam PR is updated updated to use open() now.
empty salmon
You could try doing a backup on some other machine and then doing `storage.erase...
Ran this test and it did not improve the situation.
empty salmon
I looked at the release notes for 15.6 and 15.6.1 and as usual they are useless....
It's likely unrelated but ...
today, when testing both my COIN_M0 and the Itsy Bitsy M0 running both 9.2.8 and 9.2.1 under Windows 11 (with the most recent major pack?) I was having problems saving code.py - I would throw an error and if the device had disconnected. I would have to unplug / replug to sort it but "save" would cause the problem to occur again. I tested with multiple USB cables and different USB ports.
tulip sleet
It's likely unrelated but ...
today, when testing both my COIN_M0 and the Itsy...
what editor were you using?
empty salmon
what editor were you using?
Thonny. ... I didn't think to try another method 🫣
tulip sleet
Thonny does not write to CIRCUITPY directly. It sends invisible commands to the REPL (file.write(...)) etc.
it's not a good test of CIRCUITPY access
empty salmon
I thought a "file save" from thonny did an actual "file save".
I did not use "run".
tulip sleet
it still uses REPL commands, I think. That way it runs smoothly on non-USB boards that can't support CIRCUITPY USB access.
empty salmon
I will test with a simple file copy then.
tulip sleet
maybe not, though, because CIRCUITPY would be read-only to the REPL, on second thought. I may be making this up
somber garden
IIRC, certain file editors have problems because they write to the filesystem differently.
https://learn.adafruit.com/welcome-to-circuitpython/recommended-editors
I remember Notepad++ had a lot of problems back in the day, but doesn't now.
empty salmon
I know thonny's "run" is done by streaming the Python into the REPL.
somber garden
Interestingly, it recommends Thonny
tulip sleet
I wrote that section 🙂
empty salmon
I have used Thonny extensively but to rule it out, I will just try "copy" from the file system.
somber garden
That's me taught then :)
tulip sleet
no, i appreciate it, and I didn't remember exactly how I recommended Thonny. It was a real problem earlier on with a lot of editors that they didn't flush to disk
somber garden
Back when I was using Notepad++ on Windows I noticed the bad disk flushing, so I
- saved locally
- created a desktop shortcut to a script that copied files to the circuitpython device
- Gave a the desktop shortcut a keyboard shortcut
So all I had to do is save, then do something like "CTRL+ALT+S" or whatever to upload it.
empty salmon
I teach my CircuitPython workshops using thonny. For that, I use an RP2040 based design. It's been great.
I will focus on getting a reproducible flow on Windows and test with a couple different M0 boards.
Theoretically, they should all be stable.
sharp hazel
hi friends what's a good font for small text on those 64x32 matrix leds? the default terminalio.FONT builtin is ok but if there were one only like 6 or 7px tall that was still legible i'm curious
i want to put 4 lines of text on the screen (or one large one on top and two small below) and i'm not sure how to begin with that. i'm novice to python but do have a little bit going so far just not sure how to properly address matrixportal (using the ESP32-S3 based matrixportal fwiw, the docs seem really towards the M4?)
crimson ferry
there are few small fonts around, 5pt and up, not sure where I got them (maybe various learn guides), but if you search you may find them: l_10646, LEDBDREV, MartianMono, NotoSans, PressStart2P, roboto, RobotoMono, tahoma, verdana
sharp hazel
THANK YOU! i appreciate you taking the time to distill a shortlist for me , i will search out!
i tried the roboto, tahoma, verdana 8pt and wasn't happy, didn't try smaller but lemme try the rest on that list and in diff sizes
crimson ferry
there are verdana 5-7, tahoma 5-8, several roboto 5pt and up
ah, here it is... https://forums.adafruit.com/viewtopic.php?t=203655
sharp hazel
ok yeah that's what i found too i just hadnt tried them all yet, and the first few i plucked out didn't seem to be as legible as the default terminalio.FONT builtin
ill play with some more and see... next then is to figure out how to address the screen right and i guess that's labels?
i'd like to have a simple way to address the rows of the screen and "paint" to them
crimson ferry
displayio works the same whether it's M4 or S3
sharp hazel
ok thanks, ill look there. maybe that's what ive been missing.
brand new to this ecosystem
my project is to make a SPL monitor to connect to my Smaart SPL logging system's websockets API... i figure'd there has to be some sort of portable asyncio out there right it's got wifi, could just have it live display in the front of house
crimson ferry
an alternative to splitting the screen into labels is something like this (I haven't tried it with other than builtin terminalio font, but it should? work https://gist.github.com/anecdata/0e426cc9d150c078f7f0b276afa6ff57
there is asyncio, but you're at the mercy of the blocking granularity of each module or library
but it can still be a huge help to get things multitasking
sharp hazel
which is sorta-ok cuz i dont really expect more than an update every second or three, faster refresh would be ideal but my tolerance is huge here
it's just going to be going the whole time at events as a audience display
yeah ust 3x values from a json blob from a websockets endpoint printed to the screen, that's basically it
crimson ferry
we should probably take this over to #help-with-circuitpython , this channel is more for core firmware development
sharp hazel
oh hahaha sorry yes indeed, ill post there - did not realise the distinction
empty salmon
i want to put 4 lines of text on the screen (or one large one on top and two sma...
I created a 5x7 font based on the HP100XL. It's already in the right format for use with CircuitPython. You can find it here and if you browse the parent project, you will find the CircuitPython code for using it
solar whale
Just curious. The Library Bundle has not been updated since August 10. Is that expected? Several libraries have been updatded since then.
lone axle
Just curious. The Library Bundle has not been updated since August 10. Is that...
No, that seems like something is broken. Thanks for mentioning. I will look into it
That was due to Github disabling cron actions automatically when there are no commits in a repo for 60 days. These jobs are inside of the adabot repo currently. There is an open issue to move the reports task to a different repo to avoid the problem. Looks like some more tasks will benefit from that move too.
I re-activated just now with the button so the tasks should run correctly on their next scheduled attempt.
lone axle
@solar whale we ran a bundle update manually so there is a new release now: https://github.com/adafruit/Adafruit_CircuitPython_Bundle/releases/tag/20250829.
