#circuitpython-dev
lone sandalBOT
manic glacierBOT
tulip sleet
<@&356864093652516868> Weekly meeting starts in about two hours. Add your notes to the notes doc if you'd like: https://docs.google.com/document/d/144fIOxqiH-xJLrzGLr7FSjrF0y5hylkiNM5iMPc7vpU/edit?usp=sharing
tulip sleet
<@&356864093652516868> meeting starting now
onyx hinge
that list of examples could be a new "awesome" list, to be maintained in adafruit org or circuitpython org on github
thanks dan!
slender iron
thanks for hosting dan!
manic glacierBOT
lone axle
Thanks for hosting Dan. Have a great week everyone!
tulip sleet
Here is the notes document for next Monday’s CircuitPython Weekly Meeting, July 1, 2024. 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/1GWVWuiHZLkTs9k3mCuSJaoPIpG91JJZ3cc_ogbGAXdo/edit?usp=sharing
upper radish
Not sure where else to mention this, but the Weekly Meeting video on YT has no audio from @tulip sleet . Probably not something that can be corrected, but maybe worth including a note somewhere so folks know it's not that their speakers/headphones aren't working?
lone axle
@tulip sleet I have the backup recording and at least with quick check it appears to have audio for you (and everyone else). I can share it if we want to swap the one published to YT so it will contain all of the audio.
tulip sleet
<@329766224093249548> I have the backup recording and at least with quick check ...
Yes thanks! It was showing up on the VU meter?!
lone axle
Yes thanks! It was showing up on the VU meter?!
https://www.dropbox.com/scl/fi/3xabsjoji8lm0drid8who/CircuitPython_Weekly_Meeting_2024_06_24.mkv?rlkey=nrqc7rgq2zsbgahm54m2fa89s&dl=0 you should be able to download with the link near the top right corner on this page.
tulip sleet
https://www.dropbox.com/scl/fi/3xabsjoji8lm0drid8who/CircuitPython_Weekly_Meetin...
All set, processing on YouTube. Feel free to delete from dropbox. Thanks again!
upper radish
Sounds good! 😏
midnight ember
@tulip sleet I don't really have an opinion on where examples should live. The pro of the current system is they are easily found and can be updated via PR by anyone. I really like having fast access to them in the bundle since FoamyGuy reorganized it. Another option could be a completely separate examples bundle but then all library repos wouldn't have examples included in the repo when searching github, that will make things harder for everyone all around. Because the examples live in official libraries like adafruit_requests it's ultimately up to Adafruit what to do and I'm ok with whatever folks decide. I can see the issue from both perspectives and it's not an easy decision.
slender iron
@gilded cradle https://forums.adafruit.com/viewtopic.php?p=1020439#p1020439
tender blade
is the circuitpython displayio library on github to view?
austere acorn
is the circuitpython displayio library on github to view?
displayio is a CP module, not a library, it's inside the circuitpython code repository
short tendon
@tulip sleet / @lone axle based on this gist: https://gist.github.com/justmobilize/f0ee9668d7aab6d91524b3ab53953d2c I would like to create an Adafruit Library Adafruit_CircuitPython_Base64_Stream that can be used by adafruit_io. Thoughts?
tulip sleet
<@329766224093249548> / <@382939733107408897> based on this gist: https://gist.g...
this seems ok. Is there anything equivalent in CPython?
short tendon
this seems ok. Is there anything equivalent in CPython?
Not directly, foamyguy found this: https://github.com/aws/base64io-python, but it's backwards
tulip sleet
how have people handled this in CPython? Just "by hand"?
short tendon
I think in CPython you have more memory then you need, and so you don't worry about it. The birdfeeder learn guide does this:
def send_jpeg_to_io():
# before we send the image to IO, it needs to be encoded into base64
encoded_data = binascii.b2a_base64(jpeg).strip()
# then, send the encoded_data to Adafruit IO camera feed
print("Sending image to IO...")
io.send_data(feed_camera["key"], encoded_data)
print("Sent image to IO!")
Which dies on a Pico W
tulip sleet
I think in CPython you have more memory then you need, and so you don't worry ab...
sounds like a good idea, then!
tulip sleet
I could create an empty repo with the right name, make you a collaborator, and you could populate it with cookie-cutter etc. Or you can create it in you own userspace and we can transfer it to adafruit when you're ready. ... The latter is probably easier when you're making initial changes.
short tendon
I could create an empty repo with the right name, make you a collaborator, and...
Either works for me. I have a CookieCutter with that name ready (but no other changes) that I can push up and transfer it to you (Last time I couldn't do it to the org)
tulip sleet
go ahead with that then, np
short tendon
Transfer started
tulip sleet
Transfer started
but do you want to finish the coding before transfer?
tulip sleet
ok sure
tulip sleet
Which is about ready anyways
ok transferred to me, and then I transferred to adafruit, and you are a collaborator (that seems to have carried along)
short tendon
👍
tulip sleet
<@329766224093249548> I don't really have an opinion on where examples should li...
I'm thinking some kind of central repo with categorized examples and good lists for searching would be good. But until we do anything, please don't feel inhibited about submitting examples to requests or wherever. We can always move the examples later.
short tendon
Hey @tulip sleet I'm getting this:
Successfully built adafruit_circuitpython_base64_stream-1.2.3.tar.gz and adafruit_circuitpython_base64_stream-1.2.3-py3-none-any.whl
Traceback (most recent call last):
File "/opt/hostedtoolcache/Python/3.11.9/x64/bin/twine", line 5, in <module>
from twine.__main__ import main
File "/opt/hostedtoolcache/Python/3.11.9/x64/lib/python3.11/site-packages/twine/__init__.py", line 40, in <module>
__uri__ = metadata["home-page"]
~~~~~~~~^^^^^^^^^^^^^
File "/opt/hostedtoolcache/Python/3.11.9/x64/lib/python3.11/site-packages/importlib_metadata/_adapters.py", line 54, in __getitem__
raise KeyError(item)
KeyError: 'home-page'
Error: Process completed with exit code 1.
When running actions the new library. I get the same thing when it's local on libraries I know that work
tulip sleet
have you updated all your pip-installed stuff?
short tendon
Yes. And just found this: https://github.com/pypa/twine/issues/1125
Looks like any libraries we try to update will fail...
tulip sleet
there are comments there from one minute ago! We can wait a few hours and see what happens
well, mentions from 1 minute ago
short tendon
Sounds good
tulip sleet
short tendon
Why smart people do things like: importlib-metadata >= 3.6 I will never know
stuck elbow
python ecosystem wasn't always so happy to break backwards compatibility carelessly
tulip sleet
Sounds good
if you need it to work to proceed, you can pin twine to 5.0.0 or importlib_metadata to 7.2.1
looking at https://github.com/python/importlib_metadata, the latest listed release is 7.2.1, but there's an 8.0.0 tag, and 8.0.0 is still on pypi. Maybe they are in the process of yanking 8.0.0
short tendon
I'm not in a rush. My PR is in draft and can wait. If you get bored you can take a look at it (with the millions of other things on your plate)
tulip sleet
I am looking at some crashes 🙂
short tendon
I wish I had time to dig into that more
tulip sleet
no worries, you are doing great stuff I wouldn't even think of doing
short tendon
Thanks!
manic glacierBOT
tiny peak
@tulip sleet and @onyx hinge , following up on the Monday meeting In the Weeds discussion about building project bundles for Playground guides: I made a demo repo that uses a GitHub Actions workflow to build a zip file and upload it to GitHub as a release asset:
https://github.com/samblenny/cookiecutter-playground-bundle
This is a little rough around the edges and light on docs, but the basic workflow works:
- .github/workflows/bundle_builder.yml calls
make bundleto run bundle_builder.py bundle_builder.pyreads bundle_manifest.cfg to determine which files to include in the zip bundle (note: this is not finished yet; code and README work, library copying does not work).github/workflows/bundle_builder.ymlusesgh release upload ...to upload the bundle zip file to a release (this works, see release v0.0.0-1
onyx hinge
<@329766224093249548> and <@426200547024961539> , following up on the Monday mee...
Very neat. I'll look at it later when I'm not mobile
tiny peak
Very neat. I'll look at it later when I'm not mobile
my intent is for this to be a proof of concept style contribution to the discussion on how to provide big blue "Download Project Bundle" buttons for Playground guides. If any of the code or ideas are useful, please feel free to fork, copy and paste, slice and dice, or whatever.
tulip sleet
<@329766224093249548> and <@426200547024961539> , following up on the Monday mee...
