#circuitpython-dev
manic glacierBOT
Yes, the hardware is limited to a few specific watchdog timeout periods.
You can find out the specific period that has been set by reading the timeout property. It is intended to always be at least as large as the value assigned. (however, there was a small bug there, see #5549)
>>> microcontroller.watchdog.timeout = 5
>>> microcontroller.watchdog.timeout
8.0
tulip sleet
there is already code in the hid library to try twice to get a device. DId you try just adding a time.sleep() at the top of code.py to allow time for USB enumeration?
tidal kiln
not my code. it's for this:
https://forums.adafruit.com/viewtopic.php?f=52&t=184411
guessing it's timing issue at boot
just adding a delay up top was one idea i had too
tulip sleet
I would try that. If that doesn't help, I would also try adding a time.sleep in boot.py, on the off chance that for some reason the Pi is not ready to do enumeration or something. keyboard.py already does this in __init__():
# Do a no-op to test if HID device is ready.
# If not, wait a bit and try once more.
try:
self.release_all()
except OSError:
time.sleep(1)
self.release_all()
my guess is that it's getting OSError again for some reason. Could put a try-except around k= Keyboard(), and try it again, or length the timeout above in the library.
tulip sleet
there's the one-second delay to allow going into safe mode, that's all
tidal kiln
and the second try delay above is also only 1 second
a typical pi boot is longer than that
tulip sleet
yes. Can you reproduce this?
tidal kiln
maybe...gathering things to try right now
the other scenario would be a reboot (vs. initial power)
tulip sleet
so what I would suggest is put the k = Keyboard() in a loop with a try-except, and keep trying over and over, so it works on power up
tidal kiln
since the CP board would remain powered
tulip sleet
yah
if the board remains powered, then at some point it will throw an OSError "USB Busy" exception when the host goes away. I don't know if the RPi cycles the power on its USB ports when it reboots. Maybe not since there's an internal hub
So the code.py could catch OSError and do microcontroller.reset() to restart completely. That would be different than the OSError catching at the start. It would be after it was running successfully.
tidal kiln
yep
would probably need to work in both
a try/loop at top to get going
and then catch/reboot elsewhere for the reboot scenario
lone axle
initializing the Keyboard() inside of try/except in a loop is the way I handled it at one point when working on a project where the microcontrollers were finished booting and trying to connect as a keyboard before their host was ready for them.
tidal kiln
@lone axle thanks. did you deal with warm boot scenario at all?
lone axle
<@!382939733107408897> thanks. did you deal with warm boot scenario at all?
not sure I understand what you mean. Like the microcontroller booting up while the host system is already powered on and didn't just boot up itself?
proven garnet
Hey y'all, I'm adding more typing to libraries, and I'm trying to figure out how CircuitPython handles the __exit__ method, for reference:
def __exit__(self, exception_type, exception_value, traceback)
On my little trinket, I can tell the exception_type is Optional[Type[BaseException]], and that exception_value is Optional[BaseException] since they may not get passed if the context manager doesn't fail. But I'm having trouble figuring out what traceback should be, I can only get it to return None. Does anyone know?
tidal kiln
@lone axle yep. like keep everything powered. and then just reset the host system.
tulip sleet
Hey y'all, I'm adding more typing to libraries, and I'm trying to figure out how...
I think it may not do anything, but I'm checking. Note this, btw: https://docs.micropython.org/en/latest/genrst/core_language.html#context-manager-exit-not-called-in-a-generator-which-does-not-run-to-completion
(we inherit that behavior from MicroPython)
proven garnet
I think it may not do anything, but I'm checking. Note this, btw: <https://docs....
Thanks for the reference, definitely good to have!
lone axle
<@!382939733107408897> yep. like keep everything powered. and then just reset th...
I don't think we handled that case specifically. In our project everything was powered from the same place, so everything tended to be rebooting at the same times.
This is the code we used (possibly out of date for todays API I think):
kbd_initialized = False
while kbd_initialized is False:
led[0] = (255,125,0)
try:
kbd = Keyboard()
kbd_initialized = True
led[0] = (0,255,0)
time.sleep(0.75)
except:
kbd_initialized = False
led[0] = (63,31,0)
time.sleep(1)
@lone axle thanks, yep. looks like your use case only had to worry about cold boot.
tulip sleet
Hey y'all, I'm adding more typing to libraries, and I'm trying to figure out how...