Longer term solution for the root cause is in the works.
slender iron
@tulip sleet I think I have a fix for the blanking. wait on beta.3 for me please
(blanking during esp32spi request)
tulip sleet
<@329766224093249548> I think I have a fix for the blanking. wait on beta.3 for ...
great - I am still working on release notes.
solar whale
<@343570430042308609> we ran a bundle update manually so there is a new release ...
Thank you!
empty salmon
I think I missed something with my COIN_M0.
Two questions:
Question 1:
I have Adafruit_CircuitPython_NeoPixel frozen and I have CIRCUITPY_FULL_BUILD = 0.
It seems I need to explicitly add neopixel_write to my COIN_M0 board using CIRCUITPY_NEOPIXEL_WRITE = 1.
Is this correct?
Question 2:
I have not built in a while. I now get `Minimum GCC version 14" but I do not see where I get GCC version 14 for atmel-samd boards.
tulip sleet
= 1?
empty salmon
(sorry, I am so used to Discord on mobile that on desktop, ENTER keeps submitting my post)
tulip sleet
CIRCUITPY_NEOPIXEL_WRITE ?= 1 is in py/circuitpy_mpconfig.mk, so it should be on by default. However, neopixel is not frozen in by default.
empty salmon
I add FROZEN_MPY_DIRS += to add it to my board. after the build, when I use import neopixel I get the error from neopixel.py line 20 with is the neopixel_write.
tulip sleet
if you just do an import neopixel_write in the REPL, does it fail?
tulip sleet
could you upload your mpconfigboard.mk using the + to the left or point me to it?
slender iron
@tulip sleet This is waiting your review: https://github.com/adafruit/circuitpython/pull/10499
empty salmon
https://github.com/bradanlane/circuitpython/blob/bls_coin_m0/ports/atmel-samd/bo...
this is the repository copy which does not have the explicit CIRCUITPY_NEOPIXEL_WRITE = 1
(and sorry for the slowness but me dev env is a mess because I build in a VM on Mac but mac currently can load CP so I have a remote into a Windows machine for uploads) BLECK!
tulip sleet
you can do make print-CIRCUITPY_NEOPIXEL_WRITE BOARD=bradanlanestudio_coin_m0 to see what it thinks the value is
do a make clean before a build if you change the .mk files
tulip sleet
<@329766224093249548> This is waiting your review: <https://github.com/adafruit/...
i felt like I had to test this, but maybe not
empty salmon
you can do `make print-CIRCUITPY_NEOPIXEL_WRITE BOARD=bradanlanestudio_coin_m0` ...
I have done a make clean BOARD= ...
The setting is = 1
tulip sleet
so do a rebuild an then try import neopixel_write
keen field
Is this the right place to ask about my struggles to get two pi pico cowbell can bus hats to communicate with each other? The write buffers just fill up like nothing is being sent out.
tulip sleet
Is this the right place to ask about my struggles to get two pi pico cowbell can...
better to ask in #help-with-circuitpython
this is more for core development discussions
slender iron
Is this the right place to ask about my struggles to get two pi pico cowbell can...
The support forum is a good place too: https://forums.adafruit.com/viewforum.php?f=60
@lone axle want to wait on the cron changes until next week when we can watch it to make sure it is working?
lone axle
Yep sounds good to me.
slender iron
Yep sounds good to me.
I'm not working again until Wednesday but you can do it Tuesday if you like. I'll hold off to approve though so it doesn't get done over the long weekend. Thanks for doing it!
empty salmon
sorry for all the bothering. ... I am feleing pretty dumb at the moment.
I now have the newer GCC for atmel. When I build I get an error OPT_OS_ZEPHYR is not defined. It's coming from .../supervisor/shared/usb/usb.c
tulip sleet
your submodules may be out of sync: Run this:
alias gitsubupdate='git submodule sync --quiet --recursive && git submodule update --init --filter=blob:none'
i mean run what the alias does. You might find the alias useful
tulip sleet
take out the --filter=blob:none. Sounds like your git is somewhat old
empty salmon
FYI: git is 2.34.1
tulip sleet
that is pretty old. See https://learn.adafruit.com/building-circuitpython/build-circuitpython#using-a-version-of-git-that-can-do-partial-clones-3146710 about using a newer git
it will save time on git fetches
have you consider updating to macOS 26 beta or is that too dangerous?
tulip sleet
can you back out 15.6.1? I know Apple makes that hard
tulip sleet
i do remember that from previous attempts
empty salmon
serve me right for accepting the update in the first place
tulip sleet
well, it was an important zero-day fix. were you updating from 15.6 or earlier?
this is all kinda why I use very standard linux for a dev machine
empty salmon
15.6.0 to 15.6.1
I may take an old windows laptop and nuke it for linux
I have a few that are more than good enough for CP development
as a matter of fact, I will add that to my to-do list
tulip sleet
I am going to test the FruitJam on a usable monitor...
slender iron
I suspect #10564 is the same psram issue
tulip sleet
I suspect #10564 is the same psram issue
are you looking at #10564? I will merge the lvfontinfo after I re-run the failing job (which looks like a CI glitch).
slender iron
ya, it is more complicated because PIO can't just wait for more bytes and work the same
slender iron
w00t. yay for bug fixing
tulip sleet
@slender iron ok, I will make a release unless you have anything else.
orchid basinBOT
manic glacierBOT
empty salmon
I build a new dev environment and have successfully build, deployed, and tested the COIN_M0 with 10.0.0-beta.2.
It goes without saying this is usually a rough step for me but I have submitted the PR.
(The usual thanks yet again to @tulip sleet and @slender iron for patiently working with me to get this fix working).
tulip sleet
CircuitPython 10.0.0-beta.3 Released!
https://blog.adafruit.com/2025/08/29/circuitpython-10-0-0-beta-3-released/
Highlights of this release:
- Convert mono audio to stereo when going through a stereo
audiomixer.Mixer. - Add
audiomixer.MixerVoice.panning, withsynthio.BlockInputsupport. - Add
synthio.MidiTrack.tempo. - Restore missing
audioiomodule on ESP32 and ESP32-S2. - RP2xxx DVI now works on more monitors and has fewer glitches.
thorny jay
I created a 5x7 font based on the HP100XL. It's already in the right format for ...