Very interesting! I am asking the Learn/Playground devs if they had any bundling in mind for Playground.
onyx hinge
settable CPU frequency pulls in that much more stuff? or it was just that close to the edge anyway? [ok ok I should go read the PR discussion]
brazen hatch
settable CPU frequency pulls in that much more stuff? or it was just that close ...
power management gets enabled, it was something like -500b before removing a lengthy string
tulip sleet
<@329766224093249548> and <@426200547024961539> , following up on the Monday mee...
FYI, th Learn folks say they have not included project bundling on Playground for various reasons, including security and maintenance. So this is not duplicative by any means. It is possible now to embed a single file from GitHub, but not a whole bundle.
onyx hinge
FYI, th Learn folks say they have not included project bundling on Playground fo...
I suspected as much. I'd blanche at the idea of feeding arbitrary untrusted python code to "findimports" even if it 'should be safe'!
creating tools for building bundlefly-like zips moves the responsibility to the project, which by definition can trust its own source code
tulip sleet
yes, it's a good idea
tiny peak
I suspected as much. I'd blanche at the idea of feeding arbitrary untrusted pyth...
I guess it sounds like it might be worth some additional effort to polish up my demo and get the library copying part working? FWIW, I don't have any aspirations of owning or maintaining a cookiecutter that might have the potential to become infrastructure-ish for Playground authors. If somebody else wants to build something off my code, that would be fine by me.
It will probably be another day or two before I get back to looking at the bundle_builder.py script. If nobody beats me to it though, I probably will come back around for another pass to finish up the library stuff.
tulip sleet
It will probably be another day or two before I get back to looking at the bundl...
no hurry about this! we can transfer the repo to adafruit when it's done
tulip sleet
Looks like any libraries we try to update will fail...
twine yanked 5.1.0, and released 5.1.1, so this should be fixed now. See https://github.com/pypa/twine/issues/1125 as you already linked, for details.
short tendon
twine yanked 5.1.0, and released 5.1.1, so this should be fixed now. See <https:...
That worked. Tests all passed!
brazen hatch
x86
arm
half a kb difference
same commit on both
both on arch latest
I mean they are both compiling a different architecture, xtensa..
(And yes, my new sbc is faster than my laptop)
manic glacierBOT
warm topaz
Hey, I have the ESP32-S3-Touch-LCD-1.28 (that's the touch version, I was wondering when their would be an official firmware for this board? it seems to be the same as the RP2040 version but with an S3 instead,
manic glacierBOT
tulip sleet
Hey, I have the ESP32-S3-Touch-LCD-1.28 (that's the touch version, I was wonderi...
CircuitPython has builds in its base repo for many third-party boards, but they are largely contributed by volunteers or by the manufacturers.
Sometimes we add them ourselves when they are the mfr dev boards (e.g. Espressif) and that is the first example of a port to that chip, etc. Pi Pico is another example
warm topaz
thanks, it seems to build on it fine, it's just the touch screen only seems to speak to the I2C when touched and otherwise doesnt show up under an address scan?
slender iron
thanks, it seems to build on it fine, it's just the touch screen only seems to s...
It's best to ask the board's manufacturer for help.
tulip sleet
@slender iron could you tell me again under what conditions the BLE workflow is advertised? Is it any time a BLE-capable board comes up and is not connected, or do you expect there to be some setting that will enable possible BLE workflow. (I'm fixing something about supervisor_bluetooth_start_advertising()), which is being called on every supervisor_bluetooth_background() call, and is doing too much work.)
the background call is happening on every background tasks run, I think
slender iron
it should advertise if bonded or if boot_in_discovery_mode is set
that doesn't seem right
tulip sleet
have time to talk?
slender iron
yup
manic glacierBOT
orchid basinBOT
manic glacierBOT
tranquil ether
I'm starting to bring up my Adafruit LoRa Radio FeatherWing - RFM95W 900 MHz - RadioFruit (Adafruit P/N 3231). I'm using the Prop-Maker 2040. I'm starting with the shipping example "rfm9x_simpletest.py. For wiring, I chose to connect CS to A0 and RST to A1.
CS = digitalio.DigitalInOut(board.A0)
CS.direction = digitalio.Direction.OUTPUT
RESET = digitalio.DigitalInOut(board.A1)
RESET.direction = digitalio.Direction.OUTPUT
And this is the error I'm getting:
I think it is OK to use A0 and A1 as Digital IO's - anyone disagree?
Any help is much appreciated. Must walk a way for the moment.
CircuitPythoin V9, by the way.
midnight ember
CircuitPythoin V9, by the way.
A0 and A1 should work. You'll also want to wire up the IRQ pin as it's used in most code as a hardware interrupt pin. You'll get faster transceiver switching between send/receive by using the IRQ instead of a software based IRQ.
tranquil ether
Ok. Thanks for that. However, I’m still stuck with that error. The code that I’ve started with doesn’t call for the IRQ pin
midnight ember
Hmm maybe not. The propmaker rp2040 doesn't have a digital pin alias for a0 or a1. They might be analog only.
It's best to use one of the A to E pins which go to digital pins.
tranquil ether
Ok. I can try other pins, I suppose
midnight ember
yeah most demos of the rfm featherwing use those pins.
tranquil ether
I've already used the pins that your photo suggests and I've done a fair amount of wiring to use them. Not too easy to undo that.
however, everything on the left except for the SPI port is open
midnight ember
Interesting. I didn't see those aliases when looking it up in pins.c board definition.
oh derp nvm that's the featherwing not the feather.
tranquil ether
well, my table on top of the pretty pins may not be correct
here is the dir(board...
import board
dir(board)
['class', 'name', 'A0', 'A1', 'A2', 'A3', 'ACCELEROMETER_INTERRUPT', 'BOOT', 'BUTTON', 'D0', 'D1', 'D10', 'D11', 'D12', 'D13', 'D24', 'D25', 'D4', 'D5', 'D6', 'D7', 'D9', 'EXTERNAL_BUTTON', 'EXTERNAL_NEOPIXELS', 'EXTERNAL_POWER', 'EXTERNAL_SERVO', 'I2C', 'I2S_BIT_CLOCK', 'I2S_DATA', 'I2S_WORD_SELECT', 'LED', 'MISO', 'MOSI', 'NEOPIXEL', 'RX', 'SCK', 'SCL', 'SDA', 'SPI', 'STEMMA_I2C', 'TX', 'UART', 'dict', 'board_id']
tranquil ether
oh, oh... pins.c that's a big help
I wish the pretty pins would show the same infor as the dir(_board)
midnight ember
According to the diagram it has digital capability on A0 and A1 but in pins.c there aren't digital aliases. Maybe it does it automatically?
Assuming A0 and A1 can be used as digital then the error is still saying that something isn't correct with the wiring setup.
instead of A0 and A1 try D0 and D1
tranquil ether
I could try D4 and D5. They are open on my pin assignments and they start with 'D'
midnight ember
they're listed in your dir(board)
D0 and D1 are usually the digital alias for A0 and A1.
tranquil ether
OK, AKA Tx and RX for the serial port
tranquil ether
If I'm reading pins.c correctly, D0 amd D1 are
{ MP_ROM_QSTR(MP_QSTR_D1), MP_ROM_PTR(&pin_GPIO0) },
{ MP_ROM_QSTR(MP_QSTR_D0), MP_ROM_PTR(&pin_GPIO1) },
midnight ember
ah oops, you're correct
tranquil ether
... are GPIO0 and 1. And then using the prettyPins, thats Tx, RX lower left of the pinout
midnight ember
this is why its good to pay more attention to the pins.c because that's the real list of pins and aliases.
tranquil ether
yes, yes. OK, I'll give that a try first thing tomorrow. Thanks a million for your help!
midnight ember
I could be wrong and you could end up in the same situation just with different pins tomorrow. I would solder in that irq pin too.
so don't thank me just yet, i'm not 100% sure of my advice.
tranquil ether
haha. yes, for sure, it may not be solved yet. But I have somethign pretty logical to try and that's somethign!
midnight ember
yep new avenue to try. feel free to tag me tomorrow.
tranquil ether
will do!
midnight ember
didn't realize that was all in dev chat, moved to #help-with-circuitpython
empty salmon
I am somewhat sure there is a problem within pulseio when both synthio and irremote are using it. What I do not know is what more I can do to dig into the root cause - or what my next steps should be.