I think it always passes traceback=None, though it's a little hard to tell. However, in native implementations of __exit__(), we annotate it as Optional[TracebackType]
proven garnet
I think it always passes traceback=None, though it's a little hard to tell. Howe...
So is it best if I have the type hints match that then?
tulip sleet
Yes, just in case it ever gets added.
proven garnet
Awesome, will do then, thanks for the quick resolution!
tidal kiln
is the Tim C that wrote this guide here on discord?
https://learn.adafruit.com/circuitpython-display_text-library/overview
proven garnet
The workflow just came back with a pylint failure for duplicate code, I think I saw a commit on another library with the same issue that disabled that as well as had an upgrade pylint==2.11.1 . @idle owl, is that one of the changes being/been applied across the board, or how should I handle that?
PR for reference: https://github.com/adafruit/Adafruit_CircuitPython_Motor/pull/57
idle owl
is the Tim C that wrote this guide here on discord?
<https://learn.adafruit.com/...
foamyguy.
tidal kiln
thanks. should've guessed based on the other guides 🙂
tidal kiln
@lone axle hey, nothing urgent, but have request for this guide:
https://learn.adafruit.com/circuitpython-display_text-library/overview
idle owl
@tulip sleet A bunch of libraries are failing on unspecified-encoding which is calling open() without encoding specified. Should we be specifying? Or should we simply disable the check?
lone axle
<@!382939733107408897> hey, nothing urgent, but have request for this guide:
<ht...
always on the lookout for ways to improve it, whatcha got?
tidal kiln
the creating-new-labels-in-loop-and-hitting-memory-issue thing seems to come up multiple times a day
so wanting to add some new info to discuss that, show a simple example usage code, etc.
i think that guide is a better location than the displayio guide
lone axle
Ah, good idea maybe a troubleshooting page that shows an example of "bad" code making them in a loop (with plenty of warnings that it's not the right way) and then explain the issue and show the proper way to code it with a single label getting re-used.
tulip sleet
<@!329766224093249548> A bunch of libraries are failing on `unspecified-encoding...
it's not required by CPython, so it's a style thing. We don't support varied encodings anyway.
idle owl
it's not required by CPython, so it's a style thing. We don't support varied enc...
Thanks!
idle owl
The workflow just came back with a pylint failure for duplicate code, I think I ...
Hello! Yes, we are disabling that check on PRs that fail because of it, and in some cases filing an issue about refactoring into a base class and subclassing. Please add ,duplicate-code (exactly like that, comma, no space on either side of it) to the linked line in the .pre-commit-config.yaml file. Please don't alter this file in other PRs without instruction to do so. https://github.com/adafruit/Adafruit_CircuitPython_Motor/blob/93ace61ce2788965474b4d826d3434e7c06cbab8/.pre-commit-config.yaml#L27
proven garnet
Hello! Yes, we are disabling that check on PRs that fail because of it, and in s...
Will do, thanks!
idle owl
@lone axle Oh, right. My question from yesterday. Don't do it yet, but could your label-removing script be run on the CircuitPython core repo as well? There are 22 issues that still have it, and most of them are on the CP repo. I say "don't do it yet" because I want to find out what Scott's opinion is on it, since that's his thing.
lone axle
<@!382939733107408897> Oh, right. My question from yesterday. Don't do it yet, b...
I think that it could. It is essentially a python script that runs in the root of a repo. For the bundle libraries I used git submodule foreach to make it run in the root of each of those repo's. I believe the same script could work for any repo assuming that my user has permission to add/remove labels from issues (I did already for libraries, assume it's same for core but if not I think I would need that).
idle owl
I think that it could. It is essentially a python script that runs in the root o...
