But if it wasn't connected it would wait 5 seconds before continuing...
#circuitpython-dev
short tendon
crimson ferry
Yeah, even if that gets separated out into a separate connect, I wonder if a smaller max timeout could be used
is there a better place to discuss library design? It doesn’t seem to fit really here or in #help-with-circuitpython
short tendon
Yeah, I struggle with where also. I usually do here, or open up a PR...
alpine holly
I'm toying with adding 4-wire SPI SD-Card (SDIO/MMC) support for esp32 - i.e. implementing sdioio library for esp32. I have the tech know-how to make it work, but no "administrative" understanding of what I need to do inside my "ports/espressif/boards/ai-thinker-esp32-cam" folder to cause this lib to be included, and what else I need to do to generally make this work available for anyone else with esp32... is there anyone reading this who can either lay out the steps I need to take, or point me to someplace that might explain it?
wraith crow
I had toyed with looking at 1-wire and opened an issue https://github.com/adafruit/circuitpython/issues/8498 I don't know if they are relevant but there were a couple of info links I had run across
onyx hinge
I'm toying with adding 4-wire SPI SD-Card (SDIO/MMC) support for esp32 - i.e. im...
it'd be grand to have SDIO support on espressif family microcontrollers. Modules are enabled by putting CIRCUITPY_SDIOIO = 1. This can be in the board's mpboardconfig.mk, like ports/atmel-samd/boards/same54_xplained/mpconfigboard.mk:CIRCUITPY_SDIOIO = 1 or at the port level as in ports/cxd56/mpconfigport.mk:CIRCUITPY_SDIOIO = 1.
(make BOARD=whatever clean after changing those .mk files)
tulip sleet
<@&356864093652516868> and anyone else who's interested: The weekly CircuitPython meeting will start in about two hours, at 2 pm US Eastern time. If you're planning on submitting notes but haven't written them up yet, here's the notes document: https://docs.google.com/document/d/19GmXK8VzVgchFclJTfG4C6ejqNsg9emYgCgpcv1PVZ4/edit?usp=sharing
alpine holly
@wraith crow @onyx hinge - excellent help and references - thanks heaps.
onyx hinge
The Internet here is not good enough to support me joining with voice but hello from pycon! I'll be back next week.
slender iron
panic is the company
lone axle
turbid radish
It does help
manic glacierBOT
lone axle
That sounds like an AWESOME idea for a hackathon challenge!
lone axle
if you have or are willing to make a simplified reproducer that illustrates that issue (removing some code alowing it not to crash) it would be good to post it in an issue on the nina-fw repo perhaps.
silver tapir
Thanks for hosting Dan.
lone axle
Thanks for hosting Dan. Hope everyone has a great week!
slender iron
500 boards!
tulip sleet
Here is the notes document for next Tuesday's CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord, but is a day later than usual due to a US Monday holiday. 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/1EvRvqSdQW_nMZFkgvgHDKlqHlMl08z2QehnGhNohHSA/edit?usp=sharing
short tendon
if you have or are willing to make a simplified reproducer that illustrates that...
This is the ticket opened over the weekend with the issues: https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT/issues/216
opal crystal
I see I was mentioned with my Playdate project – thanks @turbid radish ! Just one addition to what's mentioned in the notes doc (although as far as I see the newsletter has gone out already): I’m better reachable on Mastodon than X, https://mstdn.social/@isziaui.
scenic ibex
Heya! I was told it'd probably make more sense to post things over here vs #help-with-circuitpython , so I'll repost a few bits about what I'm trying to do:
Short version is I've got a custom board I've made that is super similar to the badger2040w in design, which I want to use circuitpython for (and previously, I've been successful in just using the prebuilt badger2040w distribution for it, I sorta designed it to be basically pin compatible for it), but due to some supply issues, I'd like to switch to using a SSD1680 based display vs the UC8151D that the Badger2040w and the past revision of my board uses, along with a few other more obvious tweaks. I've found a few sources and am prepared to dive into mountains of other datasheets and code to figure it out ( Waveshare's Arduino Driver and Docs being a primary resource, along with the official adafruit_SSD1680 circuitpy library, and the SSD1680 docs themselves ) , but figured I'd ask around to see if someone's done similar stuff before that I can use as a base.
I have stumbled across the Czech Maker's Maker Badge: https://circuitpython.org/board/maker_badge/ , which uses a SSD1680 based display, but in the firmware they just instantiate things on the python side vs in board.c, meaning the display doesn't show up as a "built in" part of things. I'm a bit unsure of how it works, but I assume that means it doesn't display nice error/crash messages on the screen like it would if it was properly setup/initialized in board.c, so I'd like to try and figure it out if at all possible.
tulip sleet
Heya! I was told it'd probably make more sense to post things over here vs <#537...
The arguments to the display initialization in board.c are pretty much the same as in the Python code, so you can probably copy an existing board (maybe even the badger), and replace those args (and fix anything else that's different). Also check the other files in the board directory.
lone axle
It seems that the Build and Validate Stubs task inside of the docs job in actions failed for the release of 9.0.4 https://github.com/adafruit/circuitpython/actions/runs/8709993694/job/23890925410 This resulted in pypi not getting a 9.0.4 version of circuitpython-stubs so it's still showing 9.0.3 is the latest stable release at https://pypi.org/project/circuitpython-stubs/
At the very end of the log it seems to indicate success building the stubs, but then immediately afterward it fails to find the file that would have been built.
Successfully built circuitpython_stubs-9.0.4.tar.gz and circuitpython_stubs-9.0.4-py3-none-any.whl
make[1]: Leaving directory '/home/runner/work/circuitpython/circuitpython'
WARNING: Requirement 'circuitpython-stubs/dist/circuitpython-stubs-*.tar.gz' looks like a filename, but the file does not exist
Processing ./circuitpython-stubs/dist/circuitpython-stubs-*.tar.gz
ERROR: Could not install packages due to an OSError: [Errno 2] No such file or directory: '/home/runner/work/circuitpython/circuitpython/circuitpython-stubs/dist/circuitpython-stubs-*.tar.gz'
make: *** [Makefile:281: check-stubs] Error 1
Error: Process completed with exit code 2.
All 3 of the 9.1.0 beta releases did complete their actions successfully and upload artifacts to pypi successfully so those are available as pre-release versions. beta.1 and beta.2 both were released after 9.0.4 as well
tulip sleet
At the very end of the log it seems to indicate success building the stubs, but ...
yes, this build problem was fixed post 9.0.4, but I can't regenerate the stubs easily. I could just copy them to 9.0.4 bit 9.0.3 is good enough
midnight ember
Heya! I was told it'd probably make more sense to post things over here vs <#537...
Do you have a github repo with your current efforts for your board? That would be easier for someone to take look and notice any obvious changes that need to be made assuming you already have a correct pins file.
scenic ibex
Do you have a github repo with your current efforts for your board? That would ...
Not yet, but I'm currently hacking away at things and will give it an upload soon. I need to do a once over on pins.c still, but it should be more or less correct, at least for the epaper display
I think I've been able to parse a good bit of it and at least understand what I'm looking at more, but some of the LUT config options are still going over my head at the moment
midnight ember
Sounds like you're headed in the right direction. Because the init sequence is already done in the library that's actually one of the hardest things to figure out and it's already been done. The build configs might take a lot of trial and error so don't get discouraged, keep at it.
scenic ibex
it seems like some drivers I've found do set the LUTs explicitly, but the more adafruit/circuitpython adjacent ones don't (using default settings probably?)
midnight ember
I don't have any experience with einks but just glancing over the maker badge and pybadge they're different enough from a TFT where I can't offer any solid advice on the build config. I'm used to working with TFT's. There are many eink builds in ports/boards to look into for examples even if they're slightly different from what you might need, looking into other examples helps.
scenic ibex
Yeah, I've been scouring a bunch of different examples and libraries so far, and they've been pretty helpful
midnight ember
Any boards from adafruit with ssd1680 can be found by searching the store https://www.adafruit.com/search?q=ssd1680 and then you can probably find them in ports/boards. The closest example is probably the Pimoroni InkyPhat but I couldn't find the port for it, seems like it's for the Raspberry Pi only.
Well, it might have been the closest example if there was a port for it. Maybe it's in ports but I couldn't find it.
tulip sleet
it seems like some drivers I've found do set the LUTs explicitly, but the more a...
the settings in the CircuitPython Python library supposedly should be fine
scenic ibex
Do you have a github repo with your current efforts for your board? That would ...
https://github.com/ben-hawks/circuitpython/tree/main/ports/raspberrypi/boards/aivillage_dc32_badge I went ahead and pushed stuff to a repo finally. It builds successfully and I get a REPL no problem! That being said, there are some issues with the display, it seems to be able to connect and refresh, but shows basically garbage on the screen, even with a simple test script
stuck hatch
Hello.
Sorry, not sure if's the right place to ask, but thought maybe asking here first before filling github issue would be better. 🙇♂️
I am using Circuitpython 9.0.4 on T-Display S3, to have multiple DIY air quality sensors. Along with air metrics I also measure voltage on USB/battery like voltage_in = analogio.AnalogIn(board.BATTERY) and have issue with analog reading being "stuck" after some hours of use (3~6 hours) on some random(?) high value.
>>> import analogio
>>> import board
>>> voltage_in = analogio.AnalogIn(board.BATTERY)
>>> voltage_in.value
61761
>>> voltage_in.value
61761
>>> voltage_in.value
61761
>>> voltage_in.value
61761
It either goes back to normal after few hours, or after fully disconnecting from power and restarting device again. microcontroller.reset() from code does not resolve the issue. Same issue happens on multiple boards (tested on 8 different boards).
I was wondering if it is the hardware design issue with T-Display S3, or could be related to circuitpython? I do not remember this issue happening before when I used c++ on Arduino IDE for same boards (but I have not tested it extensively back then so could be wrong).
midnight ember
https://github.com/ben-hawks/circuitpython/tree/main/ports/raspberrypi/boards/ai...
That is a lot to configure. Someone with eink experience will have to answer that one.
scenic ibex
That is a lot to configure. Someone with eink experience will have to answer th...
Yeah 😅 I'm sure its a matter of finding the right settings, things seem to be "working" hardware wise (I've seen and had some hardware failures before with this board, and the eink fails different for hardware errors in my experience), but Just something config/driver wise
scenic ibex
Yeah 😅 I'm sure its a matter of finding the right settings, things seem to be ...
This is currently what i get with what's committed to the repo
scenic ibex
Yeah, so I can confirm the board works hardware wise with the screen - if I initialize it in python using the circuitpython lib, things work file with this example: https://github.com/adafruit/Adafruit_CircuitPython_SSD1680/blob/main/examples/ssd1680_four_corners.py
but on boot as things are configured in board.c , things are still broke
manic glacierBOT
tulip sleet
Hello.