Environment: CircuitPython 9.1.0-beta3 on RP2040
Symptoms:
- irremote generating huge pulse decode sequences or generating a memory allocation error trying to allocate 65536 bytes.
- synthio (using PWM) briefly having stuttered note or note stuck for a short time before recovering.
digital shoreBOT
adafruit
orchid basinBOT
manic glacierBOT
manic glacierBOT
update - I'll have to test some more boards - I'm getting some intermittent issues with http://192.168.1.118/cp/serial today as well. I've tried an s2 mini (not my port) a lolin d1, and an esp32cam (my port) so far - I'll try some others...
I am using your esp32cam (awesome work BTW!) and /cp/serial works only sometimes, even using the IP:
BTW you might want to check what [mi...
manic glacierBOT
manic glacierBOT
alpine holly
Is pyexec_file or pyexec_frozen_module exposed to the REPL or python itself? (aka parse_compile_execute) [I started a post in the forum: https://forums.adafruit.com/viewtopic.php?p=1020937 ] - I'm wanting to efficiently run (chain) a new python program from an existing one [ending the existing one, starting the new [possibly .mpy compiled] one, without burning out the flash]. My experiments show it's only a few bytes of compiled "C" to add this in (works nicely on my esp32 test boards so far). Along the way, I noticed code.mpy does not run at boot? Seems contrary to the doc about the .mpy versions of things taking precedence?
manic glacierBOT
stuck elbow
Is pyexec_file or pyexec_frozen_module exposed to the REPL or python itself? (ak...
I do this by using supervisor.set_next_code_file and restarting
empty salmon
CircuitPython appears to take over GP0 and GP1 on an RP2040.
Is there a way to make these available as general IO pins (knowing they will not be used for UART by a CircuitPython program)?
My board does not have the board_uart_obj in its pins.c
stuck elbow
CircuitPython appears to take over GP0 and GP1 on an RP2040.
Is there a way to m...
they should be available by default, unless you specifically marked them for some other use in your board configuration
they are exposed on the pi pico, for example, and they work perfectly fine
empty salmon
Must be something else then. Thanks.
orchid basinBOT
grave parrot
Does CircuitPython currently support annotations like these : @micropython.asm_thumb
onyx hinge
Does CircuitPython currently support annotations like these : @micropython.asm_...
no, we never enable MICROPY_EMIT_INLINE_ASM for any boards. Right now a few specific boards enable the "native" emitter (CIRCUITPY_ENABLE_MPY_NATIVE), but I don't know whether anyone is using/testing this either (they are not adafruit boards). If support for asm_thumb is important you'll probably want to go with micropython.
brazen hatch
Does CircuitPython currently support annotations like these : @micropython.asm_...
Now it does. I'll open a PR soon.
alpine holly
I know this is old, but, maybe you've noticed that some ESP32 boards let you program them without you having to reset the board while holding GPIO0 down? The way those work is by hijacking the hardware flow-control lines of the RS232 driver and using those to programatically reset the chip while programatically holding down GPIO0. esptool (and other flashing programs) does this for you. On those boards, there is NO hardware flow control (because the lines were hijacked for other purposes). Unfortunately, other boards actually do honor hardware flow control, which conflicts with esptool and other serial software which expects those pins to be hijacked for programming purposes. Hence my post. WE need some way to inform all that software what those pins are used for, so the relevant software in use (e.g. serial terminals, serial file-transfer code, serial REPL, ..., and web page flashers) knows whether to use them for RESET/GPIO or for Flow Control (the latter requiring the addition of an instruction to tell the user how to put the chip into upload mode), because you can't have both at once, and you cannot guess without causing one-or-the-other to lock up.
I'm working on fixing the broken espnow issues (cannot set channels, cannot broadcast, and lots of data errors during recieve).
Does anyone know how to add a "debug output" line (a simple "print" would be fine) into core "C" modules ?
I'm trying to work out which (of many) calls is returning the error.
stuck elbow
Does anyone know how to add a "debug output" line (a simple "print" would be fin...
search the existing code for printf
alpine holly
search the existing code for printf
Worked !!! I was totally not expecting it to be that simple 🙂 Thanks for the nudge.
tulip sleet
I know this is old, but, maybe you've noticed that some ESP32 boards let you pro...
We discussed this at length at the time, I thought. "We need some way to inform all that software" -- I think this has to be in the tools, if you expect the tools to support this kind of board. The tools need to know this way before CircuitPython even runs. I said at the time I thought these boards (e.g. ESP32CAM) were ill-designed.
onyx hinge
(it's very weird to me that whatever that tinyusb change was, it affected just 2 board builds out of all of them!)
candid sun
<@&356864093652516868> We'll have our weekly meeting in about 60 minutes from now in this text channel and in the circuitpython voice channel. Please take the time to add your notes in advance to the document: https://docs.google.com/document/d/1GWVWuiHZLkTs9k3mCuSJaoPIpG91JJZ3cc_ogbGAXdo/edit#heading=h.gwb5d95e5zic -- I look forward to everyone's updates!
manic glacierBOT
tulip sleet
(it's very weird to me that whatever that tinyusb change was, it affected just 2...
those may have just been broken builds?
that is, transient
onyx hinge
I didn't look in detail because github was having trouble displaying logs (it always has trouble with long logs) so I didn't see whether there was anything that looked like network or cloning problems
since the PR changed the tinyusb submodule and the errors I could see were about tinyusb I jumped to a conclusion
tulip sleet
the failures are weird, but we can ignore them for now. The mpy-cross-mac job is hanging waiting for a runner. this has happened several times in other PRs as well. I thought you bumped the version already to macOS 12?
onyx hinge
@tulip sleet looks like that went in 9.0.x and may not be merged to main? ```diff
~/src/circuitpython$ git diff origin/9.0.x origin/main -- .github/workflows/build.yml
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index 13f758c1be..03cecae231 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -102,7 +102,7 @@ jobs:
cp-version: ${{ needs.scheduler.outputs.cp-version }}
mpy-cross-mac:
- runs-on: macos-12
- runs-on: macos-11
needs: scheduler
if: needs.scheduler.outputs.ports != '{}'
env:
ah, ok, I didn't see that it was for 9.0.x? Could you cherry pick that to main (or do a 9.0.x -> main merge)?
onyx hinge
I'll prepare a PR to merge 9 to main
tulip sleet
are the mpy-cross renamings already in main?
onyx hinge
@tulip sleet no, both were done in 9.0.x
I did it in 9.0.x because we would need the change in both versions, assuming we made further 9.0.x releases.
thanks for not ignoring today's brownout, I was all set to do so
gilded cradle
Sorry, was afk for a moment
midnight ember
Lurking, appreciate if someone can read for me.
midnight ember
Looking forward to reading that CGrover
errant grail
Looking forward to reading that CGrover
Finding a local "approved" observation station is non-trivial. Once you know which station to monitor, the API is pretty straightforward.
onyx hinge
awww feel better soon scott!
slender iron
👍
slender iron
awww feel better soon scott!
tylenol helps 🙂
manic glacierBOT
slender iron
and could be worse too
onyx hinge
thanks liz!
lone axle
Thanks for hosting Liz, hope everyone has a great week.
onyx hinge
short & sweet
candid sun
thanks folks!
slender iron
@candid sun do you have an example project it'd help with?
candid sun
<@347203168121126912> do you have an example project it'd help with?
a few 😅 but this one: https://learn.adafruit.com/4x4-rotary-encoder-midi-messenger and this one: https://learn.adafruit.com/circle-of-fifths-euclidean-synth-with-synthio-and-circuitpython in particular
midnight ember
@candid sun and the solenoid xylophone. i think i remember you talking about latency at the time, but that might have been from playing midi from a DAW?
candid sun
<@347203168121126912> and the solenoid xylophone. i think i remember you talking...
yeah that was more DAW related and BLE midi. i've noticed recently that that seems to be less of an issue. with the robot lyre i didn't have any latency issues and when i've been using the robot xylo recently it's been responsive
slender iron
a few 😅 but this one: https://learn.adafruit.com/4x4-rotary-encoder-midi-messen...
awesome! thanks!
candid sun
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1_zyNviU0e7doCx2nxfv0r2V85XiOYp7oe_lgMbxiCfE/edit
manic glacierBOT
empty salmon
@tulip sleet - thanks for sticking with me as I stumbled thru the PR 👏🏼
tulip sleet
np! That submodule thing was pretty mysterious.
manic glacierBOT
manic glacierBOT
alpine holly
We discussed this at length at the time, I thought. "We need some way to inform ...