Right on, ok. I'll get back to you on this. Scott might want them left alone. We'll see.
idle owl
@proven garnet I replied to the FeatherWing PR, if you want to integrate the linting updates from Dylan's PR into your PR, to get it passing, I'll get it merged.
blissful pollen
Random question has anyone ever used framebufferio for something other then 1 bit (sharp display) or 16 bit (RGB matrix) color? Been trying with 24 bit for a proof of concept and something seems off
idle owl
@tulip sleet Is Threading.Thread.isAlive deprecated in CircuitPython also?
tulip sleet
we don't have threading turned on. where is that?
idle owl
Turns out it might be an example for RPi.
That confused Dylan.
Yeah is for RPi.
So nevermind. 😄
It's simply .is_alive now apparently.
tulip sleet
they regularized the camelCase to non-camelcase
the actual code is really low-level, too
you could just fix the is_alive
idle owl
you could just fix the is_alive
Pretty sure that's what is happening.
idle owl
@ember iris Can you use a with inside a with?
I was perhaps mistakenly under the impression that you could not.
ember iris
I think? I'm not 100% though. I was wondering if it was avoided because of the added overhead was too much in an embedded setup
idle owl
No, I simply thought you couldn't do it. Is there more overhead? Because that should definitely be avoided in the CP lib.....
Ugh. I guess I should try it and actually test it.
ember iris
I'm mostly asking because I don't know if there's a reason it's being avoided. I don't know if the overhead is much overhead either just because all I know is that with adds a flag so if there's a problem reading the file or writing the file, and it exits that code block aggressively, it'll make sure to safely close the file
That way the user doesn't have to remember and handle all the weird edge cases they could be thrown out of a file write/read
idle owl
Though looking at it, I don't quite understand how that block would look if I withed the second bit. Like this? py with self._audio_out( board.SPEAKER ) as audio: # pylint: disable=not-callable with audiomp3.MP3Decoder( open(file_name, "rb") ) as mp3file: audio.play(mp3file) while audio.playing: pass
ember iris
But I don't know how the micro python/circuit python handles the implementation of the with
idle owl
I see you peepin', @onyx hinge. No idea if you have input on this or are bringing up something else entirely.
onyx hinge
no not really
I just typed a letter accidentally while looking through discord channels
idle owl
Hahaha fair enough.
ember iris
Here's the source block of code in case: https://github.com/adafruit/Adafruit_CircuitPython_CircuitPlayground/pull/111/commits/f85c21ec1737e2c11cdcf51977b808ca2b38d8eb#r743961588
idle owl
Sorry for the disturbance then 🙂
onyx hinge
you can also write with expression1 as name1, expression2 as name2:, it's more compact but may/may not be better
idle owl
Yeah, there's two., The one I pasted above is the second one.
oh hmm
I'm not sure how to do it that way with a pylint: disable= in the middle. I guess I can move stuff around, let Black run, and then put the disable back where it goes
ember iris
Huh. That does get to be really messy really fast
idle owl
Oh hmm, it works where it ended up anyway.
OK, I'll do that on the other one as well.
Now mpy-cross and test on CPX, I suppose.
ember iris
So what does the block look like now? (I tried typing out what I thought it would become but it got messy and I figured it was too likely to be wrong)
idle owl
One looks like this py with self._audio_out( # pylint: disable=not-callable board.SPEAKER ) as audio, audiocore.WaveFile(open(file_name, "rb")) as wavefile: audio.play(wavefile)
The other looks like this: py with self._audio_out(board.SPEAKER) as audio, audiomp3.MP3Decoder( open(file_name, "rb") ) as mp3file: # pylint: disable=not-callable audio.play(mp3file)
Dunno why one worked one way and the other only worked the other way.... 🤷🏻♀️
But its passing pre-commit
ember iris
Ok that's about what I thought it would be minus the newlines to break it up
idle owl
Black did all of that. 🙂
ember iris
haha black always has a lot of work to do when I write code, so that makes sense that I didn't get that part right
idle owl
I opened the Adafruit GitHub page for some reason, and then got caught up in this, and have no idea what I was going to do with the page.
There was a reason....
ember iris
haha I was going to do something then saw the review requested and looked and can't remember what I was doing beforehand either, something for the FAQ but beyond that who knows
I can approve the changes now or I can figure out a test setup to see if it works but I think it should be fine
idle owl
@ember iris I tested it. It works.
Pro tip: Don't put audio file playback in the loop. 🙄
idle owl
I can approve the changes now or I can figure out a test setup to see if it work...