Sorry, not sure if's the right place to ask, but thought maybe asking he...
when the battery is fully charged, what would be a typical value?
stuck hatch
when the battery is fully charged, what would be a typical value?
around 4.5 (recalculated for the graph attached to message, I did not logged the raw values) but I have not encountered this issue on battery (perhaps because battery runs out faster than it takes to trigger the behavior). The pic above is when all of the devices connected via usb.
tulip sleet
so without it being stuck (like being first plugged in), if you did the REPL code above, what raw value would you see?
there is battery charging circuitry on the board, as I think you know
if you unplug the battery, is it still stuck?
stuck hatch
just tested
>>> import analogio
>>> import board
>>> voltage_in = analogio.AnalogIn(board.BATTERY)
>>> voltage_in.value
45596
>>> voltage_in.value
47244
>>> voltage_in.value
47185
>>> voltage_in.value
47165
the ones that stuck run without battery, so basically measuring usb voltage
tulip sleet
is that the same case in what you just posted?
we do a pretty thorough reset of the ADC when the AnalogIn is created. You could try CircuitPython 9.1.0-beta.2, which has a newer version of the underlying Espressif SDK (ESP-IDF 5.2.1)
stuck hatch
just tested
```
>>> import analogio
>>> import board
>>> voltage_in = analogio.A...
I just posted measurements on T-Display S3 with fully charged battery (and usb connected)
ok, thanks, I will try flashing 9.1.0 beta and see if issue occurs again after some hours
tulip sleet
ok, thanks, I will try flashing 9.1.0 beta and see if issue occurs again after s...
another couple of ideas if it reproduces on 9.1.0:
- measure the voltage at the battery pin itself, just to check that there's not some hardware problem that is causing the charging pin to go too high.
- The BATTERY pin is GPIO4, which uses ADC1. Try measuring the voltage on another ADC1 pin and see if it is stuck too. Ground that other pin, measure again, and then check the BATTERY pin again. See http://wiki.fluidnc.com/en/hardware/ESP32-S3_Pin_Reference for which pins use ADC1.
slender iron
This is from Python? The c code should be analogous to the Python code so there must be a mismatch somewhere
scenic ibex
This is from Python? The c code should be analogous to the Python code so there ...
Yeah, I think there's something different, but I'm just trying to parse and figure out where the issue is
slender iron
Yeah, I think there's _something_ different, but I'm just trying to parse and fi...
I'll take a look at your board code
scenic ibex
I'll take a look at your board code
That would be incredible, thank you!!
slender iron
That would be incredible, thank you!!
I think it is your ramxcount and ramycount
the python library patches those values
you can hard code it for your board
scenic ibex
I think it is your ramxcount and ramycount
Ah, I think I haven't updated the repo yet, I did hardcode those values and its still failing :(
Let me push my updates really quickly
scenic ibex
I think it is your ramxcount and ramycount
Oh wait, its patching the ramx and ramycount values? I didnt realise, I thought it was patching the Driver_control values
slender iron
I think it is
you can change the driver to print out start_sequence 🙂
maybe display size?
scenic ibex
you can change the driver to print out `start_sequence` 🙂
Went ahead and did that, this is what the python driver's start sequence is:
Start Sequence:
\x12\x80\x14\x11\x1\x3\x3c\x1\x5\x2c\x1\x36\x3\x1\x17\x4\x3\x41\x0\x32\x4e\x1\x1\x4f\x2\x0\x0\x1\x3\x27\x1\x0\x22\x1\xf4
it does look like it's modifying the "driver control" values vs the ramx/ramycount values
I did go ahead and update the repo to the current code I'm using, though its changed a rreasonable bit since I've been trying a bunch 😅
short tendon
okay, is something broken with exceptions? This:
from adafruit_minimqtt.adafruit_minimqtt import MMQTTException
l = None
def x():
raise MMQTTException("test 1")
for i in range(5):
try:
x()
except MMQTTException as e:
l = e
raise MMQTTException("test 2") from l
only gives the traceback:
Traceback (most recent call last):
File "<stdin>", line 14, in <module>
MMQTTException: ('test 2', None)
Where as this:
l = None
def x():
raise Exception("test 1")
for i in range(5):
try:
x()
except Exception as e:
l = e
raise Exception("test 2") from l
gives (as I expect):
Traceback (most recent call last):
File "<stdin>", line 6, in <module>
File "<stdin>", line 3, in x
Exception: test 1
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "<stdin>", line 9, in <module>
Exception: test 2
I'd suggest you test the same thing with MicroPython.
short tendon
I'd suggest you test the same thing with MicroPython.
remind me which version we match?
tulip sleet
we most recently merged from MicroPython v1.20, last october
short tendon
we most recently merged from MicroPython v1.20, last october
MP 1.20 Warning: exception chaining not supported
tulip sleet
1.22 is probably the same, Jeff enhanced what was there
short tendon
It is, just checked
If I file a bug (which I will), do you think it would get into 9.1.0?
tulip sleet
it's not a bug, it's a deliberate deficiency, and it's not so easy to fix
short tendon
ahh
spare jacinth
If I file a bug (which I will), do you think it would get into `9.1.0`?
depends on how much it gets you to write a PR? :p
if jeff didnt do it back there, there's probably a good reason for it
whether it was about time to write it, code complexity/mainteinance, flash/ram consumption, etc... i dont know
tulip sleet
we really depend on MicroPython for the core langauge stuff. We have added a small amount of extra stuff, but not much. We don't want to diverge that much
short tendon
If I were to put a comment in code, why I'm not doing a raise/from what could I add that would make sence?
tulip sleet
short tendon
Does this make sense:
# MicroPython doesn't support exception chaining
# and CircuitPython can't chain on non-base exceptions
I'm not sure whether from ... is not implemented as you wish or what the issue is. You could file an issue but it would be long term. It might have to do with native vs non-native
@onyx hinge could speak to this when he is back. He is returning from PyCon today.
spare jacinth
we really depend on MicroPython for the core langauge stuff. We have added a sma...
given this is about making the feature better, i dont think it goes any further from the current they-dont-have-it state
but may still be quite a bit of work to get it working, and perhaps not too useful anyway...
onyx hinge
I'm not actually back to work until the 28th
spare jacinth
Does this make sense:
```py
# MicroPython doesn't support exception chaining
# a...
what about raise Base from Sub and raise Sub from Base?
tulip sleet
the spirit speaks 🙂
onyx hinge
tests/circuitpython/traceback_test_chained.py in the source tree and its associated .exp file shows what we test
spare jacinth
what about `raise Base from Sub` and `raise Sub from Base`?
and of course then cast one of them, eg something like: base = Exception(subclass.args) or the like
short tendon
So the issue is (as you can see from the code sample) an issue where we don't raise a helpful exception
spare jacinth
what about `raise Base from Sub` and `raise Sub from Base`?
as in... maybe you can get it to work if one of them is not MMQTTException but Exception and you could recover part of its contents/information
short tendon
Yeah, for now I'm just going to add the other one to the exception string. I'll let Dan help me figure it out
onyx hinge
@spare jacinth looks like some kind of bug in the implementation. you can file an issue. a reduced test case looks like it could be ```py
class C(Exception):
pass
try:
raise C("inner")
except Exception as e:
raise C("outer") from e
will let Justin make the issue as it was not me who ran into the problem
this is on a few-months-old build of 9.0, but should be good for reference: ```py
class C(Exception):
... pass
...
raise C("inner") from C("outer")
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
C: inner
raise C("inner") from Exception("outer")
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
C: inner
raise Exception("inner") from Exception("outer")
Exception: outer
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
Exception: inner
raise Exception("inner") from C("Outer")
C: Outer
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
Exception: inner
so the bug feels to be along the lines of:
code does
type(inner) == Exceptioninstead ofisinstance(inner, Exception)
(however that idea translates into the C API :p)
short tendon
I think it's more then that, since this works:
>>> raise ValueError("inner") from OSError("outer")
OSError: outer
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: inner
Will take a look at the code after dinner but will probably not get a fix done hahahaha
spare jacinth
I think it's more then that, since this works:
```py
>>> raise ValueError("inner...
Maybe looks at compile time (C-defined) classes, but not runtime (py-defined) ones, then
Time to start dumping data into github :p
spare jacinth
i dont think this even deserves to open an issue, leaving the message here: ```
~/circuitpython$ make fetch-all-submodules
python3 tools/ci_fetch_deps.py all
/home/elpekenin/circuitpython/tools/ci_fetch_deps.py:14: SyntaxWarning: invalid escape sequence '.'
version_str = re.search("([0-9].)[0-9]", version_str).group(0)
more info:
raise Exception("inner") from "Exception"
raises a Hard Fault
but:
raise MMQTTException("inner") from "Exception"
Doesn't
short tendon
WHATTTT!!!!!
short tendon
I'm not actually back to work until the 28th.....
Did he also invent the time machine?
spare jacinth
probably.
onyx hinge
pycon is sprints today so I'm at my computer and wasn't concentrating on anything anyway
short tendon
Well, thank you. I didn't even get the bug open yet
Will test as soon as the builds complete
manic glacierBOT
short tendon
@tulip sleet if the exception fix and esp32spi get into the next beta - we'll be in an awesome spot for networking
tulip sleet
<@329766224093249548> if the exception fix and esp32spi get into the next beta -...
it will - i have to merge from 9.0.x but I will do that
manic glacierBOT
lone axle
I have written out several new pages covering the usage of circup in the main learn guide for it https://learn.adafruit.com/keep-your-circuitpython-libraries-on-devices-up-to-date-with-circup/usage-new?preview_token=AHSc_2Exy7L12vsR-P3oqw This link goes to the new Usage page, it now has sub-pages for the different commands which are linked in the navigation. The new pages aren't published yet, but I think they're visible for a bit with the token that is in the URL.
If anyone is interested / willing to look over the new pages it is most appreciated. If the new pages are looking good we can make them published and swap out the old Usage page.
random junco
I have written out several new pages covering the usage of circup in the main le...
I’ll give it a read through first thing in the morning.
blissful pollen
I have written out several new pages covering the usage of circup in the main le...
Reads fine, nothing stood out for me. I did like how you gave a few common examples but felt like they didn't really show what I should expect as output in return.
E.g. for freeze you show circup freeze but just the command not what I may see in return.
Not a big deal or even a reason to not publish, but an idea for the future potentially.
Another thought was that you talk about connected and Web workflow, but maybe a note that (I'm assuming) BLE workflow isn't supported yet.
lone axle
Thank you!
I did think about adding an "Example Output" section to the commands as well. I can add that. I think it would be a nice addition. It doesn't necessarily need to show the output of all of the examples but at least the most basic usage output would be nice to see.
lone axle
Reads fine, nothing stood out for me. I did like how you gave a few common exam...