Yes - and many of the "tools" are from and by adafruit - so there needs to be a way inside out ports, to tell the system that later-on provides flashing and rs232 consoles etc, how to do that.
I disagree with your "ill-designed" opinion - you're just used to the inconvenience of hitting buttons every time you want to flash stuff: it makes significantly more sense to have the software sort that out all by itself instead of making you take manual extra steps to flash your board - which is what all these espressif designs have done. circuitpython itself disables hardware flow control (XON/XOFF is default), so it's actually all the newer boards who've made implementations using hardware flow control which are the cause of the problems (because it conflicts with the pre-existing GPIO0/RST functionality on older boards for no sensible reason). To be fair - it probably wasn't intentional - they probably didn't even know they accidentally enabled hardware flow control in their designs; the esp32 doc on serial is a mess, and "hardware" in terms of modern usb emulation (ESP32C3/ESP32S2) which doesn't even have actual physical lines for those anyhow just muddies the waters.
spare jacinth
Havent read the full conversation for context, but for the "inconvenience" aspect I think the button is a good idea. Prevents accidental flashing (or an attacker doing it without you even noticing, which lets be fair will not be a major concern)... its not like pressing a button takes a lot of effort anyway
tiny peak
@tulip sleet and @onyx hinge , an update on the cookiecutter playground bundle thing from last week... I modified my repo to have a cookiecutter template directory and json config file. Seems to work for populating files into a new empty repo, but I haven't tested it very thoroughly yet. Next up is figuring out how to copy library dependencies into the project bundle zip file.
lone axle
<@329766224093249548> and <@426200547024961539> , an update on the cookiecutter ...
You might be able to leverage circup to help with that. If you have some directory project_dir/ you can install libraries into that directory with a circup command like this:
circup --board-id whatever --cpy-version 9 --path project_dir/ install adafruit_display_text
It will automatically include any extra requirements as well i.e. this will install bitmap_font library too.
lone axle
<@329766224093249548> and <@426200547024961539> , an update on the cookiecutter ...
If you have the code.py file for the project you can also have it scan that and install anything it imports rather than needing to list them out manually. Something like this should do it:
circup --board-id whatever --cpy-version 9 --path project_dir/ install -a --auto-file project_dir/code.py
Thanks for the tip. For now, my plan is to just list out specific libraries to be included using a manifest file, then extract those from a downloaded copy of a specific Adafruit_CircuitPython_Bundle release zip file (also listed in the manifest file). That arrangement should work as a form of version pinning to give some stability and bitrot-resistance to Playground guides that may not get actively maintained. If my repo ends up getting transferred to Adafruit (seems likely?), then the door would definitely be open for other folks to come along and improve on the mechanism for dependency resolution.
manic glacierBOT
tulip sleet
@slender iron I can't reproduce the two remaining 9.1.0 issues. So maybe the last beta or the first rc can be in the offing.
tiny peak
<@329766224093249548> and <@426200547024961539> , an update on the cookiecutter ...
I think my cookiecutter project bundle thing pretty much works now? I added support for libraries today. You have to manually list the names of the ones you want included using a manifest file. Probably some folks would prefer automatic dependency resolution, but implementing such a thing is more than I care to tackle at the moment. Code is here: https://github.com/samblenny/cookiecutter-playground-bundle
manic glacierBOT
atomic summit
Hey folks - having issues writing to my TinyC6 running CP (9.0.5) - Thonny, ampy, rshell & mpremote, all failing to copy files to flash using the internal SerialJTAG peripheral.
This was discussed in this github issue a while ago, that's still open.
https://github.com/adafruit/circuitpython/issues/8658
I know Adafruit have their Adafruit ESP32-C6 Feather "coming soon" - anyone able to comment on it's usability with regards to file writing?
I seem to have success using my rPI to write to the boards - as that's what my test jigs use, and I'm able to ship boards.... But on my Mac it's almost non usable, and I have no Windoz machines here to test on.
I'm happy to send some TinyC6 to the CP team for diagnosis if that will help.
manic glacierBOT
atomic summit
Ok, I'm going to create an issue about this C6 thing, but wanted to share here first it's working as expected in MP using an upcoming PR supporting esp32c6 and IDF 5.2.1 and 5.2.2, so hopefully based on that we can find what's different in CP and sort it out 🙂
manic glacierBOT
manic glacierBOT
onyx hinge
in CP if you call sys.stdout.write("x") , my expectation is that the data will be available to read on the host computer in a timely fashion without any further Python code intervention. This would be a basic requirement for the REPL to operate properly. I'm trying to understand if #9393 is about providing an even stronger guarantee or that this basic (and probably not fully articulated) guarantee is not being met.
I was curious what pyserial & a USB serial converter did. write+flush doesn't guarantee that the bytes have actually arrived, as demonstrated by some non-blocking reading (cp2102 is looped back to itself): ```py
s.reset_input_buffer(); s.write(b"hello"); s.flush(); s.read(5)
5
b'hel'
s.read(5) # after a moment
b'lo'```
Furthermore, the sequence write/flush/set control line doesn't ensure the control line transition happens only after the full transmission, as seen by a saleae watching both lines. the 2nd toggle of RTS at +.33s is after write(b"hellohello"); flush() but it occurs before the last "o" has been output. I really didnt' expect that!
slender iron
<@252717193496756235> I can't reproduce the two remaining 9.1.0 issues. So maybe...
Sounds good to me!
tulip sleet
Sounds good to me!
There are a couple of new 9.1.0 bugs and things still to explore. I am in the process of making a beta.4 iinstead of an RC and will release that today. I think it's not quite cooked yet.
onyx hinge
if (usb_cdc_console_enabled()) {
COUNT(tud_cdc_available());
}
#endif
#if CIRCUITPY_USB_DEVICE
COUNT(tud_cdc_available());
#endif
```well that's one reason it could be returning 2 when just 1 byte is available(my version is slightly modified, COUNT printed debug info and added to the count local variable)
9.0.x returns bool from serial_bytes_available while main switched to returning int so it wasn't a problem before. true + 1 == true in C
plucky tulip
Is there currently a tool like circup, but for building .py files to .mpy, I mean I know these are the .exe files etc, but I am talking about a CLI that can accept CPY version and detects a platform, then automatically downloads correct mpy compiler?
If not, would such tool be useful?
spare jacinth
Is there currently a tool like `circup`, but for building `.py` files to `.mpy`,...
i dont think it would be very useful
i have no metrics but would expect the amount of people that compile their code into .mpy to be marginal
and just to be sure i understand the question, you would want a "version manager" for (was it named mpy-cross), right? as in... instead of going to its releases page and grab the one you need
plucky tulip
i dont think it would be very useful
i have no metrics but would expect the amou...
Yeah, that is what I am taling about. You are probably right about the amount of people that would use something like this, today I wanted to test something and I somehow assumed that this is part of circup, and when I couldn't find that option i started wondering whether this would be useful as part of circup or as a separate tool.
spare jacinth
i mean, the cross compiler exists, if that's what you want
it is used on github action runs on the bundle repositories, to generate build artifacts(.zip's) that circup downloads, extracts and copies into your board
i think that those zips only contain mpy and not py for convenience/space saving, but that does not mean that you need circup in order to create mpy
manic glacierBOT
Here's data showing correct behavior at the extremes:
While I was at it I fixed the rounding issue that I had earlier addressed for samd in https://github.com/adafruit/circuitpython/pull/9343, and tested by the same method that I used there. Here are before and after plots:
Before
manic glacierBOT
orchid basinBOT
manic glacierBOT
tulip sleet
https://blog.adafruit.com/2024/07/03/circuitpython-9-1-0-beta-4-released/
Includes much more complete support for BLE on Espressif: acting as a server, and pairing/bonding
manic glacierBOT
I have flasehd the latest build, and the name change works on nRFConnect. However on Adafruit Bluetooth Connect it still shows the device name as CIRCUITPY(id), with the Local Name property set as the new name. I tried restarting the board and the phone, without luck. Anything else I should try?