Feel free to merge it. 🙂
ember iris
Do I add anything to the comment on the merge?
ember iris
good to know, thanks!
idle owl
Thank you for the thorough review and help!
ember iris
No problem!
idle owl
If you need something to do.... there's a BUNCH of these.
ember iris
Relatedly, there's a bunch of pylint pull requests right now--why are we adding it and what should I look for?
idle owl
There's two more of mine, and Dylan is cranking them out as well.
Oh!
We upgraded Pylint from 2.7.1 to 2.11.1.
ember iris
ahhhh
idle owl
So, a bunch of libraries are failing Pylint now.
So be on the lookout for the exactly what you did on that PR - simply review it as usual.
Nothing special except we're trying to sort out Pylint updates.
ember iris
Ok that makes me feel better about going through them, I was sidelined with other stuff but saw a ton coming through and instinctively went, "Something is odd about this volume" that makes a lot of sense!
idle owl
Dylan is only tagging me on PRs, should I suggest tagging CPLibians too?
I think yes. 😄
ember iris
I'm seeing most of hers, just wasn't sure and was too busy to sit down and sort it out. Now that it's the weekend I'm a bit more flexible
onyx hinge
been meaning to do that ^, it worked very conveniently in gifio.
idle owl
I'm seeing most of hers, just wasn't sure and was too busy to sit down and sort ...
Fair enough!
idle owl
been meaning to do that ^, it worked very conveniently in gifio.
Oh nice ok
onyx hinge
I didn't test it so it's probably broken 😕
but I like the idea of getting rid of open()
idle owl
but I like the idea of getting rid of open()
Will it break everything now? Or is it backwards compatible?
onyx hinge
it should be compatible, you can use it either way
I converted it to draft since I didn't test it and probably won't test it today.
idle owl
I converted it to draft since I didn't test it and probably won't test it today.
That's fair.
I'll work with you next week to test it too, if you like.
idle owl
Another pro tip - maybe don't leave that code that plays a wav file in a loop on the board when you toss it in the drawer to not pick it up for months and get nailed with a looping wav when you pick it back up 😄
ember iris
Haha does it instantly remind you what the old code was doing? Maybe that could be a memory trick--make all your code blare loud music right before leaving on vacation
idle owl
Hah! Or the worst prank ever. "Want to hate your past self? Check out this video!"
proven garnet
<@!135415127879974912> I replied to the FeatherWing PR, if you want to integrate...
Sounds great, thanks!
idle owl
Sounds great, thanks!
Excellent, thank you! It'll be a bit of work, but a lot less work than rebasing your PR to match the updates. (Well I find rebasing difficult anyway, not everyone would agree 😄 )
@ember iris You still around? I may have done something wrong in one of the PRs. Dylan and I are now second guessing what the import foo as bar should be turned into with from foo import bar..... I thought I had it right, but she's been doing it differently...
ember iris
<@!459119350851567626> You still around? I may have done something wrong in one ...
Give me a minute to check it
idle owl
Thanks
Sorry for use of foo, heh. I try to avoid it as it's incredibly unclear to beginners what's going on with it, but I didn't have time to come up with more descriptive things.
ember iris
No worries--is there a place where the conversation is happening on the side as well--I need to double check but I think both options are valid, I need to figure out which one is more 'correct'
idle owl
Huh. Ok
I will be happy if that's the case
(I'll still fix mine if needed)
@proven garnet Your Git fu is great! Nicely done on merging branches in that PR!
ember iris
Give me a while, I want to make sure I understand both options correctly before I say anything. I have a gut feeling, but I want to be right because I know one of the options uses less memory, and I want to know if the second does as well
idle owl
Give me a while, I want to make sure I understand both options correctly before ...
OK, sounds good. Dinner will be here in the meantime, so tag me when you decide you're happy with the answer, please!
ember iris
Will do!
proven garnet
<@!135415127879974912> Your Git fu is great! Nicely done on merging branches in ...