Correct about BLE workflow as well. It's not currently supported by Circup, but it could be added theoretically.
blissful pollen
And I agree you don't need output on all the commands, even one is nice. Or part of one just to show what to expect.
Again minor on freeze maybe a "why" you would use this? It is the only one I think people may wonder why. The rest are pretty self-explanatory
lone axle
good call. I'll also add a bit about why you'd use it.
short tendon
@lone axle I think a call out on update and maybe even an example for how it looks for a major update would be a nice add. And the fact that a major update could break code.
I also think a page on fonts could be helpful, since you need to add the bundle, and getting the naming right can be fun (like only certain font sizes are available)
pure spade
Hello there, I have a question about USB VID/PID. I am trying to submit my Circuitpython board support to the mainstream repo and I didn't noticed that my VID/PID in the mpboardconfig.mk was copy pasted from another board. So I started a request for PID at https://pid.codes but I am not sure the procedure will go through since my board is not open hardware yet. So my question is: Is there a VID/PID pair that can be used for any board?
austere acorn
Hello there, I have a question about USB VID/PID. I am trying to submit my Circu...
Nope, there isn't.
The answer is right there on the page you linked
workarounds, such as using other organisations’ VIDs without permission, or simply making up a VID and PID. These solutions make things worse for everyone, by damaging the assumption that a VID/PID combination is unique to a given device
stuck hatch
another couple of ideas if it reproduces on 9.1.0:
1. measure the voltage at the...
It is still reproducible in 9.1.0 beta2. I checked as you wrote and here are the results:
ADC1_0 GPIO1 board.IO1 stuck at 61761
ADC1_1 GPIO2 board.IO2 stuck at 61761
ADC1_2 GPIO3 board.IO3 first read 61761, next always 0. Reading any stuck GPIO resets next read back to 61761, 0 after that.
ADC1_3 GPIO4 board.BATTERY stuck at 61761
ADC1_4 GPIO5 board.LCD_RST in use
ADC1_5 GPIO6 board.LCD_CS in use
ADC1_6 GPIO7 board.LCD_DC in use
ADC1_7 GPIO8 board.LCD_WR in use
ADC1_8 GPIO9 board.LCD_RD in use
ADC1_9 GPIO10 board.IO10 same as GPIO3
if I short GPIO1/2 to ground, it reads 0, if unshort immediately back to 61761. Did not had multimeter with me today to measure exactly (sensors deployed not nearby), I will check next time what are the actual voltages.
random junco
@lone axle I think the new CircUp Learn Guide looks great - and I learned a bunch. I don't have much to add that others haven't already mentioned. Two tiny things - is it CircUp or Circup? It's inconsistent in the guide. And in the first sentence under Contributing, I'd change Repo to Repository, it matches the rest of the learn guide. Looks great, nice job.
lone axle
<@382939733107408897> I think the new CircUp Learn Guide looks great - and I lea...
Thank you. Originally it was CircUp but I've never really seen much usage of that stylization with the capital 'U'. In the process of re-writing the pages I've been trying to change all instances of it to "Circup" as a proper noun for it's name, or circup when used as a literal part of a command. I have started locally a branch of the repo with that change as well but it's not complete yet, I'll submit it as a PR soon though.
orchid basinBOT
manic glacierBOT
tulip sleet
@short tendon I restarted the build. I just couldn't read your mind 🙂 . Fine to do that -- just tell us in the PR.
short tendon
<@1152454697614069811> I restarted the build. I just couldn't read your mind 🙂 ...
After it was cancelled, it couldn't start up again
tulip sleet
I restarted the wrong thing. What's wrong with your build env, could you use help with that?
short tendon
Probably. I run it under WSL. I seldom need to build, so I haven't spent much time on it
I'll push my branch back up in a bit, and add comments in it
tulip sleet
I am about to do a 9.0.x branch merge PR.
short tendon
Sounds good. I'll just wait for the next beta
tulip sleet
https://blog.adafruit.com/2024/05/22/circuitpython-9-0-5-released/
- Fix exception chaining for non-built-in exceptions.
- Waveshare RP2040-GEEK: fix pin names to match board.
- circuitpython-stubs 9.0.5 is available on PyPi (9.0.4 was missing due to a build problem)
solar whale
I have been trying to use the espcamera module with a WaveShare ESP32S3 Pico board but every time I try to initialize the camera, it crashes into Safe Mode - memory access or instruction error https://circuitpython.org/board/waveshare_esp32_s3_pico/ https://www.waveshare.com/wiki/ESP32-S3-Pico I am using one of the OV5640 pico camera boards with it. Is there anything obvious about this board or its memory devices that might make it incompatible with espcamera?
FWIW this is the init code that is crashing whn the call to espcamera.Camera is made ``` def init(self) -> None: # pylint: disable=too-many-statements
self._i2c = busio.I2C(board.IO16,board.IO15)
self._camera_device = None
self.camera = None
def init_camera(self, init_autofocus=True) -> None:
"""Initialize the camera, by default including autofocus"""
print("Initializing camera")
self.camera = espcamera.Camera(
data_pins=[board.IO17,board.IO18,board.IO33,board.IO34,board.IO35,board.IO36,board.IO37,board.IO38],
pixel_clock_pin=board.IO14,
vsync_pin=board.IO11,
href_pin=board.IO13,
powerdown_pin=board.IO12,
reset_pin=board.IO39,
pixel_format=espcamera.PixelFormat.RGB565,
frame_size=espcamera.FrameSize.VGA,
framebuffer_count=2,
i2c = self._i2c,
)
@tulip sleet, I wonder how helpful something like https://github.com/adafruit/circuitpython/issues/9265#issuecomment-2125248117 would be as a CP library, that if it was installed ConnectionManager would automatically wrap the socket...
tulip sleet
<@329766224093249548>, I wonder how helpful something like https://github.com/ad...
I think this is a useful thing, but I don't think it needs to be added to ConnectionManager. It think setting it up to log should be manual for now, to make it clear what's happening, and to avoid extra code in ConnectionManager. It's the kind of thing we could put in a Learn Guide
tulip sleet
I have been trying to use the espcamera module with a WaveShare ESP32S3 Pico boa...
Could you file this as an issue and describe the versions you are using, and what's on the board (flash, PSRAM or not, etc.). Thanks.
short tendon
So the only code that would be added to CM, would be the import and if it works, the 2 liner if/then wrap
solar whale
Could you file this as an issue and describe the versions you are using, and wha...
will do.
short tendon
otherwise, you would need to wrap the socketpool getsocket which would be more complex
Just a thought...
I can definatly build out a playground note with this though (and wrapping the socketpool)
tulip sleet
So the only code that would be added to CM, would be the import and if it works,...
creating a wrapped socketpool makes sense. You have a socketpool that will wrap the underlyling socketpool.socket in a specified wrapping class. I am just leery of doing of doing the wrapping automatically if the library is present. That's also could be a nuisance because preventing it means you have to rename or remove the LoggingSocket library.
short tendon
makes sense. so something like:
pool = adafruit_connection_manager.get_radio_socketpool(radio)
logging_pool = adafruit_logging_socket.wrap_pool(pool)
logging_pool.log_send()
requests = adafruit_requests.Session(logging_pool)
Many vendors will give out PIDs for devices with their micro. Links are at the bottom of this page https://pid.codes/faq/
manic glacierBOT
tulip sleet
makes sense. so something like:
```py
pool = adafruit_connection_manager.get_rad...
yes, that makes sense
short tendon
yes, that makes sense
Cool. Would we want that in the main bundle, or should I make it for the community bundle?
tulip sleet
Cool. Would we want that in the main bundle, or should I make it for the communi...
community bundle for now, because we don't want to necessarily be responsible for suppot
tulip sleet
It is still reproducible in 9.1.0 beta2. I checked as you wrote and here are the...
how long did you have to wait before it got stuck? I would try to reproduce this locally, but I don't have that board. I hope it would be possible to reproduce otherwise. It does seem like something is going wrong with the ADC rather than the pin itself. I looked for similar issues in the esp-idf repo but didn't find any
stuck hatch
how long did you have to wait before it got stuck? I would try to reproduce this...
its kinda random, can happen anytime from 3~6 hours and then randomly fix itself or get broken again, but it happens on all 8 devices that i have. There are other stuff happening on devices, like wifi connection, sensor reading over I2C, etc. - not sure if it can be related. I will also try reproducing it isolated without any other additional code.
tulip sleet
thanks for trying that
just writing a loop to poll the value over and over might do something
manic glacierBOT
tulip sleet
@slender iron I made 9.1.0-beta.3 release notes. It includes your Espressif BLE work and all C file header changes, and the frozen modules update fixes a PyPortal issue. If you have time to approve https://github.com/adafruit/circuitpython/pull/9271 (update froze modules), I can make a beta.3 release.
orchid basinBOT
tulip sleet
https://blog.adafruit.com/2024/05/22/circuitpython-9-1-0-beta-3-released/
Initial implementation of BLE GATT server support on Espressif chips. Pairing/bonding is not yet implemented.
Update TinyUSB; includes USB host fixes.
Fix exception chaining for non-built-in exceptions.
scenic ibex
maybe display size?
I got the display initalization working! https://media.discordapp.net/attachments/342354513338040322/1242988849551114320/IMG_20240522_185216.jpg?ex=664fd713&is=664e8593&hm=0e23cd2cd4f70c3497c4651c7b62f29abec7b5549b6fd54bdf83ffd846abaddf&=&format=webp&width=682&height=910
The issue in the end was really dumb, it ended up being that busy_state in the ...epaperdisplay_construct was set to False from the previous display I was using, but the driver code for the SSD1680 sets it to True, I just missed it when reading/transferring the settings 🤦♂️
I also had to find that start_col was set to 8 for this board, which in set in the driver code only if it's not passed in kwargs, so I also missed that bit first time around
with both those issues solved, it seems to work like expected! I still have some cleanup and addtional changes to make to the board def, but i'll likely be opening a PR to add the board soon :D
short tendon
Something I've been playing with. Created a playground note for a wrapper to get the radio for a device. I can now run the same networking based code with no changes on ALL my CP devices!
https://adafruit-playground.com/u/justmobilize/pages/get-radio
feedback appreciated
stuck hatch
just writing a loop to poll the value over and over might do something
upd, I ran board with nothing attached and only AnalogIn read cycle for a day. From logs I can see it restarted a few times, but readings did not get above 60k like the other boards.
It looks like either i2c sensor or wifi or something else in code might affect that 🤔
I guess I will try by the method of elimination test what may cause it.
tulip sleet
upd, I ran board with nothing attached and only AnalogIn read cycle for a day. F...
was there a battery attached?
stuck hatch
no, but the same is in the other sensors that get stuck.