fervent steeple
Feather M4 Power Problem:
Hello, this is my first time working with programmable LED's and I am using the Feather M4 Express with the FeatherWing board. I successfully programmed the chip to do the LED animation and it works great when it is plugged into my computer. But when I disconnect the USB plug and insert my 3.7V battery into the battery port, nothing happens. I've confirmed the battery has power as it is a rechargeable one but nothing turns on when I plug it in. What am I doing wrong? Below is my code for reference:
import digitalio
import board
import neopixel
from adafruit_led_animation.animation.pulse import Pulse
from adafruit_led_animation.animation.sparklepulse import SparklePulse
from adafruit_led_animation.helper import PixelSubset
from adafruit_led_animation.group import AnimationGroup
from adafruit_led_animation.sequence import AnimationSequence
from adafruit_led_animation.color import (
AMBER,
ORANGE,
)
NUM_PIXELS = 116
NEOPIXEL_PIN = board.D5
POWER_PIN = board.D10
ONSWITCH_PIN = board.A1
strip = neopixel.NeoPixel(NEOPIXEL_PIN, NUM_PIXELS, brightness=.5, auto_write=False)
strip.fill(0)
strip.show()
enable = digitalio.DigitalInOut(POWER_PIN)
enable.direction = digitalio.Direction.OUTPUT
enable.value = True
group1 = PixelSubset(strip, 0, 15)
group2 = PixelSubset(strip, 15, 116)
pulseOrange = Pulse(group1, speed=.05, color=AMBER, period=6)
sparkle_pulse = SparklePulse(group2, speed=2, period=5, color=ORANGE)
animations = AnimationSequence(
AnimationGroup(
pulseOrange,
sparkle_pulse,
sync=True,
),
AnimationGroup(
pulseOrange,
sparkle_pulse,
sync=True,
),
)
while True:
animations.animate()
If this isn't the right place to ask this question, please let me know. Thanks for the assistance.
random junco
Hello, this is my first time working with programmable LED's and I am using the ...
Hi, I'd recommend asking in #help-with-circuitpython - this channel is for development of CircuitPython itself. Cross-posting is usually discouraged, but in this case I'd recommend it. Also see #welcome for how to post code.
fervent steeple
Thank you!
digital shoreBOT
adafruit
manic glacierBOT
wanton stirrup
Guys I need a computer
stuck elbow
computers were a mistake
manic glacierBOT
misty garnet
@tulip sleet FoamyGuy fixed the test and approved the MiniMQTT PR if you can give it a look when you have time.
fair dagger
hi folks, I'm curious why adafruit-blinka requires numpy as a dependency. It's not used in the codebase AFAIK. https://github.com/adafruit/Adafruit_Blinka/blob/main/setup.py#L96
I'm mostly curious because it's unpinned upperbound, and numpy==2.0 (released a few weeks ago) is breaking some downstream stuff for me.
Edit: actually it's not numpy 2.0 that breaking things, it's that the numpy install via pip doesn't work well for RPis. I use the apt-get install python3-numpy, but now blinka ships with numpy (as of 8.44.0), and this is causing pip install blinka to install numpy via pip.
Anyways, my solution (I think) is to pin blinka to 8.43.x
atomic summit
I just grabbed the latest "main" from the cp repo and building it gives me 9.1.0-beta.3-62-g121d9ee42b and not 9.1.0-beta-4 - is there something I am missing?
onyx hinge
make fetch-tags?
atomic summit
`make fetch-tags`?
hmm, yeah, that was it, but why was that needed on a brand new clone of latest main?
The merged changes for beta4 are in the repo.
manic glacierBOT
midnight ember
I'm having some weird issues with a Gemma M0. Just downloaded the UF2 adafruit-circuitpython-gemma_m0-en_US-9.0.5.uf2flashed it, everything seems fine. boot_out.txt Adafruit CircuitPython 9.0.5 on 2024-05-22; Adafruit Gemma M0 with samd21e18 Board ID:gemma_m0 and getting an error that it cannot find the module. the module is definitely in lib. code.py output: Traceback (most recent call last): File "code.py", line 7, in <module> File "adafruit_dotstar.py", line 19, in <module> ImportError: module not found and it's specifically asking for the .py file which I gave it the PY file from the bundle but still throws that same error.
can someone try a Gemma M0 with latest stable before I submit a bug report. this is weird and i cannot explain it.
no idea if the latest build date showing from May has any bearing. I just downloaded the latest stable 9.0.5, is Gemma M0 not building properly since May?
onyx hinge
I'm having some weird issues with a Gemma M0. Just downloaded the UF2 `adafruit-...
Line 19 of adafruit_dotstar.py is import adafruit_pixelbuf. is adafruit_pixelbuf.mpy (or py) there? We had to simplify error messages on those tiny old boards so it doesn't actually say the name of the module that couldn't be found, which is confusing but we don't have a great alternative
midnight ember
yeah that's an issue. it doesn't come with pixelbuf and no mention it requires it. :/
manic glacierBOT
midnight ember
also the Gemma M0 is still shipping with Circuit Python 5.
onyx hinge
this is why newer products don't ship with circuitpython anymore, it gets out of date. fwiw I tried to update a pyruler project to 9 recently and just gave up, it couldn't import usb_hid 😕
tulip sleet
@atomic summit whose micropython fork are you using for C6? I just want to look at the code.
Never mind, I see https://github.com/andrewleech/micropython/tree/esp32c6
atomic summit
<@195765176035835904> whose micropython fork are you using for C6? I just want t...
Yup, that’s the one.
manic glacierBOT
nimble sparrow
@onyx hinge Great work polishing up audiomp3 stream support, covers so many use cases
nimble sparrow
<@426200547024961539> Great work polishing up audiomp3 stream support, covers so...
Now we just need to pass those espressif AAC decoder bindings through 😛
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
tulip sleet
<@&356864093652516868> The weekly meeting will start in a little over two hours. If you have hug reports and status updates, add them to this week's notes doc: https://docs.google.com/document/d/1_zyNviU0e7doCx2nxfv0r2V85XiOYp7oe_lgMbxiCfE/edit?usp=sharing
blissful pollen
I haven’t been able to attend the weekly in a while but I saw this at a park here recently and thought I’d share! Group hug to all
wraith crow
What was the make command to reset/remove an old make fetch submodules command?
tulip sleet
make remove-all-submodules
manic glacierBOT
onyx hinge
@tulip sleet obs isn't screen-capturing for me, restarting ...
candid sun
@onyx hinge @tulip sleet i can run a back-up recording
onyx hinge
what do I do to fix names showing up shortened?
ah there it is
I think I'm ready to record
random junco
turbid radish
Thanks all who contribute and subscribe
slender iron
they allude to it but don't have an explicit section for it
onyx hinge
@tulip sleet let me know if you want my recording, I think it went OK but I'll check it now.
slender iron
lemme find an example
onyx hinge
my recording was fine in OBS after I restarted. This happens sometimes, it just isn't able to screen capture and I don't know why
tiny peak
@tulip sleet do you want me to transfer that cookiecutter repo?
slender iron
for example:
tulip sleet
my recording was fine in OBS after I restarted. This happens sometimes, it just ...
did you mean you restarted the OS or just OBS?
onyx hinge
@tulip sleet I restarted the whole OS (linux)
tulip sleet
I'll try that and see what happens tnx
tulip sleet
<@329766224093249548> do you want me to transfer that cookiecutter repo?
sure, you can go ahead and initiate it. I think you need to transfer it to an individual like me (@dhalbert) and then I'll transfer it to Adafruit
thanks!
tiny peak
sure, you can go ahead and initiate it. I think you need to transfer it to an in...
sounds good. I'll give it a try.
tiny peak
sounds good. I'll give it a try.