Haha thanks! Gotta constantly remind myself to check upstream and the like 
proven garnet
took a few merges but I got it!
idle owl
@proven garnet Excellent job! PR merged 🙂
idle wharf
You might be gone for the day.. but @idle owl when reviewing library PRs like the linting ones... if we approve should we also go ahead and merge it or leave to the PR submitter to merge?
idle owl
You might be gone for the day.. but <@!330227457296957440> when reviewing libra...
Ok, so in this case, hold off, simply because I'm not sure which one of us has the import style correct on the ones that are left. However! Usually, you would absolutely go ahead and merge the PR. I prefer not to merge my own PRs, and prefer not to have others do the same. In the case of Limor reviewing, she often approves and leaves it to us to merge - she gets to do it how it works for her. 🙂 But that's not what I try to do otherwise.
The other thing is, if the author isn't paying attention, and no one merges it, then it sits. Which also isn't great. At the moment, Dylan's doing a bunch, so it would definitely sit.
idle wharf
Ok, so sounds like linting ones are under some sort of organized process and I'll skip those.
idle owl
Ok, so sounds like linting ones are under some sort of organized process and I'l...
Don't skip them, I need the help 😄 - the only organisation right now is that I'm waiting on someone to get back to me on how to make import into a from properly.
idle wharf
OK, I can help. I finished all my work PR Reviews so I'm sort of in the mode to do some more
idle owl
Because either Dylan's got it right, or I've got it right, or both of us have it right. And I might need to fix some PRs.
ember iris
Ok, so in this case, hold off, simply because I'm not sure which one of us has t...
On that--At the end of the day it appears to be functionally equivalent. I think it comes down to a style choice. But I need to write a test to see if one uses more memory than the other and that's not super in my wheelhouse so it's slow progress
idle owl
On that--At the end of the day it appears to be functionally equivalent. I think...
That's ok! We'll go with Dylan's for now since she did so many, and we can always grep it and update it later if needed.
@idle wharf That's what I needed to hear! Review away!
idle wharf
At the end of day its just text files... can totally update them again
idle wharf
That's ok! We'll go with Dylan's for now since she did so many, and we can alway...
So DO merge them then ?
idle owl
So DO merge them then ?
Yes please!
hoary moat
Does anybody know of any CP examples using the RMT peripheral on espressif boards?
idle wharf
Tip for those who do work from GitHub, you can filter out Adafruit PRs when you're looking at your PRs that need review.
crimson ferry
Does anybody know of any CP examples using the RMT peripheral on espressif board...
looks like pulseio uses it (based on grep of common-hal)
I don't think it is directly exposed to user CircuitPython code
hoary moat
idle owl
Tip for those who do work from GitHub, you can filter out Adafruit PRs when you'...
My search is a bit different, but same concept. Good suggestion!
idle wharf
Suggestion Box 🗳
Adafruit already knows what you've ordered (from their store) why not invite them to opt-in for library update notifications and\or to help to test PRs ?
Hmm... that sticker happened automagically
ember iris
OK, sounds good. Dinner will be here in the meantime, so tag me when you decide ...
I'm going to keep poking at it, but to my understanding the difference between the two imports is fairly minor and it evaluates to the same thing (just maybe with a slightly different number of steps).
https://docs.python.org/3/library/functions.html#__import__
There's a minor chance one option is better than the other from a memory usage point of view, but I can't tell.
I'd suggest one import is selected over the other just for style consistency
I think from foo import bar is better than import foo.bar as bar but I'm staunchly in one camp over the other.
Thoughts?
blissful pollen
I have the glasses running native with displayio! Runs the same as the rgbmatrix or sharpdisplay using framebufferio.
idle wharf
I'm going to keep poking at it, but to my understanding the difference between t...
I agree the from package import module is best syntax to teach people to use. But if there was some memory benefit that’d be cool to know about.
stoic rain
I'm trying to have a pin on a SAMD51 board go high when the board starts, and stay high. It's needed to get the external flash on that board to work. Would this do it if added to the appropriate board.c?
void board_init(void) {
gpio_set_pin_direction(PIN_PB22, GPIO_DIRECTION_OUT);
gpio_set_pin_level(PIN_PB22, true);
gpio_set_pin_pull_mode(PIN_PB22, GPIO_PULL_OFF)
}
gilded pewter
It will be great if we could get this for circuit python. Then the feather sense could be used more easily https://learn.adafruit.com/how-to-fuse-motion-sensor-data-into-ahrs-orientation-euler-quaternions/overview
blissful pollen
It will be great if we could get this for circuit python. Then the feather sense...