Thank you for the fonts.
PS: Maybe you should switch to settings.toml
manic glacierBOT
manic glacierBOT
frail widget
Hey team, I've been working through some type issues with 2 minor PRs ^ and a question about BlockDevices over on CircuirPython_Typing. https://github.com/adafruit/Adafruit_CircuitPython_Typing/issues/51
tulip sleet
Hey team, I've been working through some type issues with 2 minor PRs ^ and a qu...
I think the BlockDevice protocol description was swiped from MicroPython, and includes things that are not used in core CircuitPython and in libraries like Adafruit_CircuitPython_sd. For instance, ioctl() talks about littlefs, which we don't use at all. I think you can remove the ioctl method. As for start_block vs block_num, they mean the same thing but start_block is more descriptive. For ReadableBuffer / WriteableBuffer vs bytearray, the ...Buffer is more precise.
frail widget
I think the BlockDevice protocol description was swiped from MicroPython, and in...
Thank you for this. I have put up a proposed resolution in https://github.com/adafruit/Adafruit_CircuitPython_Typing/pull/52
tiny peak
Is there some magic to enabling CIRCUITPY_DEBUG_TINYUSB ? I tried uncommenting the #define lines in ports/raspberrypi/boards/adafruit_fruit_jam/mpconfigboard.h for CIRCUITPY_CONSOLE_UART_TX, CIRCUITPY_CONSOLE_UART_RX, and CIRCUITPY_DEBUG_TINYUSB. Tried it with the default ..._TINYUSB 0 and also with ..._TINYUSB 1. In both cases, I can get a copy of the normal console output on the A4 (GPIO44) pin with 115200 serial into a rpi debug probe. But, there's no sign of TinyUSB debug stuff.
tulip sleet
Is there some magic to enabling `CIRCUITPY_DEBUG_TINYUSB` ? I tried uncommenting...
i'd look at what that value really controls. There are further logging options in tusb_config.h, I think.
tiny peak
i'd look at what that value really controls. There are further logging options i...
Thanks. Figured it out. The magic is that you have to do #define CIRCUITPY_DEBUG_TINYUSB 2 or perhaps #define CIRCUITPY_DEBUG_TINYUSB 3. The number you pick there selects the log level.
hmm... as I suspected, the USB glitch stuff is a Heisenbug. Turning on debugging appears to add enough delay that the glitch goes away.
tulip sleet
My last long bug investigation was also a Heisenberg 😀
tiny peak
supposedly tinyusb debug over RTT is possible. might try that.
supposedly RTT is really fast
tiny peak
Ended up using this procedure: https://adafruit-playground.com/u/SamBlenny/pages/circuitpython-core-dev-debug-tricks#tinyusb-debug-log-3206041
full plume
Not exactly sure why, but looking at that made me slightly nostalgic for the old days where my "debug prints" went straight to text video memory, back when a "multiple monitors" meant you stuffed a CGA and a Hercules Mono card into the same case. Which I guess might be ever so tenuously relevant to debugging code on a FruitJam
tiny peak
From what I read of RTT, I think that's sorta how it works? Like, it just stuffs entries into some big circular buffer in RAM, where some hardware peripheral hooked up to the debug circuitry can get at them?
full plume
hmm... as I suspected, the USB glitch stuff is a Heisenbug. Turning on debugging...
that's why I just write code without any bugs in it, so I don't need a debugger. Of course I'm mostly kidding, but not all kidding. When I got my first computer, it took a while (another six months or so?) before I could save up enough to by any external storage - so whatever I wrote was gone as soon as the pwer went off. And if my code got stuck end an endless loop or poked something it shouldn't have and froze things, oh well... I think this is partly responsible for my tendency to write in a more defensive style which does a lot of internal "sanity checking" as things move along. Those habits have proven particularly useful for embedded programming - I do make mistakes, but the surrounding code usually catches and identifies things fairly well.
tiny peak
yeah... there are times and places where it really pays off to think pretty hard about what your code is gonna do before you turn it loose to do those things.
full plume
From what I read of RTT, I think that's sorta how it works? Like, it just stuffs...
OK, I want one.... what other CircuitPython options supports RTT? (besides the FruitJam)
full plume
yeah... there are times and places where it really pays off to think pretty hard...
trying to figure out how to say this, but in a way I think more about what my code is not going to do, or at least what it's not intended to do yet, and actually set traps to make sure it doesn't go there. Kinda like (and including) typical preconditions / postconditions / asserts, but more... ???
tiny peak
People also spend a lot of time thinking about such things in the realm of Ada / SPARK language
tiny peak
and cars. I think some of that grew out of investigations in to stuck Toyota accelerators or something like that? Might be totally mis-remembering the details though.
But, yeah, generally for times and places where stuff actually needs to work
full plume
Personally, for anything beyond moderate complexity I'd just say don't try to write it in C
crimson ferry
wait, you mean my CircuitPython microcontroller hobby project isn't MISSION CRITICAL!?? 😉
full plume
C++ can be a much better environment for building high performance, high reliability code. Of course in can also lead to unmitigated disaster faster than you can say printf( "hello world" ); if you try and use all the options or constantly chase the shiny new things
full plume
wait, you mean my CircuitPython microcontroller hobby project isn't MISSION CRIT...
maybe not your project. I don't write things any other way...
tiny peak
If you've never experimented with trying to write some stuff in Ada / SPARK, it might blow your mind. Forces you to consider all sorts of edge cases and specify what behavior you intend.
Rust is sorta that way too, but it aims in a bit of a different direction.
Thing about C++ is it can be pretty much anything and everything. Some of those things being safe, and others very much not
Trouble with all of it is that ultimately you have to link against C libraries if you want to get stuff done in a reasonable amount of time.
full plume
Actually I thought twice before saying that. I've thought about placing (and most likely some time soon will) a VERY prominent warning section and some extremely blatant clauses in whatever "license" I end up with for my current project stating that this is NOT a "safe" system, and you are strictly FORBIDDEN from using it in ANY project which is channeling enough power or force (i.e. motors, etc.) to cause injury.