I initiated the transfer to @dhalbert
manic glacierBOT
tulip sleet
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1C4XIYb0tcKxlzNpwMI5pGf2OmRlzvSLOz7nbUt_Jo3s/edit?usp=sharing
wraith crow
I just ran a pre-commit and ended up with a bunch of formatting changes for files I didn't modify. I'm wondering if the extra formatting was a result of running an old pre-commit? Or should I just go ahead and submit the unrelated formatting changes with my PR?
slender iron
It can be a mismatch of a formatter dependency. I'd just include the files you changed
thorny jay
Oups. Well done danh.
onyx hinge
(hi david!)
tulip sleet
@onyx hinge I inquired on the keebio discord, and the flash chip they said is used is W25Q16JVxQ, not the EXTERNAL_FLASH_DEVICES = "ZD25WQ16B, W25Q32JVxQ" in the maple elite-pi board def
onyx hinge
The web page https://docs.keeb.io/elite-pi-guide also says W25Q16
Flash Memory: 16Mbit (W25Q16JVUXIQ)
tulip sleet
i'll build another UF2 to test
I didn't think it would work at all or it would work at reduced performance with the wrong chip, but I don't know the details of what happens when the chip is wrong.
onyx hinge
the other report said "EK68Custom" whatever that is
doesn't seem to be a firmware we ship
tulip sleet
we can leave it at that until they test again. If the Pi Pico UF2 works, that would be telling
manic glacierBOT
manic glacierBOT
manic glacierBOT
lone axle
@tulip sleet I'm curious if you had to do anything "special" in order to request review from circuitpython librarians on https://github.com/adafruit/Adafruit_CircuitPython_Base64_Stream/pull/1 ? I noticed yesterday (just before my hard drive died) that arduinolibrarians was tagged initially and went to change it but circuitpython librarians was not in the list anywhere that I could see. There were a dozen or so different entries with various people and the arduino librarians, but no circuitpythonlibrarians specifically.
tulip sleet
I mistakenly tagged the repo for review by arduino librarians: I just saw "librarians" and though it was circuitpython librarians. So I went to https://github.com/adafruit/Adafruit_CircuitPython_Base64_Stream/settings/access and did "Add teams" and added CircuitPythonLibrarians.
When this repo got transferred, it should have gotten set up as an Adafruit repo, but that didn't happen. That might be on me.
lone axle
Ahhh, I see. I didn't realize the teams worked that way. I was quite confused to see all of the names except for circuitpython librarians. Thank you.
tulip sleet
Ahhh, I see. I didn't realize the teams worked that way. I was quite confused to...
do you have privs to add teams?
lone axle
do you have privs to add teams?
I do not. I don't see the settings page at all.
but an familiar with it from my own repos.
tulip sleet
if you click that link what do you get?
lone axle
github 404 page.
tulip sleet
@slender iron If EXTERNAL_FLASH_DEVICES in mpconfigboard.mk is incorrect on an RP2040 board, what happens? Does it use some generic settings? I am trying to understand why the setting for the board in #9413 is wrong but it still works.
midnight ember
After reading CGrovers excellent playground note on transitioning from openweathermaps to weather.gov I double checked openweathermaps to find this notice. The One Call API 2.5 has been deprecated on June 2024. Please read the detailed guide for transitioning to the more advanced One Call API 3.0 which includes several useful upgrades like 45 years of history archive, daily aggregation and human-readable daily summary in comparison with the One Call API 2.5 version. Read more about One Call API 3.0 on the product documentation page and FAQ.
There are Adafruit Learn Guides that use that API that it might affect. I'm still using the 2.5 Free API. Even though it says it's been deprecated it is still working for me. Might actually have to deprecate the requests api example if openweathermaps goes to a paid only API.
The difference being if you go over the limit with 3.0 it will charge you. With 2.5 free API if you go over the limit it will temp ban you for 24 hours. I'd much rather be temp banned then get charged especially if a script messes up and continually tries to connect every x milliseconds... I'm not sure if it's configurable to ban rather than charge. This is a concerning policy change in their API.
and there are more..
slender iron
<@252717193496756235> If `EXTERNAL_FLASH_DEVICES` in `mpconfigboard.mk` is incor...
It depends on what way it is wrong
and how the flash chip settings differ
slender iron
@onyx hinge at my desk now
tulip sleet
It depends on what way it is wrong
In this case it is "ZD25WQ16B, W25Q32JVxQ" but the chip is really a W25Q16JVxQ.
so the W25Q mfr stuff would probably match, but the size is wrong
onyx hinge
@slender iron hi! I'll head down to my desk in a second. I've been trying to find a good way to debug to USB CDC on espressif...
@slender iron I hooped in Amelia
hopped even
onyx hinge
seems CONFIG_IPV6 is already enabled to some degree. interesting. ```build-adafruit_metro_esp32s2/esp-idf/sdkconfig
1137:CONFIG_LWIP_IPV6=y
I was hoping it'd be easy!
@onyx hinge looks like thread can just be another network interface: https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-reference/network/esp_openthread.html#id6
slender iron
@onyx hinge got it connected
tulip sleet
@onyx hinge @slender iron if you are able to review https://github.com/adafruit/circuitpython/pull/9422 and https://github.com/adafruit/circuitpython/pull/9419, I'll merge those and make 9.1.0-rc.0. I have made release notes. I have pushed https://github.com/adafruit/circuitpython/issues/9362 to 9.x.x because it's not critical and I can't reproduce.
slender iron
<@426200547024961539> <@252717193496756235> if you are able to review <https://g...
Done!
tulip sleet
thanks!!
slender iron
@onyx hinge js only version of matter: https://github.com/project-chip/matter.js
there is a rust version too
wireshark dissector https://github.com/project-chip/matter-dissector
orchid basinBOT
tiny peak
For anybody interested in ways of sending video (or bitmaps in general) from a circuitpython board to a host PC, I came up with a technique that works pretty well using base64 and Web Serial. Turns out it even works to run the static web GUI page off the CIRCUITPY drive as an index.html file (so the whole setup can be packeged in a project bundle). A couple potential advantages of this approach (framed base64 over serial port) compared to usb_video are that it doesn't require special USB stack support, and it would be easy to adapt for sending various types of data (frame buffer, color palettes, status text, or whatever). Anyhow, here's the code and here's the Playground guide.
ornate breach
For anybody interested in ways of sending video (or bitmaps in general) from a c...
Would you mind moving this to #show-and-tell ? This channel is mostly for discussing the development of circuitpython, show and tell generally gets more traffic for cool stuff like what you’re sharing 🙂
That way it doesn’t get buried in GitHub feed stuff
deep haven
if len(instruction) > 1 and instruction[-2] == "side":
if sideset_count == 0:
raise RuntimeError("No side_set count set")
sideset_value = int(instruction[-1], 0)
if sideset_value >= 2**sideset_count:
raise RuntimeError("Sideset value too large")
delay |= sideset_value << (5 - sideset_count - sideset_enable)
delay |= sideset_enable << 4
instruction.pop()
instruction.pop()
if instruction[0] == "nop":
``` the pioasm sideset instruction check pops the instruction empty, causes the next nop check to out of indexadd if instruction: before next if block
tiny peak
Would you mind moving this to <#334905778857050125> ? This channel is mostly for...
yeah, sure, I can mention this over in show and tell... However, I was thinking this might be of interest to people ( @ jepler ?) who are following the ongoing dev work on usb_video, such as https://github.com/hathach/tinyusb/issues/2439
The point being, if library developers need a way to get video or bitmaps from a board to a host computer, this is a technique that would work
onyx hinge
```python
if len(instruction) > 1 and instruction[-2] == "side":
...
I am not 100% sure I correctly interpreted your message, but I created a Pull Request to improve the diagnois of the incorrect pio assembler line side 1: https://github.com/adafruit/Adafruit_CircuitPython_PIOASM/pull/65 -- if you have anything to comment, add or clarify, please do!
ornate breach
yeah, sure, I can mention this over in show and tell... However, I was thinking ...
You’re fine, I just didn’t want it getting swallowed up by GitHub activity posts in this channel.
tiny peak
You’re fine, I just didn’t want it getting swallowed up by GitHub activity posts...
thanks!
tulip sleet
@slender iron rc.0 has only been cooking for about 18 hours, but maybe it's just time to release 9.1.0 final. If I don't do it today might want to wait until next week, and, well, why?
slender iron
fine with me! Can talk it up today
tulip sleet
yah, exactly, on S&T
onyx hinge
>>> s.connect(("fd5f:3f5c:fe50:0:6d9:f5ff:fe1f:ce10", 22))
>>> s.setblocking(False)
>>> s.recv(100)
b'SSH-2.0-OpenSSH_9.2p1 Debian-2+deb12u3\r\n'
I'm poking at mdns now
onyx hinge
recvfrom doesn't succeed in showing the peer address ```>>> s.recvfrom(100)
(b'SSH-2.0-OpenSSH_9.2p1 Debian-2+deb12u3\r\n', None)
I don't think we have an ipv6 type for addresses yet
slender iron
@onyx hinge did you need to change anything to get an ipv6 address on the network?
onyx hinge
haha :egg on face emoji: that was from python3, not circuitpython.
but it really does resolve addresses
slender iron
😄
onyx hinge
>>> s.connect(("fd5f:3f5c:fe50:0:6d9:f5ff:fe1f:ce10", 22))
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
OSError: [Errno 22] Invalid argument
```here's what it really doesthat's what circuitpython does? at least you have something to fix 🙂
orchid basinBOT
slender iron
@tulip sleet this is the git setting https://speakerdeck.com/schacon/so-you-think-you-know-git?slide=69
atomic summit
Hey @tulip sleet have some good news and some bad news re 9.1.0 😉
Good news is the serial comms stuff is working great - my test jig can now write to each TinyC6 about 3x faster than before, without any fails, so big thanks for that.