When it is stuck (while being connected with usb), it does not matter if I attach battery, power from usb, or both, it still stuck at same value over 60k
tulip sleet
that is pretty odd. Add back the wifi first, because the wifi uses ADC2 (thought that should not affect ADC1). If that does nothing, yes, try the I2C and disable wifi (so we are trying one thing at a time)
stuck hatch
on devices that get stuck, I run sensor on I2C with GPIO1 and GPIO2 which on ADC1, I suspect maybe that could be the issue?
tulip sleet
... maybe ... though that ADC channel is not activated, so in theory it should not make any difference.
stuck hatch
Thanks. I will try different ways to come up with minimal reproducible setup.
short tendon
@tulip sleet if I wanted to get some devices with less ram (have a m4, pico-w) would you recommend this one: https://www.adafruit.com/product/5323? Any others to make sure I have a good set to play with?
tulip sleet
<@329766224093249548> if I wanted to get some devices with less ram (have a `m4`...
Yes, Feathers, or the QT Py's.
short tendon
Is there a specific QT Py that has the smallest?
crimson ferry
all S2s have the same internal RAM, same for all S3s ...the difference is amount of PSRAM, there are some variations without PSRAM
(or with less, but 2MB I think is the min PSRAM, so not helpful if you're looking for less memory)
tulip sleet
there is 8MB flash / 0 MB PSRAM (no PSRAM), and 4/2. It has to do with what's embedded in the little silver modules.
crimson ferry
STM32F405 has 128KB available to user, I think M0 is the only thing smaller (32)
I think you can typically load a no-PSRAM UF2 onto a similar board with PSRAM (flash should match though), and CircuitPython won't recognize the extra memory
burnt tangle
Hey, could someone please point me to the magic that enables the press() calls here to actually make sound? I'm not seeing how DMA is still going when play() is called without loop=True
audio = audiobusio.I2SOut(board.I2S_BIT_CLOCK, board.I2S_WORD_SELECT, board.I2S_DATA)
synth = synthio.Synthesizer(sample_rate=44100)
audio.play(synth)
while True:
synth.press(65) # midi note 65 = F4
time.sleep(0.5)
synth.release(65) # release the note we pressed
time.sleep(2)
Synthesizer aways produces audio samples even if no notes are playing, so it's effectively an infinite stream of audio. Loop=true/false only matters when an audio stream ends
The samples are produced by a call of a function like "get buffer" in shared-module synthio. I can't be more specific right now, I'm mobile
burnt tangle
Thanks, details would be appreciated when able, not seeing how things don't grind to a halt like in audio_dma.c:
if (get_buffer_result == GET_BUFFER_DONE) {
if (dma->loop) {
audiosample_reset_buffer(dma->sample, dma->single_channel_output, dma->audio_channel);
} else {
// Set channel trigger to ourselves so we don't keep going.
dma_channel_hw_t *c = &dma_hw->ch[dma_channel];
c->al1_ctrl =
(c->al1_ctrl & ~DMA_CH0_CTRL_TRIG_CHAIN_TO_BITS) |
(dma_channel << DMA_CH0_CTRL_TRIG_CHAIN_TO_LSB);
if (output_length_used == 0 &&
!dma_channel_is_busy(dma->channel[0]) &&
!dma_channel_is_busy(dma->channel[1])) {
// No data has been read, and both DMA channels have now finished, so it's safe to stop.
audio_dma_stop(dma);
dma->playing_in_progress = false;
}
}
}
Because it doesn't return DONE
burnt tangle
ah thanks, that's a good toe-hold
orchid basinBOT
devout jolt
I'm working with a custom RP2040 and I think I have a flash-related issue. What I observe:
- Getting to RPI-RP2 works and I can upload a CircuitPython UF2 no problem
- CircuitPython runs well for as long as the board stays powered on
- On reset , CircuitPython does not come up (no USB device)
- Programming the board with
arduino-picousing the "Pico" board type works after reset
The board uses a 16MB Winbond W25Q128JVPQ. I've recompiled the standardraspberry_pi_picoCircuitPython to include this chip inmpconfigboard.mkwithEXTERNAL_FLASH_DEVICES = "W25Q128JVxQ"
My questions: - Did I change mpconfigboard.mk correctly?
- How do I turn on UART serial debug messages again?
- How can I further debug this?
ornate breach
I trying to open a bin file, and I know that memory is a bit tight, is there a way to get a little bit larger memory allocation?
the bin file is 104kb, gc.mem_free() says I have 154880 free
allocation fails by 56320 or 72448 bytes respectively
just depends on reload
code.py output:
Mem Free: 160608 Mem Alloc 2976```ah wait, I realized I built for a different fpga.. lemme try and use the right file..
that makes a world of a difference lol..
onyx hinge
@devout jolt a stab in the dark but try using this setting https://github.com/adafruit/circuitpython/blob/8.2.x/ports/raspberrypi/boards/adafruit_feather_rp2040/pico-sdk-configboard.h
tulip sleet
I'm working with a custom RP2040 and I think I have a flash-related issue. What ...
this is familiar and can be due to the clock oscillator not starting in time. There's an adjustment for that, which we added to all the adafruit boards... hold on while I find it.
oh, there, it is , I hadn't scrolled down. jepler points to what I was going to
devout jolt
this is familiar and can be due to the clock oscillator not starting in time. Th...
Ah hah! That’s the setting I was vaguely recalling. Thanks @tulip sleet & @onyx hinge! I’ll try it out
slender iron
@devout jolt let me know if the oscillator fix doesn't solve your issue. I can help with flash bringup
devout jolt
Yep @slender iron that was totally it. I remembered doing this on another board and totally forgot about it.
So follow up questions:
- Could we add
#define PICO_XOSC_STARTUP_DELAY_MULTIPLIER 64to theraspberry_pi_pico/pico-sdk-configboard.hboard? - Generally, what about having a "generic RP2040" and maybe "generic ESP32 S3" and similar boards? (Like what arduino-esp32 does)
When bringing up a new RP2040 board, I often use theraspberry_pi_picoboard because it has all GPIOs inboardand usually its flash choice works well enough.
(also didn't it used to be thatmicrocontroller.pins.*wasn't completely filled out, just being aliases forboard.*? But I see that's no longer the case at least on RP2040. maybe I'm thinking of a different chip arch)
tulip sleet
Yep <@252717193496756235> that was totally it. I remembered doing this on anothe...
We could add that to raspberry_pi_pico, but it seems like it's not needed on those boards. Either their choice of crystal-related components is very good or they toss the ones with this issue during QA. So that's why we didn't add it.
Maybe a generic board is appropriate, as it might benefit people who have some off-brand thing that they don't necessarily want to create a custom board for.
devout jolt
We could add that to raspberry_pi_pico, but it seems like it's not needed on tho...
There are a ton of Pico clones of various shapes. When people ask about the 16MB clones, I usually point them to the gnd_yd_rp2040 but it doesn't have all the GPIOs brought out to board. Lately, I've noticed CircuitPython is autodetecting the 16MB flash even when a 2MB flash is configured. Maybe since v9?
slender iron
flash size has always been autodetected. you still risk the wrong access settings though
tulip sleet
so a generic board would have to include all known expected flash chips
sounds like
EXETERNAL_FLASH_DEVICES in mpconfigboard.mk is required.
manic glacierBOT
bright robin
Is there a way we can get a visible cursor on the device LCD when running the REPL.
I can do line editing with the Cardputer's built in keyboard, but there's no cursor on the built in display so you have to guess where you are on the line.
manic glacierBOT
onyx hinge
Is there a way we can get a visible cursor on the device LCD when running the RE...
Right now this is not possible, sadly. I don't know if there's a specific issue requesting this functionality though. @lone axle did a guide that created a fullscreens editor and I think it created the cursor programmatically, but I'm not sure of the details, and I don't think it worked with the >>> REPL
manic glacierBOT
manic glacierBOT
short tendon
Not sure where to post this. Looking it's a bug with with CircutPython 9.1.0 Beta 3 and MicroPython 1.22.2:
class I:
@property
def i(self):
print("why hello there")
and then
>>> i = I()
>>> dir(i)
why hello there
['__class__', '__module__', '__qualname__', '__dict__', 'i']
So calling dir on an object actually calls the properties... Which can be bad if those properties do things
burnt tangle
short tendon
Looks like I'm not the first to find it. Thanks!
urban knoll
Yeah, goes all the way back to #2179 apparently
short tendon
And it is a MP issue, so not a whole lot to be done...
urban knoll
There's already a CP patch in there, so maybe... https://github.com/adafruit/circuitpython/blob/781c577745338128e1994d0b63fe95d75f3ef786/py/modbuiltins.c#L173
tulip sleet
And it is a MP issue, so not a whole lot to be done...
I tried fixing this a few years ago, and ran into more problems than I would have expected.
short tendon
I'm surprised they haven't. I can imagine a ton of hidden bugs from calling a property without realizing it
tulip sleet
MicroPython discourages use of properties because it's less efficient than methods. CircuitPython was more interested in typical Python style, so we use properties more and fixed them up slightly to work better, though not dir()
short tendon
Oh, that's weird. But I guess it makes sense
On a random note. I found this working through the socket logger. Which I now have in a pretty good spot.
onyx hinge
@stuck elbow @devout jolt @split ocean you may wanna take a look at https://github.com/adafruit/circuitpython/pull/9279 and have opinions
sly mist
I am trying to get a module to compile and include in my firmware package, but the I keep getting linker errors.
I have my files in shared-module/aurora_epaper and shared-bindings/aurora_epaper. I think I have added the necessary compile flags and files to py/circuitpy_mpconfig.mk and py/circuitpy_defns.mk to have it included, however I still get linker errors because some of my functions are used in shared-module/displayio/__init__.c and It cant find them.