I'm probably being paranoid, but I know a bit about building high risk systems and I also know that my current project isn't exactly playing by the rules.
tiny peak
Good luck with figuring out how to handle that. Always tricky to decide how much time and effort to put in and when to call it good and ship the thing.
full plume
Good luck with figuring out how to handle that. Always tricky to decide how much...
I highly doubt anyone would even think of using it for anything like that anyway. Well, except for some guy building a life size 40K Space Marine Cosplay and wanting power assist for leg motion....
tiny peak
mmm... yeah, you never know what people will try.
full plume
But I really, really want to, just need to find some volunteers. Which is a bit challenging, considering that anyone who participates in discussions on most of the places I frequent (certainly anyone already lurking - much lest posting - in circuitpython channels) is already outside of the primary "non-programmer" target audience for my project. Still, I could always throw out some YouTube videos. (sadly, that's not a joke...)
crimson ferry
We seem to have fewer sockets available in CP 10 (espressif 10.0.0-beta.3)? In CP 9, I could use 8 TCP sockets. in CP 10, only 4```py
import time, wifi, socketpool
pool = socketpool.SocketPool(wifi.radio)
socks = []
while True:
socks.append(pool.socket(pool.AF_INET, pool.SOCK_STREAM))
print(f"{len(socks)}") # 8 in CP 9; 4 in CP 10
so I don't think you could have for example, an HTTPS client and an HTTPS server with a connected client (this would not be an unusual use case)
crimson ferry
CONFIG_LWIP_MAX_SOCKETS=8 is still 8. The limit of 4 is reached at socket creation time, presumably before the CONFIG_LWIP_MAX_ACTIVE_TCP=4 and CONFIG_LWIP_MAX_LISTENING_TCP=4 limits kick in. (Those haven't changed in 2 years either, but probably at least `_MAX_ACTIVE_TCP should?)
full plume
We seem to have fewer sockets available in CP 10 (espressif 10.0.0-beta.3)? In C...
Wow, that's unexpectedly tight. Even 8 seems restrictive.
crimson ferry
yes, but many competing demands for esp-idf memory
full plume
yes, but many competing demands for esp-idf memory
sure, but that can usually be better handled by allocating on the heap with a smaller default but allowing more on demand
crimson ferry
(I raised it to 8 from 4 in a PR years ago)
some esp-idf options have to be configured in at build time, like how many stations can connect to a wifi AP, and probably other wifi and ble things
full plume
looks like the main affected bit would be static uint8_t socket_fd_state[CONFIG_LWIP_MAX_SOCKETS]; and static socketpool_socket_obj_t *user_socket[CONFIG_LWIP_MAX_SOCKETS]; in socketpool/Sockets.c
crimson ferry
but the config constant hasn't changed, so I'm at a loss why only 4 STREAM sockets can be created now
full plume
that's in "our" ports/espressif/common-hal, not an ESP-IDF config
crimson ferry
right, something could have changed under the covers in esp-idf
full plume
but the config constant hasn't changed, so I'm at a loss why only 4 STREAM socke...
that's kinda weird...
crimson ferry
or some new default or config option we didn't tweak yet
I had to upgrade to CP 10 to fix that socket hard fault, but lost sockets in the process
full plume
right, something could have changed under the covers in esp-idf
I suppose that's possible - a "new" option seems plausible but if so it's really hard for me to imagine them setting the default that low. I can see it being unusual for CircuitPython apps to bump up against an 8 socket limit, but it would cause far more noticeable problems for C/C++ apps using ESP-IDF
crimson ferry
Espressif defaults are quite a bit higher... 10 for LWIP_MAX_SOCKETS and 16 for the other two params above (which don't use memory if not used)
full plume
I can poke around a bit - the httpServer listen loop needs to come from that list, as would each active Websocket (although you're only supposed to have one at a time). Shouldn't be too hard to throw enough (near) simultaneous requests at it to see where it breaks on my system.
crimson ferry
I think all it would take is 1 HTTPS client, then 1 HTTPS server which then tries to either open a websocket or accept a client connection
4 is painful, 8 is livable
full plume
I'm currently running a 10.0 beta-ish fork build for a Lolin S2 Mini. I can switch to 10.0.0-beta.3 (hopefully it won't break my code) - what board/build are you seeing these problems on?
crimson ferry
S3, but any espressif board should have the same behavior
full plume
10.0.0-beta.3?
crimson ferry
yes
(QT Py S3 N4R2, but I doubt flash config or PSRAM have anything to do with it)
full plume
OK, running with 10.0.0-beta.3 and it so far everything seems to be working (my fork provides support for __all__, so some from XYZ import *s might have gotten a little wonky with out it)
crimson ferry
nothing is jumping out at me from the esp-idf release notes
if you run my code above, do you get 4?
full plume
Need to hack togeher a load test
realizing it's not quite that simple... I was going to hit it with multiple concurrent REST requests from Postman and my "proxy" request base wrappers in a Jupyter notebook, but I think the current (non-async) adafruit_httpserver.Server will just "serialize" those
SO guess it's time to finally break the rules and see what happens when I try to open a second Websocket without closing the first one..
crimson ferry
or do an HTTP[S] request from within the server route
we should have taken this to a thread o_O
gotta step away for a while
full plume
Never too late for fancy threads, we can spin one up when you're ready
full plume
so.... multiple active Websocket connections is - at least for my current use case - working
full plume
we should have taken this to a thread o_
crimson ferry
Is there a way to easily revert a [library] PR? Looks like there is a use case on the VEML7700 for using on-chip interrupt functions, including on-chip smoothing, by polling a register (there is no interrupt pin). https://github.com/adafruit/Adafruit_CircuitPython_VEML7700/pull/31#issuecomment-2963463113
lone axle
Is there a way to easily revert a [library] PR? Looks like there is a use case o...
I'm not aware of an easy button it the GitHub interface anywhere. But it should be fairly easy (depending on comfort with git) checkout the last commit before the PR and then submit that as a new PR that would undo whatever came from the PR commits
I can do that tomorrow if you don't get to it, or can't get git to cooperate.
tulip sleet
I'm not aware of an easy button it the GitHub interface anywhere. But it should ...
there is a revert on the PR page:
lone axle
Ooh, nice. Thank you!