The bad news is... specific to the C6 - WiFi connectivity is at least 2x slower than in 9.0.5, if it works at all.... boo hoo 😦
For context - on my testjig process, for every board I ship, part of my test is to connect to my test wifi AP rig that is running 3x WiFi AP's in MicroPython, one on channel 1, one on 6 and one on 10.
and every board scans for all of those APs and then connects to each, one at a time, recording the connection time for each - my hard cap for any board is 4sec for any channel. If it takes longer I'll retry once only, per channel, and then fail the board.
Every TinyC6 I have in stock that was flashed with 9.0.5 during the test, passed the stock standard test process - and every board I am now re-flashing with 9.1.0 (so I don't ship out boards that people can't communicate with) fails every wifi test on every channel.
So I tested a TinyS3, FeatherS3 and ProS3 with 9.1.0 and all pass the wifi test, for every channel in < 4 secs
Here's an example of what's happening on a TinyC6
DUT: Found UM-C6, channel 6, RSSI -60
DUT: Found UM-C1, channel 1, RSSI -49
DUT: Found UM-C10, channel 10, RSSI -70
DUT: Testing connections to the following AP"s
DUT: [("UM-C6", 6), ("UM-C1", 1), ("UM-C10", 10)]
DUT: Connecting to UM-C6 on channel 6
DUT: wifi connected to UM-C6 (192.168.4.14) in 7.318 secs
DUT: Connecting to UM-C1 on channel 1
DUT: wifi connected to UM-C1 (192.168.4.14) in 5.846 secs
DUT: Connecting to UM-C10 on channel 10
DUT: wifi connected to UM-C10 (192.168.4.14) in 14.821 secs
DUT: ... took too long, retry #1
DUT: \x1b]0;\xf0\x9f\x90\x8dWi-Fi: off | BLE:CIRCUITPY516b42 | code.py | 9.1.0\x1b\\\x1b]0;\xf0\x9f\x90\x8dWi-Fi: No IP | BLE:CIRCUITPY516b42 | code.py | 9.1.0\x1b\\Connecting to UM-C10 on channel 10
DUT: wifi connected to UM-C10 (192.168.4.14) in 5.818 secs
I'm very sure this will be an IDF issue as the C6 uses WiFi 6, so it would have a different IDF stuff than other ESP32's.
CP is using IDF 5.2.1 correct? I'll have a dig through changes in 5.2.2 and 5.3 to see if there's been anything that addresses wifi performance issues specific to the C6.
tulip sleet
I'm very sure this will be an IDF issue as the C6 uses WiFi 6, so it would have ...
Hmm - sorry to hear that. 9.1.0 is using ESP-IDF 5.2.2.
Do you have a Espressif C6 dev board, like ESP32-C6-DEVKitC-1-N4 or N8? Does it show the same problem?
I suggest a simple thing to try on the TinyC6, which is to reduce the default tx power, using a new .mk constant that was added, for instance here: https://github.com/adafruit/circuitpython/pull/9369/files. The default is 20. Try 10 to start and if that helps, bump it up a little at a time Reducing the power fixed connection problems on various C3 boards that had possible antenna mismatches, maybe due to minimal keepout or matching network choices.
onyx hinge
meanwhile https://github.com/espressif/esp-idf/releases/tag/v5.3-rc1 v5.3 has reached release candidate stage
tulip sleet
I am going to work on merging MicroPython updates, and updating to v5.3 could go along with that.
tulip sleet
@slender iron @onyx hinge I have not made a 9.1.x branch yet, because I thought near future changes might still be on main as we make 9.1.1, etc. But if you want to add ipv6 stuff for matter in prep for a 9.2, then we can split 9.1.x off right away.
also i would certainly put micropython merges and ESP-IDF upgrades in 9.2
onyx hinge
I think it's fine to hold off branching for now
orchid basinBOT
slender iron
I agree. No need to branch yet
tulip sleet
@slender iron I thought you mentioned someone who was trying to get microdot to work on CircuitPython. Who was that? (this is re https://github.com/adafruit/Adafruit_CircuitPython_asyncio/issues/35)
slender iron
I think it was the person who created microdot
slender iron
haha, yup. just got there too
tulip sleet
i can certainly fix the u stuff
tulip sleet
i'll look at updates to asyncio from upstream as I do the merge
slender iron
it'd be good to support it
atomic summit
Hmm - sorry to hear that. 9.1.0 is using ESP-IDF 5.2.2.
Do you have a Espressi...
I’ll try it, but it’s not a RF mismatch or HW design issue. If it was, it would happen in 9.0.5 and in MP and Arduino, which it doesn’t. Only happening on CP 9.1.0.
tulip sleet
I’ll try it, but it’s not a RF mismatch or HW design issue. If it was, it would ...
probably not, but I was just wondering if 5.2.2 changed something about the radio handling
is your settings.toml empty?
or not there?
atomic summit
probably not, but I was just wondering if 5.2.2 changed something about the radi...
Yeah, it’s definitely an IDF thing. But so far I’ve not found anything obvious that stands out.
atomic summit
or not there?
Not there. I don’t add one on my boards. Never got around to looking into it tbh
slender iron
@atomic summit probably worth doing a git bisect
atomic summit
<@195765176035835904> probably worth doing a git bisect
👍 lol, no idea what that is. I’ll look it up when I get into the office.
slender iron
👍 lol, no idea what that is. I’ll look it up when I get into the office.
Its a way to figure out which commit broke you
slender iron
@onyx hinge you around?
is it weird to have a class that abstracts onto an underlying buffer instead of just parsing it into variables?
onyx hinge
@slender iron I'm not working today
atomic summit
probably not, but I was just wondering if 5.2.2 changed something about the radi...
Ok, in the office now and pulling latest CP. I'll let you know how tests go once I've built some firmware with reduced TX power.
atomic summit
Ok, the times are a little less erratic with TX power of 15 (only tested 3 boards so far), but connections are still 2-3 seconds slower than previously. They'd still all fail my standard tests 😦
DUT: Connecting to UM-C6 on channel 6
DUT: wifi connected to UM-C6 (192.168.4.3) in 5.818 secs
DUT: Connecting to UM-C1 on channel 1
DUT: wifi connected to UM-C1 (192.168.4.3) in 5.82 secs
DUT: Connecting to UM-C10 on channel 10
DUT: wifi connected to UM-C10 (192.168.4.3) in 5.826 secs
Here's my same test running on a TinyC6 with 9.0.5
DUT: Found UM, channel 1, RSSI -50
DUT: Found UM-C1, channel 1, RSSI -56
DUT: Found UM-C10, channel 10, RSSI -57
DUT: Testing connections to the following AP"s
DUT: [("UM-C6", 6), ("UM-C1", 1), ("UM-C10", 10)]
DUT: Connecting to UM-C6 on channel 6
DUT: wifi connected to UM-C6 (192.168.4.4) in 3.414 secs
DUT: Connecting to UM-C1 on channel 1
DUT: wifi connected to UM-C1 (192.168.4.4) in 3.417 secs
DUT: Connecting to UM-C10 on channel 10
DUT: wifi connected to UM-C10 (192.168.4.4) in 3.417 secs
There seem to be quite a few changes in wifi/wifi MAC (and all radio stuff) in 5.2.2, including PHY changes 😦
It could be one or many of these things or none of them. Way beyond my understanding 😦
Is there someone at Espressif we can ping to ask them about this issue and what might be causing it?
onyx hinge
their public tracker is our main point of contact, I think
atomic summit
their public tracker is our main point of contact, I think
hmm, ok, that's sad considering the influential level of this project.
tulip sleet
There seem to be quite a few changes in wifi/wifi MAC (and all radio stuff) in 5...