Here is the relevant make output:
xtensa-esp-elf/bin/ld: warning: build-offsummit_2024/firmware.elf has a LOAD segment with RWX permissions
xtensa-esp-elf/bin/ld: warning: build-offsummit_2024/firmware.elf has a LOAD segment with RWX permissions
xtensa-esp-elf/bin/ld: build-offsummit_2024/shared-module/displayio/__init__.o:(.literal.common_hal_displayio_release_displays+0x8): undefined reference to `aurora_framebuffer_type'
xtensa-esp-elf/bin/ld: build-offsummit_2024/shared-module/displayio/__init__.o:(.literal.common_hal_displayio_release_displays+0x10): undefined reference to `common_hal_aurora_epaper_framebuffer_deinit'
xtensa-esp-elf/bin/ld: build-offsummit_2024/shared-module/displayio/__init__.o:(.literal.reset_displays+0x0): undefined reference to `common_hal_aurora_epaper_framebuffer_reset'
xtensa-esp-elf/bin/ld: build-offsummit_2024/shared-module/displayio/__init__.o:(.literal.displayio_gc_collect+0x4): undefined reference to `common_hal_aurora_epaper_framebuffer_collect_ptrs'
xtensa-esp-elf/bin/ld: build-offsummit_2024/shared-module/displayio/__init__.o: in function `common_hal_displayio_release_displays':
__init__.c:(.text.common_hal_displayio_release_displays+0x11c): undefined reference to `common_hal_aurora_epaper_framebuffer_deinit'
xtensa-esp-elf/bin/ld: build-offsummit_2024/shared-module/displayio/__init__.o: in function `reset_displays':
__init__.c:(.text.reset_displays+0x156): undefined reference to `common_hal_aurora_epaper_framebuffer_reset'
xtensa-esp-elf/bin/ld: build-offsummit_2024/shared-module/displayio/__init__.o: in function `displayio_gc_collect':
__init__.c:(.text.displayio_gc_collect+0x33): undefined reference to `common_hal_aurora_epaper_framebuffer_collect_ptrs'
xtensa-esp-elf/bin/ld: __init__.c:(.text.displayio_gc_collect+0x86): undefined reference to `common_hal_aurora_epaper_framebuffer_collect_ptrs'
I was loosely following this article but it is out of date: https://learn.adafruit.com/extending-circuitpython?view=all
tulip sleet
I am trying to get a module to compile and include in my firmware package, but t...
are you sure your source files are being compiled? You can use the V= options to see all the compilation lines, and also check in the build directory for the proper .o files.
small comet
(just checking) is there a "standard" stanza one can currently opt into (say) the pre-commit yaml file for a library, to run mypy type checks?
sly mist
are you sure your source files are being compiled? You can use the `V=` options ...
Your right it doesn't seem to be compiling my module, which is strange because I had compiler errors before. Okay I found it, SRC_PATTERS += aurora_epaper/% mistyped PATTERNS. 🤦♂️
Thanks!
small comet
hmm... <https://jaredkhan.com/blog/mypy-pre-commit>
Ack, thanks for the pointer!
atomic summit
Hey @tulip sleet Not a big thing, but the release notes for 9.1 betas don't have ESP32-C6 listed under port status.
It definitely should - C6 has been working great since 9.0.2 and I've been shipping 9.0.x on my TinyC6 boards without issue 🙂
tulip sleet
Hey <@329766224093249548> Not a big thing, but the release notes for 9.1 betas d...
I’ll add it!
tulip sleet
Hey <@329766224093249548> Not a big thing, but the release notes for 9.1 betas d...
I've updated the published release notes for 9.0.5 and 9.1.0-beta.3. I also added ESP32-C2, which is present in beta.3, but not 9.0.x.
atomic summit
I've updated the published release notes for 9.0.5 and 9.1.0-beta.3. I also adde...
Nice!
steady mulch
Hi, my team has a hardware design with a Pico W in it. RAM is getting tight on the Pico W image on CircuitPython 8, and currently blocking upgrade to 9. There are savings to be made in code that was written when there seemed to be plenty of RAM, but may not go far enough.
If we were looking to add an official board, is there a minimum quantity to be considered, and would it require its own USB VID/PID despite being a Pico W? (Noting that Badger 2040 W has one.)
If we were looking to release a modified CircuitPython ourselves, are there guidelines about that?
candid sun
<@&356864093652516868> We'll have our weekly meeting in about 90 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/1EvRvqSdQW_nMZFkgvgHDKlqHlMl08z2QehnGhNohHSA/edit -- I look forward to everyone's updates!
tulip sleet
Hi, my team has a hardware design with a Pico W in it. RAM is getting tight on t...
If you're making your own board that you'd like to release as an official board (as opposed to some product where CircuitPython is not visible to the user, say), then it's fine to make that a contributed board. You can get a USB PID from https://github.com/raspberrypi/usb-pid or from https://pid.codes.
If you want to fork and maintain your own version, we ask that you not call it "CircuitPython". See https://docs.circuitpython.org/en/latest/README.html#branding
I think a regular board definition with more features turned off (PICO DVI, USB host, etc.) may get you back enough RAM.
Or you can ship with 8.2.x for now.
manic glacierBOT
steady mulch
If you're making your own board that you'd like to release as an official board ...
Raspberrypi implies that you need a company and a minimum quantity, and that you should use a standard one if possible. pid.codes says it needs to be open-source hardware, which I'm not sure if it would be if it has a Pico W in it? Maybe that just counts as a component?
steady mulch
If you want to fork and maintain your own version, we ask that you not call it "...
Is there any concern about "CircuitPython" text inside it / instructions about what to do with version info etc?
tulip sleet
Is there any concern about "CircuitPython" text inside it / instructions about w...
you can change the strings that print "CircuitPython"
i'm confused, are you distributing this? So there is more than one?
steady mulch
There are currently 135 as far as we know
tulip sleet
if your board is open-source, that should be fine. we have our own open-source boards that have proprietary chips in them
it's a component
is the badger a similar example?
steady mulch
I think a regular board definition with more features turned off (PICO DVI, USB ...
I was wondering about freezing additional adafruit libraries too. Though we accidentally packaged adafruit_bus_device even though it was frozen, and removing it only seemed to save 30 bytes.
tulip sleet
adafruit_bus_device is not a python library, it's C code. Did you also put it in FROZEN_MPY_DIRS or whatever?
it's fine to freeze our libraries
steady mulch
No I'm comparing with and without adafruit_bus_device on the CIRCUITPY drive on stock PicoW image
lone axle
There is a python version of bus device too I think https://github.com/adafruit/Adafruit_CircuitPython_BusDevice. It was converted to C and added to the core more recently.
steady mulch
Deleting it saved 30 bytes of RAM
tulip sleet
are you leaning toward a fork or just submitting a board?
steady mulch
Undecided. Submitting a board would make some things easier but there could be issues with not having control over lib versions if it does have extra frozen libs in it
Or indeed if we can make it work with stock PicoW
Yes I'd say the Badger is similar
Generally if you freeze a lib, do you expect to save the whole size of the mpy in RAM or is the difference much less?
tulip sleet
i think whether we remove some stuff or not from the current Pico W build is still up in the air. Also I want to look at the RAM consumption.
tulip sleet
Generally if you freeze a lib, do you expect to save the whole size of the mpy i...
almost the whole size. IMporting it adds a few dict entries but nothing else
tulip sleet
@candid sun i can finish the meeting if you want
candid sun
Sorry I think my computer dropped internet
lone axle
Thanks for hosting Liz. Have a great week everyone!
tulip sleet
thanks!
@lone axle the thing I sent you in email about the readthedocs default change isn't private, it was just the easiest way to get it to you. If there's anything to discuss here is fine
short tendon
@tulip sleet am I good to merge and release this: https://github.com/adafruit/Adafruit_CircuitPython_AWS_IOT/pull/27?
tulip sleet
<@329766224093249548> am I good to merge and release this: https://github.com/ad...
Yes. To make a release, title it "x.y.z - something something", and use the Generate Release Notes button. That is all you need do to get it into the bundle and pypi
short tendon
Yes. To make a release, title it "x.y.z - something something", and use the Gene...
Done!
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/1A-m8lDphnjKaPfIkQtyfb1Tb0lk9HHPnEFO8wNgLTIQ/edit?usp=sharing
onyx hinge
sorry I missed it but I had to catch up with some local friends over lunch. ❤️
manic glacierBOT
stuck elbow
on topic of speech synthesis and SAM, I just learned about https://components.espressif.com/components/jmattsson/picotts
orchid basinBOT
solar whale
@onyx hinge Thanks for merging the ov5640 PR -- i't's been awhile since I did a PR, Should I generate a new release or is that something that will be done by someone else?
onyx hinge
<@426200547024961539> Thanks for merging the ov5640 PR -- i't's been awhile sinc...
@lone axle may be doing "sweeps" and releasing things as needed, or if you have the necessary access you are welcome to do it manually. (the instructions for doing so were reviewed recently if you're feeling rusty: <#circuitpython-dev message>)
lone axle
I typically do a pass and release any with new commits on monday's (but haven't done one this week due to the holiday).
Making the release directly after PRs merged is good especially if you want it to make it into bundle and circup ASAP. But if it gets skipped no worries because I'll get to it during that weekly pass.
onyx hinge
thanks both!
orchid basinBOT
tulip sleet
@slender iron @onyx hinge I would be up for another 9.x.x issue (and maybe some PR's) triage meeting. Do you have time today?
slender iron
Yup, I'm mostly around. Just gonna do lunch and a short walk sometime
(before our internal meetings)
onyx hinge
I'm around as well
tulip sleet
now is fine, or if you want an early lunch, can wait until after that. I have no more scheduled events today
slender iron
I'm around as well
We're in the video of our shared DM
onyx hinge
Give me 5
onyx hinge
oh, cool, debian is back to updating gcc-arm-none-eabi. stable has 12.2, testing and sid have 13.2. I thought they had stopped/gotten way behind
also huh: TIL that google sheets has trouble exporting slides presentations with emoji, even when exporting as PDF. there are 2 missing emoji in the PDF of my presentation.
Also, I have now posted the slides of my presentation: https://emergent.unpythonic.net/files/sandbox/Connecting Old to New with CircuitPython.pdf
and you can also see the version inside google sheets with speaker notes and a few animations which also don't come through in PDF: https://docs.google.com/presentation/d/1524EXIJRLf5XJSetKCbnN5ilNuicdJ7i1ZHUpgpIevQ/edit?usp=sharing
devout jolt
also huh: TIL that google sheets has trouble exporting slides presentations with...
Was video recorded so we can watch it too?
onyx hinge
@devout jolt the video was recorded and will be released on the pycon youtube channel sometime in the coming months
devout jolt
<@352910176736772096> the video was recorded and will be released on the pycon y...
awesome!
orchid basinBOT
alpine rapids
Hey folks! I forget, is the neopixel library ever included with a board (like as a frozen library), or does it always need to be installed separately?
Just asking b/c I'm working on adding a new board and it has an onboard ws2812
tulip sleet
Hey folks! I forget, is the neopixel library ever included with a board (like as...
A few boards freeze it to save RAM or because the manufacturer has chosen to freeze it.
solar whale
@lone axle I made a new release for the OV5640 library yesterday (1.2.0) and it looks like it got into the bundle OK and it shows up in the repo under releases, but in the github repository, it still shows 1.1.6 with the "latest" tag. Did I miss some step?
lone axle
<@382939733107408897> I made a new release for the OV5640 library yesterday (1.2...
I think it's controlled by this checkbox from the create (or update) release page. I'll try checking that box now and updating it.
solar whale
Thanks -- I missed that!
lone axle
That seems to have done the trick, the main github page for the repo shows the newer release now.