I wrote an AHRS module in the community bundle. https://github.com/gamblor21/Gamblor21_CircuitPython_AHRS
It can be a bit slow as it is in pure python but it worked
gilded pewter
I wrote an AHRS module in the community bundle. https://github.com/gamblor21/Gam...
first of all Thanks for the contribution. When you say slow, how slow would you say?
manic glacierBOT
manic glacierBOT
blissful pollen
first of all Thanks for the contribution. When you say slow, how slow would you ...
Depends what else your code is doing and how accurate you want it to be. Honestly it has been a while since I touched the code
dusk mauve
I’ve started looking into helping out by addressing some issues with I2C device drivers but the one area I’m still foggy on is the best pattern(s) for register read/write and bit manipulation.
Does anyone have recommendations for recently written/updated drivers that can provide good examples of the latest, most CircuitPythonic patterns for register read/write and bit manipulation type stuff? Or is their some doc on standards I should peruse for this? I’d like to avoid perpetuating bad habits or old, deprecated patterns
ornate breach
I’ve started looking into helping out by addressing some issues with I2C device ...
You might look at any of the circuitpython driver libraries on GitHub. Usually methods will be tuned towards the specific device it’s set up to support. But overall the general template is the same.
dusk mauve
You might look at any of the circuitpython driver libraries on GitHub. Usually m...
I’ve been browsing a lot of them as I’ve been tinkering with the pile of Stemma-QT boards I’ve accumulated. Thanks for that bit of context, helps explain why there are things that look like inconsistencies at first glance - makes sense that things would need to be customized a bit for each device’s quirks.
ornate breach
Yeah, it can be confusing especially when you go and write your own driver libraries for circuitpython
I’ve written a few for boards I’ve developed and it’s tricky sometimes
dusk mauve
Yeah, it can be confusing especially when you go and write your own driver libra...
I can see how picking a register manipulation pattern could be a difficult part of writing a fresh driver. It’s a skill I’d like to pick up, eventually. I’ve got a bin full of random I2C sensors/devices I’ve accumulated or salvaged over the years that might be fun/educational to write a proper driver for.
At the very least it’d give me more excuses to break out the logic analyzer and get deeper into the intervals of the base libraries 🙂
ornate breach
Definitely a worthwhile endeavor. Adafruit has done great in taking care of the i2c messages, where all we have to do is make sure we are writing to the right registers with the right size data. 🙂
And then reading the right registers with the right buffer size
dusk mauve
And then reading the right registers with the right buffer size
Right-sizing buffers (and not declaring large ones for lesser-used features unless they're needed) was something that came to mind while I was testing the APDS9960 driver before/after adding type annotations for this PR.
There's a 129-bit bytearray that is only instantiated if a certain feature (gesture recognition) is enabled and it was making my linters angry.
That and a PR I cooked up last night for the SparkFun Qwiic Twist (basically their take on a Seesaw-esque rotary encoder) got me interested in learning more about best-practice patterns around registers.
https://github.com/adafruit/Adafruit_CircuitPython_APDS9960/pull/36
https://github.com/fourstix/Sparkfun_CircuitPython_QwiicTwist/pull/7
ornate breach
Yeah, best practices seem to vary by chip manufacturers
Some are okay with 8-bit value registers while others want a full 16-bits
Texas Instruments uses a lot of 16-bit registers in their i2c devices (2x 8-bit registers) and Microchip does too
There is seems to be no true consistency except between similar parts numbers
dusk mauve
Yeah I did notice some interesting variations on register structure/size while reading datasheets. I2C might be a healthy-ish standard but reading/writing registers is a wild world.
ornate breach
Yeah, even on one device. Some values are only 8-bit and others are 16-bit and an 8-bit value like a status register might be right next to a 16-bit input value register.