It may need to have conflicts resolved in this case. It outputs this error
crimson ferry
OK, thanks, I'll do a new PR and back out the changes (re-insert the deletions)
thorny jay
Thank you very much. The artifact works for me Adafruit CircuitPython 10.0.0-beta.3-3-g50922c9a74 on 2025-09-03, but something when wrong in the git action: Build CI / windows (not sure what it is, but it was in error and does not seems related to your change).
tulip sleet
Thank you very much. The artifact works for me `Adafruit CircuitPython 10.0.0-be...
Thanks for testing!
jolly roost
Hello, we've created a circuitpython library (https://github.com/sensebox/CircuitPython_VL53LxCX) for VL53L5CX and VL53L8CX ToF sensors based on this: https://github.com/mp-extras/vl53l5cx We are able to bundle the library and also release it but these types of sensor need a firmware file which will be transfered to the sensor on startup. Is there any good way how we could integrate the firmware file to the CircuitPython Package release?
spare jacinth
Hello, we've created a circuitpython library (https://github.com/sensebox/Circui...
havent read the code too closely, but couldn't you simply convert each of the .bin files to a Python bytes/bytearray object (perhaps each one still on its own file), and import them at runtime, instead of reading a file? effectively the same, but likely simpler to configure the packaging of those files
would need some refactor of the ConfigDataFile class, but shouldn't be too hard
lone axle
Hello, we've created a circuitpython library (https://github.com/sensebox/Circui...
I believe it's possible to bundle .bin files with CircuitPython libraries already. I'm not sure if there is anything "special" needed beyond just including them in the repo though. There is an examplere here: https://github.com/adafruit/Adafruit_CircuitPython_PyCamera/tree/main/adafruit_pycamera where this is done. You could check the code and actions in that repo to see how the bundling work and if it requires anything additional.
spare jacinth
havent read the code too closely, but couldn't you simply convert each of the `....
as in...
class ConfigDataFile:
def __init__(self, blob):
self.blob = blob
def _read_offset_data(self, offset, size):
return self.blob[offset:offset + size]
# not sure how fw_data would be adapted
# in VL53LxCX.__init
if self._id == VL53L5CX_ID:
from vl53l5cx import fw
else:
from vl53l8cx import fw
self.config_data = ConfigDataFile(fw)
@lone axle I can merge https://github.com/adafruit/adabot/pull/403 and https://github.com/adafruit/circuitpython/pull/10596, but should I do it in a particular order? These are your cron workflow migrations.
lone axle
<@382939733107408897> I can merge <https://github.com/adafruit/adabot/pull/403>...
I think the order they are merged shouldn't matter so long as they are merged on the same day. It's a cron task so it will trigger only at the time specified, not based on merging if I understand correctly.
tulip sleet
great, will do it now
lone axle
thank you
spare jacinth
I believe it's possible to bundle .bin files with CircuitPython libraries alread...
i thought setuptools ignored non-py files, but that's apparently wrong based on this link 🤣
seems like it packages the whole folder's content, which should work just fine for the usecase above :)
lone axle
I believe it may not have done that before and was only added when the PyCamera library was in the works.
lone axle
(our bundling tools at least)
spare jacinth
i thought setuptools ignored non-py files, but that's apparently wrong based on ...
i'd still go with the "convert to python" approach, so that i can simply import things, relying on relative paths can always become a footgun. but it is more cumbersome
spare jacinth
Hello, we've created a circuitpython library (https://github.com/sensebox/Circui...
btw, i believe the hardcoded 0x15000 on line 185 was intended to be _FW_SIZE
jolly roost
okay cool! Thanks for pointing to the PyCamera Library, I was already looking for other libraries who might have the same problem but didn't found it.
lone axle
okay cool! Thanks for pointing to the PyCamera Library, I was already looking fo...
One handy trick to know for finding stuff in CircuitPython libraries is using git submodule foreach in the bundle repo https://github.com/adafruit/Adafruit_CircuitPython_Bundle All of the libraries are submodules so this allows you to use CLI tools like find, git grep etc to search all libraries at once. e.g. this command in the bundle will search for .bin files:
git submodule foreach "find . -name '*.bin'||true"
Though it does also spam "Entering ___" for each repo which makes the output harder to find without searching. So tacking on an extra grep to filter those out can be helpful too
git submodule foreach "find . -name '*.bin'||true" | grep -v Entering
On linux. Sorry, I'm unsure of other platform commands but something similar should be possible.
lapis fjord
Hello, Working with a BMP581 on CircutPython9.2.8.
The adafruit_BMP5xx library doesnt seem to be in the 9.x library bundle.
So I assumed that I could pull it from github, and compile the .mpy (never done that, docs maybe a little crusty, but mpy-cross generated a .mpy:
(base) mperino@Aku:~/CircutPython/BMP5x/Adafruit_CircuitPython_BMP5xx$ mv /home/mperino/Downloads/mpy-cross-linux-amd64-9.2.8.static /home/mperino/bin/mpy-cross-9.2.8
(base) mperino@Aku:~/CircutPython/BMP5x/Adafruit_CircuitPython_BMP5xx$ chmod 755 /home/mperino/bin/mpy-cross-9.2.8
(base) mperino@Aku:~/CircutPython/BMP5x/Adafruit_CircuitPython_BMP5xx$ mpy-cross-9.2.8 ./adafruit_bmp5xx.py
(base) mperino@Aku:~/CircutPython/BMP5x/Adafruit_CircuitPython_BMP5xx$ ls
adafruit_bmp5xx.mpy adafruit_bmp5xx.py CODE_OF_CONDUCT.md docs examples LICENSE LICENSES optional_requirements.txt pyproject.toml README.rst README.rst.license requirements.txt ruff.toml
however when I copy over adafruit_bmp5xx.mpy and try an import (Adafruit CircuitPython 9.2.8 on 2025-05-28; Raspberry Pi Pico 2 W with rp2350a) :
from adafruit_bmp5xx import BMP5XX
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "./adafruit_bmp5xx.py", line 37, in <module>
ImportError: no module named 'adafruit_register'
lone axle
Hello, Working with a BMP581 on CircutPython9.2.8.