They put that check mark kind of inconveniently close to the "submit" button IMO. I know I have accidentally clicked the check while aiming for the button. It defaults to checked for me when the page loads, but I'm not sure if I configured it to do that somehow. For me it means if I click it then it becomes unchecked and I need to click again.
spare jacinth
exploring a weird idea, would anyone use something like this? 🤣
debating whether to add some more functionality and PR'ing to the bundle
import utest
class Example(utest.TestCase):
def test_foo(self):
with self.assertRaises(ValueError):
raise ValueError
def test_bar(self):
with self.assertRaises(ValueError):
raise OSError
@utest.skip
def test_skip(self):
self.fail("should have been skipped")
Example(failfast=False)
code.py output:
test_foo: success
test_bar: failed
OSError was raised while expecting ValueError
test_skip: skipped
I think there is a similar one here: https://github.com/mytechnotalent/CircuitPython_Unittest
spare jacinth
ha, my code looks almost the same
ornate breach
Is there a way to change whether CS is toggled on every byte with FourWire?
I was looking and it looks like someone made a patch to do that in bus_core, but perhaps that can be an optional argument?
manic glacierBOT
This looks like normal SPI to me:
I think there is a bug on your side. I'd encourage you to use a logic analyzer to compare what CP does to your working Arduino example.
ornate breach
@slender iron I did see the difference in spi data when I looked at it a month or two ago with my logic analyzer. I didn’t think to look what the color format was 😀
slender iron
it isn't too hard to add another color format
ornate breach
After I finish some stuff adding a new board to CP I’ll see about adding support for RGB565
slender iron
the trickiest bit will be that its 3 bytes
@ornate breach are you making a custom board for it? you could switch to parallel connection
it has RGB565 parallel support
ornate breach
No, I just have a spi one that I’d love to use on a project
slender iron
ornate breach
Plus a few people have come to my library asking if it was working yet
willow totem
If GH says the ble bonding was merged 9hrs ago, shouldn't the S3 blobs have something dated the 31st May or 1st June? GB+ US same.
https://adafruit-circuit-python.s3.amazonaws.com/index.html?prefix=bin/adafruit_feather_esp32s3_tft/en_US/
Ahh, I've just seen the assets in the build action, but they don't contain the s3 related tft feather builds, that must be it, not ideal / thought concerning.
short tendon
I believe If they were skipped, that would mean the code change doesn't affect that board...
onyx hinge
@willow totem looking at the related PR run https://github.com/adafruit/circuitpython/actions/runs/9327213101 I can see that adafruit_feather_esp32s3_tft was not built. you can see a list of all artifacts on that page. One that _was_built on github is also on s3: https://adafruit-circuit-python.s3.amazonaws.com/index.html?prefix=bin/adafruit_feather_esp32_v2/en_US/
If PR9289 "should have" made a difference on the adafruit_feather_esp32s3_tft board then there's a bug in the tooling which should be reported as an issue.
however, I think it's "correct": adafruit_feather_esp32s3_tft does not enable _bleio, but the only changes in PR9289 are in _bleio
willow totem
in my head, all the s3 boards should be affected, but I could well be wrong
ah, got it, I expected most s3 boards to have _bleio switched on
I appreciate those with displays lack more space
onyx hinge
ifeq ($(CIRCUITPY_ESP_FLASH_SIZE),4MB)
CIRCUITPY_BLEIO = 0
endif
``` maybe it's this stanza?in ports/espressif/mpconfigport.mk
willow totem
Thanks, that allows me to play if desperate, for now I'll probably drop the tft board BLE testing (he says reaching to update his fork)
onyx hinge
some code to efficiently calculate sin/cos to ~4.5 digits that might come in handy for making synthio filters tweakable in real time: https://gist.github.com/jepler/fe74b3f0a0f409f208b94fb3b840a7cb
no need to call 2 expensive trig functions, just calculate two related degree-5 polynomials. About 9 float multiplies & 6 float add/subtracts for sin & cos.
hopefully that'd make it fast enough to re-calculate the filter parameters every 256 audio samples, so the filter parameters can be controlled via LFOs...
not that I'm continuing work on this right now, but the exercise of making a quick sin+cos approximation was interesting to me
typical error of approximation, note the scale
ornate breach
is there a way to disable FLOPPY in rp2040 builds?
ah wait, CIRCUITPY_FLOPPYIO = 0
tiny peak
For people who are considering the question of how to deal with building sequencers to use with synthio, I've been working on a CircuitPython sequencer thing that you might find interesting. It more or less works now, but I haven't tested it very thoroughly: https://github.com/samblenny/txtseq
I'm planning to write up a Playgrounds guide about it. [edit: the Playground guide is up]
manic glacierBOT
empty salmon
How often are PRs processed on Learning Guides? and/or Whom do I communicate with to answer any open questions on a PR? (2813)
manic glacierBOT
TIL: sending image attachments as part of email replies doesn't work.
My phone sees this:
And on Desktop:
tulip sleet
How often are PRs processed on Learning Guides? and/or Whom do I communicate wit...
You can add another comment to the PR asking if they are any questions. When I saw this PR I thought it should be tested on, say, a CPX, to make sure the problem you are seeing is across all IR implementations. It is worth investigating whether/why this wasn't caught before. I'd want to test a remote control to see what it sends, for instance.
tulip sleet
How often are PRs processed on Learning Guides? and/or Whom do I communicate wit...
i am now looking more carefully. It looks like they are indeed reversed. We'll need to change some text in the guides, too.
empty salmon
@tulip sleet - thanks for the advice. The PR mentions what is was tested on. I could have added that the tests were not don't on other hardware because I didn't have them.
My initial experience confused me - how could the error be in all the guides, it must be me - until I realized the error was in the learning guides but not is the library example code.
tulip sleet
<@329766224093249548> - thanks for the advice. The PR mentions what is was teste...
The library code was fixed really recently by tkomde's PR's. But that change should have been carried over into the Learn Guides, and wasn't.
The library code changes the 550's to 560's. Could you push a change that does that too?
I approved but then realized they should be 560's to be consistent
empty salmon
OK. I will update the PR with the duration changes.
tulip sleet
let me check the header values too
header values are 9000, 4500 in the library
maybe retest on your own setup with those header values to make sure they work ok
@empty salmon ^^
empty salmon
OK, after an embarrassing copy/paste error, he learning guide PR has been updated 🫣
lone axle
<@382939733107408897> the thing I sent you in email about the readthedocs defau...
Regarding the RTD deprecation of README.html I'm taking a closer look into it now with the intention of making any changes we need this week if possible. However I'm unsure of a few things:
Are we actually using the deprecated technique? Inside a local copy of a library repo (asd1x15, the 1st one in the lsit) when I build the docs the file that gets output is index.html as far as I can see there is no README.html involved anywhere in our docs build. If I try loading a URL like https://docs.circuitpython.org/projects/ads1x15/en/latest/README.html I get a 404 page, Whereas https://docs.circuitpython.org/projects/ads1x15/en/latest/index.html loads the same page as the same URL sans index.html.
I'm wondering if they have some tool or metric they used to generate that list of repos, and if so why it only includes some libraries and not all of them. AFAIK the configuration is fairly consistent across libraries, but perhaps this subset still has something different about it? Or maybe the rest of the libraries are now considered sub-projects of circuitpython, so they aren't listed out specifically in the list?
tulip sleet
is ads1x15 a subproject?
lone axle
One of the 'how to update your project' steps mentions:
Migrate your Sphinx documentation to use master_doc = "index" instead of master_doc = "README". This can be done by re-structuring your documentation or by creating a symlink from of index.rst -> README.rst.
We already have master_doc = "index" in the repo here: https://github.com/adafruit/Adafruit_CircuitPython_ADS1x15/blob/main/docs/conf.py#L44 but based on the second sentence there about restructuring or creating a symlink I'm not sure if this literal variable in the code is what they are talking about.
lone axle
is ads1x15 a subproject?
I'm not sure how to tell honestly.
tulip sleet
We already have `master_doc = "index"` in the repo here: https://github.com/adaf...
I think it is. I didn't bother to check. I am trying to log in now to rtd
lone axle
The RTD configuration is something I don't have much insight into. A few of the libraries that I started, I created the docs for and still have access to admin/settings on. But the vast majority I do not have access. I'm also not exactly sure how the sub-project / parent-project relationship works in RTD, but in my mind there is one that is at least tied to the way the custom docs URLs work.
tulip sleet
ok, it is a subproject
lone axle
ok, it _is_ a subproject
Is something like Display_Text a subproject?
tulip sleet
yes, every CPy library should be a subprojevct
lone axle
okay, I'm at a loss for why only those ones in the email were listed then and not all of them.
That makes me think they had some tool to "check" for ones affected, but it's unclear to me what about these ones would make them affected by the deprecation. vs other like display_text for instance.
tulip sleet
I'm wondering if an older version is using README.html. That library has multiple version of the docs available. See https://docs.circuitpython.org/projects/ads1x15/en/latest/ and click on v:latest at the bottom of the left sidebar
tulip sleet
i'll see whether that library is unusual in having all those versions. it might be a setting issue. Or maybe it was one of a few libraries long again that used README.html.
lone axle
it looks like only the oldest one 0.2.1. README.html is 404 on 0.3.0
I think typically libraries show just latest and stable. I don't recall ever seeing that big list of versions. Though I also don't open that pull up thing too much.
it does also look like all the ones in the list are similar with old numbered versions existing.
tulip sleet
I think typically libraries show just `latest` and `stable`. I don't recall ever...
I think there are many libraries with multiple numeric versions listed. I'm not sure why some have it and some don't. Maybe it's the webhook settings, but it's not obvious to me.
tulip sleet
I think typically libraries show just `latest` and `stable`. I don't recall ever...
I started deleting the numeric versions on some libraries, but it's very tedious because a single library might have several dozen such versions. I'm thinking it's not worth it at all: the README.html links are on really old versions that nobody is going to click on anyway. At some point we should figure out why some have numeric versions and some don't, and maybe clean that up with some RTD API scripts, but it doesn't seem very important after all.
lone axle
<@&356864093652516868> the weekly meeting will be occuring here on discord at the usual time of 2pm eastern / 11am pacific, which is about 1 hour from now. The notes document is available here: https://docs.google.com/document/d/1A-m8lDphnjKaPfIkQtyfb1Tb0lk9HHPnEFO8wNgLTIQ/edit?usp=sharing for you to put in hug reports and status updates. We looking forward to hearing from you in a bit!
stuck elbow
I keep getting /home/sheep/.espressif/tools/xtensa-esp-elf/esp-13.2.0_20230928/xtensa-esp-elf/bin/../lib/gcc/xtensa-esp-elf/13.2.0/../../../../xtensa-esp-elf/bin/ld: __init__.c:(.text.common_hal_espidf_get_total_psram+0x12): undefined reference to esp_psram_get_size'` when trying to compile main or 9.1.0-beta.0 for the esp, any ideas?
slender iron
I keep getting `/home/sheep/.espressif/tools/xtensa-esp-elf/esp-13.2.0_20230928/...
is your esp-idf folder up to date?
stuck elbow
I did make fetch-port-submodules and did esp-idf/install.sh afterwards, that should do it?
slender iron
check git status too
stuck elbow
seems good
onyx hinge
@stuck elbow could be a mismatch about whether psram is enabled in esp-idf-config?