I will do some diff'ing between 5.2.1 and 5.2.2, and look at the release notes. If you re-created the problem with one of the esp-idf/examples: "look here, my connect on C6 is taking 3 times as long", I think that would bea succinct reproducible case to open an issue in esp-idf about
and if you can't reproduce, then there's maybe some sdkconfig value, new or old, that might need changing in CircuitPython, etc.
i looked at the 5.3-rc.1 release notes and I didn't see anything that might be a bug fix for this
tulip sleet
hmm, ok, that's sad considering the influential level of this project.
it might be helpful to make a C6 debug build and see what's printed out in the log as the wifi connection is tried
atomic summit
it might be helpful to make a C6 debug build and see what's printed out in the l...
How do I do a debug build again? sorry, been a while 😉
tulip sleet
make BOARD=... DEBUG=1
atomic summit
oh, lol, sorry.
tulip sleet
you may need to disable some things to make room, e.g. CIRCUITPY_ULAB = 0
do a clean build
atomic summit
been in HW space for a long time - the SW side get's stomped 😉
tulip sleet
are you adept at building ESP-IDF examples? I have to keep reviewing how to do that
atomic summit
are you adept at building ESP-IDF examples? I have to keep reviewing how to do t...
Unfortunately yes, but not at writing code in IDF - simulating what I do on the test jig in IDF land will tax me, hahaha
tulip sleet
i looked at recent C6 wifi issues - I don't see anything relevant
atomic summit
i looked at recent C6 wifi issues - I don't see anything relevant
Thanks for looking
agile tinsel
i really hope that the next update of circuitpython will be more optimized
stuck elbow
i really hope that the next update of circuitpython will be more optimized
optimized for what?
shadow vigil
new PR -- add I2C support for SH1106 displays -- https://github.com/adafruit/Adafruit_CircuitPython_DisplayIO_SH1106/pull/19
shadow vigil
and while i was at that https://github.com/adafruit/Adafruit_CircuitPython_DisplayIO_SH1106/issues/20 (i would tackle that one myself but i'm a little lost in the buffering code at the moment)
manic glacierBOT
onyx hinge
I didn't successfully get my esp32s2 to claim an ipv6 address (even though my usual wifi AP offers them), or else my newly written routine common_hal_wifi_radio_get_ipv6_addresses doesn't work. next week I'll try to get useful debug prints and see if I can see whether the dhcp6 client is running. Have a good weekend all!
manic glacierBOT
onyx hinge
Pico W doesn't even have the "internal watchdog timer", does it? It has a watchdog that can be enabled from Python code but the code shown doesn't enable it, and anyway the microcontroller.watchdog doesn't reboot into safe mode, it reboots normally
kind river
I'm starting to look at adding openthread as a network interface. My current plan is to pretty much just copy the wifi module and replace the implementation and auth api with openthread
manic glacierBOT
onyx hinge
I'm starting to look at adding openthread as a network interface. My current pla...
@slender iron and I have recently been looking at Matter with the aim of supporting it someday in CircuitPython. I have been working on adding IPv6 support to our Espressif wifi bindings, though I haven't yet gotten it to DHCP6, let alone exchange packets on the network. Thread support is also something we'd like eventually. fwiw we're focusing on Espressif family microcontrollers for now. I'm sure Tannewt would be happy to chat with you at some point about this! I don't know how openthread relates to what's on espressif but I see the URL of the doc says "openthread" https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-reference/network/esp_openthread.html
kind river
<@252717193496756235> and I have recently been looking at Matter with the aim of...
I was inspired the the Matter deep dive. That's what I'm looking at now. I'm not sure if it make more sense the just use Espressifs openthread wrapper or "stock" openthread that could run on any chip with either an internal radio or external radio co-processor
onyx hinge
Agreed. For instance, with the ipv6 over wifi stuff I'm faced with potentially implementing it twice (once on espressif, once on rp2040) since the espressif and rp2040 use different APIs. But espressif is using lwip under the covers, so it might be possible to do similar to what we did with ssl and merge the SocketPools of the two systems to one set of lwip code...
(for ssl both were using mbedtls, but the espressif side was going via espressif wrappers)
kind river
I'm leaning toward bare openthread as that would also allow development using the Unix port and a RCP dongle
For now, I'm just working on a module API definition
manic glacierBOT
BTW, this is what the code looks like when it's running. A view of the web page on my iPhone. Nothing fancy at all because I stripped out a lot for this test code.
manic glacierBOT
kind river
From what I've read, it looks like the right approach would be a shared-bindings module that wraps the standard openthread api and calls a port specific radio init
manic glacierBOT
manic glacierBOT
kind river
@onyx hinge @slender iron My plan for openthread is to have a shared-bindings module that wraps the standard openthread api and a common-hal module that wraps the radio and potentially port specific socketpool (eg. Espressif integrates openthread with lwip)
onyx hinge
It'd be grand if someone else could review https://github.com/adafruit/Adafruit_CircuitPython_DisplayIO_SH1106/pull/19 -- the reviewer author & I got off on the wrong foot and I'm wary of writing more.
shadow vigil
It'd be grand if someone else could review https://github.com/adafruit/Adafruit_...
@onyx hinge i have no problem with your amended comments! 😀
onyx hinge
<@426200547024961539> i have no problem with your amended comments! 😀
thanks, it's good to hear that.
lone axle
Has anyone made it through the setup process for pre-commit as it's used by the libraries under Ubuntu 24.04 or python 3.12? During the pylint environment setup step it crashes out with a stack trace ending in AttributeError: module 'pkgutil' has no attribute 'ImpImporter'. Did you mean: 'zipimporter'?
I understand is a change with python 3.12, but I'm unclear on the right resolution for usage with pre-commit. I'm going to go digging, but thought to mention it here in case someone has already worked through this.
spare jacinth
which repo is the pre-commit config targetting?
lone axle
Ah indeed, a few folks have made it through. Looks like there are some solutions to try within this and the linked issues. https://github.com/pre-commit/pre-commit/issues/3041
spare jacinth
which repo is the pre-commit config targetting?
as in... perhaps it doesn't support 3.12 (yet?)
lone axle
which repo is the pre-commit config targetting?
I'm not certain tbh, whatever is default within the libraries / cookiecutter.
spare jacinth
according to the issue, it is indeed an issue with Py's vs libraries's versions incompatibility:
you have a too-old version of virtualenv which is pulling in an old setuptools which did not support 3.12
lone axle
I think this one is related to pylint and our libraries as well: https://github.com/adafruit/workflows-circuitpython-libs/issues/31#issuecomment-2084152441
Looks like it's probably time to consider upgrading the version of pylint that we're pinned to.
spare jacinth
side note: might be worth looking into using ruff instead of pylint, it runs quite faster and has some more features
lone axle
ruff has been discussed a little bit ago and some of the libraries have PRs that illustrate the changeover to it, but I think it hasn't had much more movement since.
Definitely worth circling back to that discussion while on the topic of potentially upgrading the pylint version.
onyx hinge
ruff has been discussed a little bit ago and some of the libraries have PRs that...
specifically, pip install git+https://github.com/pycqa/pylint@v2.17.4 (which is what pre-commit does internally) does not work starting in Python 3.12. Updating pylint across libs would be great, as would taking the leap to ruff.
without tuning the list of included & excluded tests, ruff defaults to doing much less than pylint, so if we don't have a lot of time to do it right we should probably bump pylint instead
the last in the 2.17 series, 2.17.7, has the same problem
spare jacinth
i think there is something like extend = ["ALL"] for ruff's config file, but running every check on each project sounds like an overkill 😛
another good thing is that it can also format files (eg could replace black too)
onyx hinge
2.8.3 and newer work
@spare jacinth you CAN enable all but some rules even conflict with each other and some are way too opinionated
spare jacinth
oh, didn't know they conflict themselves, that's... fun
onyx hinge
since pre-commit requires that ALL diagnostics be resolved, we have to 100% whatever checks we enable
warning: `one-blank-line-before-class` (D203) and `no-blank-line-before-class` (D211) are incompatible. Ignoring `one-blank-line-before-class`.
warning: `multi-line-summary-first-line` (D212) and `multi-line-summary-second-line` (D213) are incompatible. Ignoring `multi-line-summary-second-line`.
```at least these two pairs are contradictorySounds good! The other thing to consider for the API is how it relates to the socket API. Ideally we'll be able to drop it in as a replacement.
manic glacierBOT