The adafruit_BMP5xx library d...
You need to also install the adafruit_register library. You can find it in the library bundle or use circup.
There are also pre-compiled .mpy files on the github release pages: https://github.com/adafruit/Adafruit_CircuitPython_BMP5xx/releases/tag/1.0.0 though in this case I only just made the first release a few hours ago so this library is still brand new.
lapis fjord
I copied yours over, 4byte difference, added adafruit_register, and it's at least importing:
import adafruit_bmp5xx
I opened an issue under: https://github.com/adafruit/Adafruit_CircuitPython_BMP5xx/issues
LMK if I should close it.
lone axle
my best guess is maybe the difference is based on the version being filled in our not. But I am not sure, perhaps there are specific versions of mpy-cross used that differ slightly.
lone axle
I opened an issue under: https://github.com/adafruit/Adafruit_CircuitPython_BMP5...
I'll respond over there and close it. That way it can potentially be helpful if anyone else happens upon it before it gets added to the bundle.
lapis fjord
Also looks like there is a change which dropped the generic BMP5XX and is using the more specific and helpful BMP5XX_I2C
from adafruit_bmp5xx import BMP5XX_I2C (good)
from adafruit_bmp5xx import BMP5XX (depricated i asssume)
lone axle
Also looks like there is a change which dropped the generic BMP5XX and is using ...
correct BMP5XX_I2C is the class name moving forward. If we end up adding SPI support there would be a BMP5XX_SPI that is used for that. Possibly at that time BMP5XX class might return as a super class, but that would be an implemenation detail that doesn't change the API. user code would still use BMP5XX_I2C or BMP5XX_SPI
lapis fjord
Greatly prefer the new class makes more sense. Thank you for the speedy support!
tulip sleet
One handy trick to know for finding stuff in CircuitPython libraries is using `g...
I do this kind of thing with find . -name '*.bin' at the top level of the bundle repo. I don't bother with the git submodule foreach, since the submodules are just directories anyway.
lone axle
Ahhhh yeah that is way easier. I got honed in on submodule foreach using it with git grep. But now that you mention it I think that could work the same way with grep and without the submodule loop.
Thank you!
tulip sleet
i know you are using submodule foreach All The Time 🙂 to do patches or whatever. Makes sense you would continue the idiom
lone axle
@tulip sleet I'm going to push button to run the update bundles workflow manually. Having a new release will make it possible to get correct versions of the files screenshot picures used in the guides. I figure it will also be nice to know now if there is any reason it can't complete from it's new home.
full plume
havent read the code too closely, but couldn't you simply convert each of the `....
Just a few extra things to consider - The b"initializer" in the python source could take up more space than the .bin file contents. However, if you distribute it in .mpy format, that overhead might be removed. And there's also the per-file "file-system", overhead so if an actual .bin file needs to exist somewhere on the CIRCUITPY drive, that will likely add overhead.
spare jacinth
Just a few extra things to consider - The b"initializer" in the python source c...
I would expect the 2 solutions to be equivalent in terms of speed and resources usage... But the main point of my suggestion (being easier to pack Python files than "other" files) seems to be wrong, at which point it would be best to just use the existing bin files
wraith crow
I've been playing at getting CircuitPython working with all the peripherals of the Olimex RP2350pc but like the SolderParty RP2xxx Stamp Carrier XL the USB A ports/hub has been wired in parallel to the dedicated USB_DM/USB_DP RP2350 pins. I've mostly got a version of CircuitPython working that respects the CIRCUITPY_USB_DEVICE=0 parameter which allows the device to use the attached USB A ports for USB Host functions (keyboards, mice) at the expense of having to use something like an FTDI serial cable and something like Thonny to transfer files. While I would like to see boards like this supported on Circuitpython.org I know not having a CIRCUITPY drive has been a hard stop for boards in the past. Given that, I'm thinking if I make this firmware available on github for public use, my understanding was that Adafruit would prefer I didn't call it "CircuitPython". I was thinking of simply referring to it as something like RP2350pc Python but my question is, is there an obvious place in the core where the CircuitPython branding can be adjusted?
manic glacierBOT
lone axle
All of the cron tasks seem to have run correctly from in the core repo
lone axle
In 10.x it appears that os.uname().sysname for some devices changed. The Matrix Portal M4 is one I konw of for sure that changed from samd51 to samd51j19. Getting more specific about the chip type it looks like.
Is that change intentional and likely to stay? If so I found some libraries that use that mechanism that will need to get updated with the new version.
although I guess the library should account for it for some period of time even if it was untentional and going to change back since there are beta releases out with that value.
tulip sleet
In 10.x it appears that `os.uname().sysname` for some devices changed. The Matri...
This was https://github.com/adafruit/circuitpython/pull/10145, I'm pretty sure. It was meant to make the names more consistent.
fleet hollow
Does anyone know why most code using the TLV320DAC3100 sets the sample rate to 11030 rather than 11025? Is that a quirk of the dac? I've used 11025 in some projects and that seems to work fine.
tulip sleet
Does anyone know why most code using the TLV320DAC3100 sets the sample rate to 1...
I do not know. @lone axle e.g. this line https://github.com/adafruit/Adafruit_CircuitPython_FruitJam/blame/e67239b37b9c82fd3d2557cd3bec3ba5e812e8f7/adafruit_fruitjam/peripherals.py#L160 uses 11030, which is in your initial commit
manic glacierBOT
lone axle
I do not know. <@382939733107408897> e.g. this line <https://github.com/adafruit...
@fleet hollow it is the sample rate used by the boot animation audio file. I used that file a lot for early audio testing it's likely I copied it around to a few places.
tulip sleet
maybe that 11.03 is just a rounded 11.025
fleet hollow
maybe that 11.03 is just a rounded 11.025
Exactly what I was thinking
lone axle
I don't really know much about sample rates though, perhaps 11.03 is rounded? If 11025 is more of a standard that is probably what it should be.
fleet hollow
I don't really know much about sample rates though, perhaps 11.03 is rounded? If...