@slender iron @tulip sleet I have my gpio profiling data ^^^ give me a moment to explain it
stuck elbow
this is for seeed_xiao_esp32s3 that is not yet merged, so maybe something about the board definitions changed?
tulip sleet
this is for seeed_xiao_esp32s3 that is not yet merged, so maybe something about ...
can you try it on beta.3 instead of .0
the ESP-IDF submodule might not be up to date in your clone
stuck elbow
I added CIRCUITPY_ESP_PSRAM_SIZE = 8MB CIRCUITPY_ESP_PSRAM_MODE = oct CIRCUITPY_ESP_PSRAM_FREQ = 80m to the mpconfigport.mk, now I get CMake Error at esp-idf/tools/cmake/project.cmake:587 (message): SDKCONFIG_DEFAULTS '/home/sheep/dev/circuitpython/circuitpython/ports/espressif/esp-idf-config/sdkconfig-psram-oct.defaults' does not exist.
slender iron
I think it is OPI
stuck elbow
thanks
slender iron
iirc
stuck elbow
compiling...
onyx hinge
D1 is "get_buffer" from the mp3 stream. it should occur regularly, though there is some variation (not sure why). D0/D2/D3/D4 are related to the buffer fill of the MP3 coded audio from network, drilling down to a progressively smaller portion of the code. D4 is just around the line mp_uint_t out_sz = stream_p->read(MP_OBJ_FROM_PTR(stream), buf, len, &errcode);: After the socket has been determined to be readable we ask for a read of up to len bytes, and it takes quite a long time, nearly 50ms, to complete the read.
I haven't eliminated the possibility that this is GC or something, but my program's not generating garbage and socket recv shouldn't generate garbage either. it feels like something in the network stack...
stuck elbow
success, thanks everyone
onyx hinge
hmmmm maybe it is my bug. let me think about this a bit more. There's a loop around stream_p->read and it does NOT check readability at each iteration.
slender iron
I'd probably be more help when talking it over video
onyx hinge
yeah just doing a quick brain dump
willow totem
Re: Micrpython1.23 - OpenAMP sounds really interesting, C<->PY dual cpu coordination. Also the Frozen module size reductions
gilded cradle
Sorry about that. All sorted out now.
slender iron
This code should work for circup ble: https://github.com/adafruit/Adafruit_CircuitPython_BLE_File_Transfer/blob/main/examples/ble_file_transfer_listdirs.py
onyx hinge
[debian stable's version of uncrustify is still OK for circuitpython but for those running testing the pip-installable version of uncrustify is probably the way to go]
willow totem
Would that G0 i2c coprocessor work in a similar OpenAMP (RPC) way, or more like a seesaw with limited predefined i2c methods (while still running cpy on the G0)?
slender iron
Would that G0 i2c coprocessor work in a similar OpenAMP (RPC) way, or more like ...
Somewhere in between. Not sure we need full OpenAMP but it is intended that you load firmware as you need it
I have tooling I intend to bind C state to a python object for access
onyx hinge
Would that G0 i2c coprocessor work in a similar OpenAMP (RPC) way, or more like ...
the blurb about openamp talks about shared memory scheme, but the G0 wouldn't share memory with the CircuitPython CPU
willow totem
oh didn;t read it yet, thanks
onyx hinge
[thanks for reading that]
onyx hinge
tiny peak
re: in the weeds python versions... macOS Ventura on M1 is python 3.9.6
slender iron
thanks for hosting foamyguy!
gilded cradle
Thanks
onyx hinge
yay -- gpio profiling helped me notice where I was doing a blocking read in the background context. now you can still get a glitch if you cause LOTS of packet loss (putting a finger on the metal can of the module does this pretty reliably). remaining problem: crash when hitting ctrl-c on an mp3 stream
lone axle
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/1UroZPgqf8H2sqgK9PbootZYyfOgK2PySqTj5470siHA/edit?usp=sharing
tulip sleet
@willow totem did you cancel the new PR build?
willow totem
i didn't think so, but maybe accidentally, I was deleting the adafruit branch (not my remote branch tyeth/blah), so was stopping the jobs still using it (the others listed tyeth:blah instead of blah). I'll avoid that from now on.
devout jolt
Same energy 😄
(I don't think this is a bad solution, but it did make me chuckle)
austere acorn
Ohai,
I wasn't sure I would be able to make the CircuitPython meeting today, so I'm posting my update now.
I started working on porting CircuitPython to the Electrosmith Daisy Seed board https://electro-smith.com/products/daisy-seed a couple of months ago, and I gotta say, it was more of a challenge than I imagined 😅 Since the internal flash for the MCU is just 128kB, CircuitPython needs to run entirely from the external flash, and I had a lot to learn about how to do that.
After weeks of figuring out how to get the external flash working in QPI mode (no examples out there for this ISSI flash chip), how to write an external loader for the ST programmer, and how to write the bootloader that will make the jump from internal to external flash, I finally have "Hello World" running on Daisy Seed: REPL working, and on board LED is blinking!
https://www.youtube.com/watch?v=n6aPXO9yjxk
This is still very much a work in progress, but I'm happy I got the first results, and I intend to write a full documentation of my process.
slender iron
Ohai,
I wasn't sure I would be able to make the CircuitPython meeting today, so...
Nice! Did you get the external ram supported too?
austere acorn
Nice! Did you get the external ram supported too?
Not yet, I wanted to figure out the flash thing first, since the H750 has plenty of ram for basics.
slender iron
Not yet, I wanted to figure out the flash thing first, since the H750 has plenty...
Makes total sense! Great progress!
austere acorn
Makes total sense! Great progress!
Thank you! 😺
slender iron
@hybrid scarab how does the ulab CI work now for these?
hybrid scarab
<@273936060608675842> how does the ulab CI work now for these?
Hmm it's actually copying the existing config file and changing a setting with sed, which ... doesn't exist 😬
specifically:
sed -e '/MICROPY_PY_UHASHLIB/s/1/0/' < circuitpython/ports/unix/mpconfigport.h > circuitpython/ports/unix/mpconfigport_ulab.h
Then, during make, supplies CFLAGS_EXTRA="-DMP_CONFIGFILE=\"<mpconfigport_ulab.h>\"
slender iron
power is blinking so I may disappear
hybrid scarab
Ooof, hope all is okay!
slender iron
ya, just windy
UPS ftw
I feel like I need more context for what you are trying to do
hybrid scarab
I feel like I need more context for what you are trying to do
I don't think it's relevant anymore, though the issue in the CircuitPython codebase- I think- still applies
slender iron
I get your point that it is seding a value not in that file
hybrid scarab
The issue being, specifically, that py/mpconfig.h offers a means to override mpconfigport.h which is not respected at all points in the codebase
slender iron
that only matters if we used it
I suspect we'll reorganize a bunch of that in the future
because it means non-port stuff is dependent on a port specific header
hybrid scarab
So it's just a vestigial non-feature, then! I had guessed as much
In that case I'll back up a little from this rabbit hole and drop its usage from the CI
Boy howdy I sure spoiled myself obsessing over build times in our CI 🤣 🤣
slender iron
ya 🙂 its on our "it'd be nice" list
hybrid scarab
You need less planning and more ADHD, works wonders!
manic glacierBOT
atomic summit
Hey folks - I want to add a new board or 2 to CP, and I'm curious about how far away 9.1 is, or if I should add them to both 9.0x and 9.1 ?
sleek relic
Sent from Mars?
OK
08:33:07.9326 MTC Sol 53473
lone sandalBOT
tulip sleet
@willow totem re #9301 above: did you turn on verbose logging or something? I've never seen so much in the log
it is failing earlier with another fatal error: https://github.com/tyeth/circuitpython/actions/runs/9358284335/job/25759793899#step:9:148
(too many @Tyeths 🙂 ) could you add an avatar or maybe be Tyethg or something? )
tulip sleet
As core developers, we very rarely run CI jobs in our own fork
willow totem
<@375674153203662851> re #9301 above: did you turn on verbose logging or somethi...
good morning 😄 yeah I turned on git trace 🫣
And sadly someone stole the #tyeth name once discord switched over
willow totem
it is failing earlier with another fatal error: <https://github.com/tyeth/circui...
Sorry, ignore that error, it's where I publish both local and global git configs, but the git config global file doesn't exist
tulip sleet
what's the git command that is actually causing the failure?
willow totem
(too many `@Tyeth`s 🙂 ) could you add an avatar or maybe be `Tyethg` or somethi...
I'm both Tyeths on this server (.tyeth and tyethgundry)
tulip sleet
@tyeth
willow totem
ah, that might be the imposter 😄 she's an artist that liked the tag and was quick to notice the new username requirements
I'll stick an image on, like the GH one (a toucan)
tulip sleet
could you point me to an earlier run without the verbose logging?
I'm not sure which git version is being used, but we could start using --filter=blob:none instead of shallow clones, though that may not have anything to do with this.
is this all so you don't have to build locally?
willow totem
right, let me give you minor background, I saw the transport 'file' not allowed and saw that was symlinks being denied, so I added that config option and that error went away but the user-fork remote being mentioned in the submodule fetch still happened, so I removed that config option. Probably a red herring.
https://github.com/tyeth/circuitpython/actions/runs/9356074069/job/25752665288
willow totem
is this all so you don't have to build locally?
yes
I think the filterblob:none may resolve the file transport error you see there.
This is the first job where I enabled the allow symlinks and you see it moans about not finding "fork" (the remote name of parent repo) for submodule:
https://github.com/tyeth/circuitpython/actions/runs/9356365363/job/25753659619
Only by turning on Git Trace did I then see it doing all sorts of fetch things for the recursive submodule init/update at which point it then references the "user-fork" or "fork" remote name for the parent repo instead of remote name for submodule. Why it specified remote at all for submodule fetching I do not know.
tulip sleet
I would try to reproduce this locally. It seems like overriding and allowing the file protocol is a bad idea. So the q is why is it trying to do that, and we don't see that in adafruit PR builds
willow totem
No, agreed, and I don't see it (transport 'file' not allowed) if the remote is origin (when doing a non-fork normal 500board CI job),** although I do see the remote name in submodule fetch**
tulip sleet
if we are still running on 22.04 then the git version will be too old and the git ppa would need to be set up to fetch a newer git
willow totem
if we are still running on 22.04 then the git version will be too old and the gi...
oh fascinating, and a good point
tulip sleet
i wonder if this is at all due to the "detached head" thing you see when you check out a remote that is not the main remote for your clone
can I step up a level and ask why you don't want to do builds locally?
willow totem
i thought I'd get around that detached head by doing a branch for the remote fork branch (checkout -b bla tyeth/bla)
willow totem
can I step up a level and ask why you don't want to do builds locally?