Could you open the file in Audacity and check the sample rate there? 11025 is the standard. 11025x2 = 22050x2 = 44100, standard cd audio
lone axle
yeah, looks like it is 11025
fleet hollow
I just wasn't sure if there was a strange quirk with that DAC that make it incapable of reproducing 11025, but my guess is that isn't the case.
I'm working on some other updates to the FruitJam library. I'll add that fix in there.
lone axle
as far as I am aware no. I just chose that value based on the properties I think.
thank you
tulip sleet
did you see 11030 anywhere else, in a guide example, etc.?
I only see it in that library in the bundle. looking in the fruit jam os...
nope not in fruit jam OS (there is a BASIC line number with that value 🙂 )
I don't see 11030 in learn.adafruit.com anywhere, based on a google site search
so maybe it's just that line
fleet hollow
I think it was originally in Fruit Jam OS and got moved into the library. So, we're good! It's fixed in this PR: https://github.com/adafruit/Adafruit_CircuitPython_FruitJam/pull/20
lone axle
I've just noticed that directories without python code in them seem to take precedent over python code files. i.e. I have:
CIRCUITPY/
boot_animation/
a_spritesheet.bmp
...
boot_animation.py
When I do import boot_animation in the terminal it seems to be trying to import the folder and doesn't execute the code inside of boot_animation.py
In the short term I plan to change the name of the assets folder to work around this. But I'm also curious if it is somehow to possible to make import use the python file instead of the directory that doesn't contain code?
hmm, maybe there is something else wrong in my situation. Having trouble re-creating that now.
devout jolt
I've been playing at getting CircuitPython working with all the peripherals of t...
There are many CircuitPython boards that don't have a CIRCUITPY drive now, mostly ESP32. I've been itching to try romkey's new circremote that uses a pure-serial workflow like Thonny, mpremote, or code.circuitpython.org
willow totem
I've just noticed that directories without python code in them seem to take prec...
Tried import .filename? [Module name so No .py]
lone axle
Tried `import .filename`? [Module name so No .py]
looks like CircuitPython doesn't support that syntax
>>> import .fake_lib
Traceback (most recent call last):
File "<stdin>", line 1
SyntaxError: invalid syntax
Ah maybe that was it, I must have had to clear that up for sen5x base driver or something
wraith crow
There are many CircuitPython boards that don't have a CIRCUITPY drive now, mostl...
Yes, that has given me some hope and I do plan on submitting the board once I work out the last of the issues but as far as I know the boards that don't have CIRCUITPY so far have all had the WiFi workflow available to them so It's possible Adafruit won't want to make this leap yet 😁
tulip sleet
I've just noticed that directories without python code in them seem to take prec...
that's import logic. you can import a module from a directory or a single file. It looks for a directory first. In general, you shouldn't have a directory with the same name as a .py file. I'll bet CPython does the same thing
lone axle
that's `import` logic. you can import a module from a directory or a single file...
it doesn't, I tried it.
under CPython it will execute the code in the .py file even if there is a folder by the same name with bitmap file in it (same conditions tested on CircuitPython)
Under 3.12 at least.
tulip sleet
(.py) halbert@cod:~$ mkdir foo
(.py) halbert@cod:~$ cat >foo.py
print("foo.py")
(.py) halbert@cod:~$ cat >foo/__init__.py
print("foo/__init__.py")
(.py) halbert@cod:~$ py
Python 3.12.3 (main, Aug 14 2025, 17:47:21) [GCC 13.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
> import foo
foo/__init__.py
>>>
CPython looks for an __init__.py file and if present, will import the directory module
so MicroPython/CircuitPython probably don't bother to look for __init__.py
stuck elbow
CPython looks for an `__init__.py` file and if present, will import the director...
that was with python2, python3 doesn't need __init__.py anymore
though you can have it if you need code at the top level of the module
tulip sleet
i was just trying to figure out how to indicate that the dir was being used first
tiny peak
@wraith crow In your github comment earlier, you mentioned wanting to discuss USB host stuff here. Sounds good to me. Let's start a thread for it.
tulip sleet
though you can have it if you need code at the top level of the module
but if I rename __init__.py to blah.py, then it indeed imports foo.py. So we could do better.
lone axle
it may not be worth the overhead of additional logic to fix. I was surprised by the behavior at first and couldn't figure out why the script wasn't running. TBH I thought I had tested it this way before, but I guess I have not. In any case it turned out to only be the first rung of the ladder down the rabbit hole I started on 🙂
lone axle
Now that I know it, I will try to avoid naming directories the same name as python files. I opened an issue so it's tracked somewhere but totally fine to close if it's not a priority to change the behavior.
slender iron
I've been playing at getting CircuitPython working with all the peripherals of t...
Not having CIRCUITPY isn't an issue anymore.
Your product supports at least one standard "Workflow" for serial and file access:
With a user accessible USB plug which appears as a CIRCUITPY drive when plugged in. With file and serial access over Bluetooth Low Energy using the BLE Workflow. With file access over WiFi using the WiFi Workflow with serial access over USB and/or WebSocket.Boards that do not support the USB Workflow should be clearly marked.
wraith crow
Cool 😁
slender iron
We should probably call out the serial workflow their explicitly. I'd want it to work with code.circuitpython.org
wraith crow
I'll make sure testing code.circuitpython.org is on my todo checklist....
drifting token
I'm trying to get a espressif esp32-c3-lyra board working, and I've been stuck for a couple of days on getting the status led neopixel working. Is there anyone around who might be able to offer some advice?
fiery pecan
Has anyone tried using https://github.com/czietz/etm-trace-rp2350 ? I think it might be a less intrusive alternative to https://github.com/adafruit/circuitpython/issues/10563#issuecomment-3240829108.
tiny peak
Has anyone tried using https://github.com/czietz/etm-trace-rp2350 ? I think it m...
woah... that sounds interesting
lone axle
How does SPIDevice end up getting write(), readinto() functions that are callable from python code? They aren't listed in https://github.com/adafruit/circuitpython/blob/main/shared-bindings/adafruit_bus_device/spi_device/SPIDevice.c, but the example code calls them. I assume it is something to do with the enter function, but don't quite understand what it is doing to "tack on" those functions to make them callable. I see the python implementation has a similar thing going on: https://github.com/adafruit/Adafruit_CircuitPython_BusDevice/blob/main/adafruit_bus_device/spi_device.py
jaunty juniper
How does SPIDevice end up getting `write()`, `readinto()` functions that are cal...
I believe calling with on a SPIDevice returns the busio.SPI instance
(handling the locking and unlocking via the context manager)
lone axle
Ahhh. Thank you
crimson ferry
What's the simplest way to generate a lot of interrupts into the circuitpython core (raspberrypi)?
full plume
What's the simplest way to generate a lot of interrupts into the circuitpython c...
I'm pretty sure the GPIO inputs on the Raspberry Pi are interrupt capable. You should be able to set up an interrupt handler triggered by state change a digital input, then drive it with a square wave generator to get all the interrupts you could want.
Unfortunately I'm not finding any simple API examples to actually do this - if you're running Linux/Raspbian on the Pi then the kernel tends to handle interrupts, and examples for direct hardware access without a linux kernel seem thin.
crimson ferry
I'm not wanting to change the core code
just have a lot of background interrupts happening behind the CircuitPython code
full plume
Even the arduino bits I've found mention that it's only "faking" it on Raspberry Pis
crimson ferry
I can set up various bus protocols or digital streams like PulseIn or countio, but hoping there was an easier way. Tried watchdog, but RAISE isn't implemented on raspberrypi
full plume
just have a lot of background interrupts happening behind the CircuitPython code
Just to be clear, you're running CircuitPython directly on Raspberry Pi hardware, not Blinka in Linux/Raspbian or on a Pi Pico, right?
full plume
Ah, well that's different ....
crimson ferry
the port naming is confusing (Big RPis are broadcom port)
stuck elbow
any particular kind of interrupts?
crimson ferry
anything really, I just want to stress test some CircuitPython code by having interrupts happening in the background, was hoping for something with minimal setup
stuck elbow
I think pio can trigger interrupts
crimson ferry
that sounds like a half day research job 😉 I'll try some countio or something
stuck elbow
I think that uses pio on the raspberry pi
crimson ferry
some things. like keypad poll in the core, so that's no use
full plume
I think pio can trigger interrupts
almost certainly at the hardware level, I think question is what you can do from CircuitPython to "exercise" something which will cause lots of interrupts
crimson ferry
yes
stuck elbow
create an spi peripheral and toggle the cs pin very fast?
crimson ferry
or maybe I can PulseOut to a PulseIn
full plume
it's plausible - maybe probable - that the wifi stack on a Pi Pico uses interrupts, so websocket torture tests might help
stuck elbow
thing is, if your interrupt-generating code is in Python, then it will be self-regulating, because a lot of interrupts will slow down python vm execution and result in generating fewer interrupts
crimson ferry
I'm actually looking for non-wifi non-socket interrupts to stress test wifi/socket code 🙂
stuck elbow
so you probably want to toggle that pin with another board
full plume
(well, a Pi Pico W anyhow...)
stuck elbow
there is also a way to generate a lot of usb interrupts, I think, but I'm fuzzy on the details
crimson ferry
I'm fuzzier
fiery pecan
thing is, if your interrupt-generating code is in Python, then it will be self-r...
Can GDB be used to mitigate that? I was able to get GDB debugging working with my Adafruit Fruit Jam and Raspberry Pi Debug Probe. It involved compiling CircuitPython from source with the DEBUG=1 argument passed to make, and using the firmware.elf file with GDB in addition to the firmware.uf2 file for flashing.
full plume
there is also a way to generate a lot of usb interrupts, I think, but I'm fuzzy ...
Wouldn't be surprised if just doing a bunch of console over USB IO requires interrupts - although it could plausibly avoid the need for lots of interrupts with a bit of hardware buffering, DMA, etc... so it would be implementation dependent
stuck elbow
Wouldn't be surprised if just doing a bunch of console over USB IO requires int...
no, but usb is polled by the host, and I think there is an interrupt every time the host does the poll, so you could probably poll very often?
crimson ferry
thanks, I've got a few things to try now, and will verify before I get too far that the feature actually uses interrupts and not polling in the core
stuck elbow
with somethinng like wireshark
@crimson ferry by the way, the wifi module is connected over spi on the pico w, so any "wifi" interrupts would be either spi or pin change interrupts anyways
crimson ferry
on espressif I could use Monitor mode
full plume
thing is, if your interrupt-generating code is in Python, then it will be self-r...
and if the interrupt code ever does any GC managed allocations, it will be extra self-regulating 🤯
stuck elbow
and if the interrupt code ever does any GC managed allocations, it will be _extr...
you can't do gc inside interrupts
ah, sorry, I misunderstood
sure
crimson ferry
this arose because LWIP allocates within an isr, but it's protected now with a new pr
full plume
this arose because LWIP allocates within an isr, but it's protected now with a n...
C level allocations can - with great care, sometimes be used safely within an interrupt. But even then it's seriously sketchy, and any simple malloc/free calls will usually (and I only say that to be complete - I have no counter examples...) blow things up.
stuck elbow
C level allocations can - with _great_ care, _sometimes_ be used safely within a...
it's a bit different in micropython/circuitpython, because even the C-level allocations are managed, malloc is replaced
there is a number of specialized malloc-like functions you can use in special cases
tiny peak
I'm actually looking for non-wifi non-socket interrupts to stress test wifi/sock...
Are you able to use picodvi on your board? I think it uses interrupts to set up dma. Rate of interrupts might increase at higher video resolutions?
It does look like countio is actually using an interrupt rather than PIO: https://github.com/adafruit/circuitpython/blob/main/ports/raspberrypi/common-hal/countio/Counter.c#L45
pulseio.PulseOut looks like it also does interrutpts, so maybe you could do a PulseOut to a counter
crimson ferry
I may have a board that can do DVI, but I haven't messed with that at all. I'm trying PulseOut now, no need for countio (maybe) since PulseOut is generating its own interrupts.