Just cos my machine hates life, and I'm often switching between 2 low power laptops and a stationary gaming laptop
misty garnet
@tulip sleet is there any more work on my end for the minimqtt PR: https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT/pull/210 ?
tulip sleet
<@329766224093249548> is there any more work on my end for the minimqtt PR: http...
sorry it fell off the list, I'll take a look today
misty garnet
np
tulip sleet
Just cos my machine hates life, and I'm often switching between 2 low power lapt...
are you not chez vous a lot of the time?
willow totem
@tyeth
I migrate between rooms and the greenhouse 😄 Oh and actually it turns out I'm all three Tyeths in this server
the main rig is also crap, but better than the others
tulip sleet
maybe remote desktop access to a stationary higher-powered machine...
willow totem
That's a very sensible idea.
Don't worry for now Dan, I just thought I'd stick the issue up for visibility and in case anyone had seen similar before, I'll play with runner versions and see what GH say
I do really appreciate you taking a look / thinking it through with me though 🌟
tulip sleet
happy to cogitate on this
willow totem
love that word
onyx hinge
yuck, there are lots of crashes if you start fuzzing our mp3 decoder 😕
it looks like some kind of memory corruption tends to occur when I exercise the mp3 code using a soma.fm stream. Usually it terminates decoding with an error that indicates bad MP3 format, but sometimes it safe-modes. but I can't capture the stream and decode it too [well I might be able to with tcpdump or ???? but I haven't] so I don't know if it's something about the content of the stream or something about the way it's delivered. It happens much more often than every 5 minutes so it's not TOO bad to iterate on, but I don't have many clues at this point.
I thought maybe fuzzing (automated program that creates varied inputs trying to find a program crash) would point me at one obvious thing but instead I have 59 distinct crashes it's discovered 😕
I guess I can curl/wget a portion of the stream, then serve it from a local http server. If that errors, I can repeat it as many times as I like; if it doesn't, that tells me it's something about the server.. I think I'll do that next
tulip sleet
love that word
DM'd you re something as well
onyx hinge
or something in the network stack I suppose
manic glacierBOT
slender iron
Hey folks - I want to add a new board or 2 to CP, and I'm curious about how far ...
We were talking about an RC soon
so maybe stable in the next week or two?
tulip sleet
@slender iron (he/him) I just read that ESP-IDF 5.2.2 was released
ornate breach
Thanks @slender iron for merging my PID request 😎
Now all RPGA feather needs is for the mini panels to arrive, and a circuitpython official build and things will be golden
short tendon
@lone axle question for you. The work you added to use circup to install something from local, could that be expanded to grab something from a URL? like https://raw.githubusercontent.com/adafruit/Adafruit_CircuitPython_Requests/3.2.9/adafruit_requests.py?
Ideally mirroring the format of pip
lone axle
<@382939733107408897> question for you. The work you added to use circup to inst...
I think it could. It could use a bit of logic right before here that checks for the argument to be a URL and if found, downloads it to a tmp location and then overwrites name with the path to the locally downloaded copy.
I'm not certain about pip's syntax though. I think I've seen it used like this once or twice, but not familair with the exact syntax.
although modules that aren't a single file would be more tricky
manic glacierBOT
short tendon
I think it could. It could use a bit of logic right before here that checks for ...
Awesome. Are you doing any work around there, or am I safe to try to add it and open a PR?
lone axle
Awesome. Are you doing any work around there, or am I safe to try to add it and ...
feel free if you'd like. I'm working on BLE support which shouldn't intersect with this code too much I think.
ornate breach
Hey @tulip sleet any chance you can get rpga feather PR looked at? 🙂
I saw you were active on gh and figured it was worth an ask
spare jacinth
does someone need to manually trigger a release for circuitpython-stubs? as in, it is not shipping with #9221
or does it rely on some stable version of CP itself being released?
lone axle
does someone need to manually trigger a release for `circuitpython-stubs`? as in...
I think it depends on when and to what branch the PR was merged. It was merged to main and I think that when it was merged we had already started the 9.1 beta releases, so new code from main is going into the 9.1 "lineage" not the 9.0 lineage. 9.0.5 was released after #9221 was merged, but I don't think 9.0.5 could have contained a merged from main, just 'fix' stuff for the 9.0 lineage, not new stuff which is going toward 9.1.
You can install the pre-release version now that includes the list command with:
pip install circuitpython-stubs --pre
# Or with a direct version specifier like
pip install circuitpython-stubs==9.1.0b3
Whenever 9.1 gets a stable release, that will be the first non-pre version that has the list command I think.
spare jacinth
ok, thx for the quick and complete answer 🙂
figured it was a release kind of thing, but wanted to raise the question in case it relied on some manual trigger 😄
--pre doesnt work, also grabs 9.0.5
urban knoll
Hey all, wondering if you've got recommendations for tools, config and workflow for JLink debugging on RP2040?
lone axle
`--pre` doesnt work, also grabs 9.0.5
that is weird, I don't have any idea what could cause that, maybe try updating pip or something? for me right now it's getting the beta:
❯ pip install circuitpython-stubs --pre
Defaulting to user installation because normal site-packages is not writeable
Collecting circuitpython-stubs
Using cached circuitpython_stubs-9.1.0b3-py3-none-any.whl (844 kB)
Installing collected packages: circuitpython-stubs
Successfully installed circuitpython-stubs-9.1.0b3
maybe i needed to also use --upgrade to skip cache
anyway, i got it with the specific 9.1 tag, no worries
lone axle
Ah, that could be if it was already installed. I explicitly uninstalled before.
spare jacinth
maybe not only uninstalling, i believe pip's caching survives the uninstall, but anyway i dont feel like testing this, i already have the latest hahaha
huh, takes 1.5-2 seconds to generate/print the whole list, too many boards
ornate breach
Thanks @tulip sleet 😄
urban knoll
Hey all, wondering if you've got recommendations for tools, config and workflow ...
@slender iron @tulip sleet I see some old docs for gdb on samd21, do you use similar procedure for all ports?
tulip sleet
<@252717193496756235> <@329766224093249548> I see some old docs for gdb on samd...
gdb works well on SAMD and nRF. It seems more problematic on Espressif due to very limited hw breakpoints. We tend to do a lot of printf debugging anyway. Sometimes we do pin toggling if things are time critical. Sometimes writing to a UART if we are debugging USB, etc.
urban knoll
How about rp2040?
I keep wanting to like fancy debugging tools, then falling back to pin toggling and printf's
tulip sleet
i have not tried the rp2040 debugging tools (picotool, etc.), but it's supposed to work
slender iron
<@252717193496756235> <@329766224093249548> I see some old docs for gdb on samd...
I use the gdb approach for arm cores when I need to
slender iron
but usually start with print debugging
(github copilot is good at autocompleting print debug statements)
short tendon
@lone axle or @tulip sleet can I add https://github.com/jelmer/dulwich to the circup requirements? It's what poetry uses. This way it could clone a branch
tulip sleet
<@382939733107408897> or <@329766224093249548> can I add https://github.com/jelm...
i missed the motivation for doing a clone instead of just fetching files or directories from github directly
short tendon
I'm still playing with things. If we clone, then we have more control. I'm imagining where I want to pin azure and it has a ton of files. Adding each one and it knowing they need to be in a directory together would be complicated
tulip sleet
if you use a clone then you have to make sure it's up to date
i thought that maybe the github api would do the file tree enumeration you want
short tendon
It would clone, copy files and delete the clone
It probably does, but I didn't want to reinvent the wheel
Or assume git was installed
tulip sleet
and this is a special case for a branch, when you want something not in the bundle?
lone axle
Peronally I lean towards just using requests if possible. That is used to pull the bundles already internally anyhow. Or if that can't work for some reason then I'd lean toward assuming git is installed if user wants to use this feature rather than introducing a seperate requirement for it.
short tendon
So I see it for a few cases:
- Something not in a bundle (like a personal repo)
- Pinning a version (for cases like breaking changes)
- Grabbing a specific file
Since many users aren't developers, they won't have git
onyx hinge
you can also get a branch as a zip from github: ```$ wget -q -O pioasm-main.zip https://github.com/adafruit/Adafruit_CircuitPython_PIOASM/zipball/main; unzip -l pioasm-main.zip | grep README
2969 2024-05-29 13:19 adafruit-Adafruit_CircuitPython_PIOASM-73170eb/README.rst
tulip sleet
but I think most of these cases are for sophisticated users who will have git installed. doing cp -r locally is not that hard. I'm thinking that these fancier features are not going to get used that much.
lone axle
the users installing specific libraries or versions from github directly via URL strike me as more likely to be developers than more general end users.
short tendon
Yeah, but then you need to make sure it can unzip...
tulip sleet
unzip is built into cpython, I think
onyx hinge
the bundle is a zip file, so circup can handle zips right now
of course there are no mpy files if it's not a release
short tendon
So my goal was to make circup match pip with the requirements
Which I would find super useful. Sounds like you all think maybe not
I'll probably keep this in a local fork
tulip sleet
i think circup does not need to be as good as pip. I see it as an end-user tool. I'm more frustrated by having to remember "Is it upgrade or update, etc.
i also see that when Melissa finishes the current work on code.circuitpython.org, that web tool could have library installation/updating added to it, with a list of libraries to click on, etc.
right now circup needs python installed, which is a big lift for the average user who is not a developer
you could look at MicroPython's mpip tool for inspiration
lone axle
The download with requests and unzip route seems like a good one to me. That would allow the similar or same functionality if pip even if implemented differently if I'm understanding correctly. And as Jeff noted that is also how the bundles are operated on by circup already so it would keep the new functionality working similar to existing.
short tendon
Sounds good. When I finish my fork, I'll share it. See if there are parts you would like in the main version
I'll look at the zip stuff and see if it's as easy as what I've done with dulwich
onyx hinge
https://github.blog/2024-06-03-arm64-on-github-actions-powering-faster-more-efficient-build-systems/
onyx hinge
[but not for free actions, if I understand correctly; these are grouped under "larger runners": https://docs.github.com/en/actions/using-github-hosted-runners/about-larger-runners/about-larger-runners]
@slender iron how can I get debug tracebacks without changing the optimization level? right now playback doesn't keep up in a debug build
onyx hinge
I'll ping you if I don't get it sorted out .. it'd be after my lunch, it's 11:53 here