Or the ADS7138 is a good example of weird I2C registers
dusk mauve
Haven't seen that one before, looks like a pretty neat SAR ADC. Was that one you wrote a driver for?
ornate breach
I’ve been working on the driver off and on for some time
Haven’t lately. I’ve got a partially working Arduino driver for it but the circuitpython driver is very much a work in progress
dusk mauve
So you've been down this road before, wrestling with registers. 😂
At least the APDS-9960 looks like its almost all 8-bit registers, with the 16-bit ones divided into high/low bytes. Some useful data is spread across a bunch of registers, of course, but at least it looks the read/write/read-and-write is all one byte at a time.
dusk mauve
Ahhh, now I see.
lone sandal
tulip sleet
I'm trying to have a pin on a SAMD51 board go high when the board starts, and st...
Check out *never_reset* in the board.c code for various boards, if you are reserving that pin for your own use. But if you are using regular QSPI or SPI flash memory, you should just be able to use the flash pins the way existing boards do.
We tie any flash chip CS lines to the right polarity, because we are using an exclusive QSPI or SPI bus to talk to the chip.
stoic rain
We tie any flash chip CS lines to the right polarity, because we are using an ex...
That makes sense. Unfortunately, it's about the hardware of the flash and the MCU (this is to fix the issue with external flash on the MicroMod SAMD51). The WP and HOLD/RESET pins on the flash chip are often just tied to 3.3V, but on that board, they're attached to PB22/PB23. To get the chip to work right, those pins have got to get pulled high. Really it's just PB22, but having both HIGH is ideal. Since they're not broken out and are dedicated to the flash chip, using never_reset makes sense.
The flash pins work as expected, if WP and HOLD are high.
west hinge
I just want to share how to set the ESP32 hostname using circuitpython on a MatrixPortal M4 device. Here is code snipet on how to set the hostname such that when viewed on router you can easily see which device it is.
from adafruit_matrixportal.network import Network
_SET_HOSTNAME_CMD = const(0x16)
SOME UTILITY FUNCTIONS AND CLASSES ---------------------------------------
def wifi_setHostname(NETWORKobj, hostname):
""" Sets the hostname of the ESP32 """
WiFiobj = NETWORKobj._wifi
ESPobj = WiFiobj.esp
resp = ESPobj._send_command_get_response(_SET_HOSTNAME_CMD, [hostname])
del ESPobj
del WiFiobj
if resp[0][0] != 1:
raise RuntimeError("Failed to set ESP32 hostname")
NETWORK = Network(status_neopixel=board.NEOPIXEL, debug=False)
myHostname = bytearray('xxx_MatrixPortalStockTicker_xxx')
wifi_setHostname(NETWORK, myHostname)
NETWORK.connect()
This code requires ESP32 firmware update 1.7.1 or 1.7.2 or higher.
crimson ferry
@west hinge With a minor change to the ESP32SPI library, NINA's setHostname could be exposed, e.g., esp.set_hostname()
but looks like there was a NINA fix in 1.7.4, so that may be the dependency for this approach
west hinge
yes, setHostname was exposed for arduino interface, but not circuitpython. yes, I was using 1.7.4 for this code.
crimson ferry
just tested it out with a ESP32SPI change, works fine with Station and shows up on the router, I'll do a PR
west hinge
That would be great. Thank you 
👍
crimson ferry
oh, there's already an open PR for this, held up by failed in black: https://github.com/adafruit/Adafruit_CircuitPython_ESP32SPI/pull/141
onyx hinge
I wonder if we can afford to break struct.pack('d') & struct.unpack('d') -- operating on double-precision values. They bring in double-precision arithmetic that costs over 400 bytes.
onyx hinge
https://github.com/adafruit/circuitpython/compare/main...jepler:space-savings-ideas?expand=1 we have some space savings available if/when we want to take it. I think taking the fat "mkfs" size savings would let us finish enabling rainbowio everywhere. The other commits in the stack all have trade-offs, modest regressions in how number formatting or binary packing work.
manic glacierBOT
onyx hinge
I think maybe my boot_out.txt changes broke safe mode 😦
ornate breach
boot_out.txt has changed its identity to break_out.txt and has acquired new skills.. breaking out of safe mode
tulip sleet
https://github.com/adafruit/circuitpython/compare/main...jepler:space-savings-id...
I guess that when we drop busio.OneWire completely in favor of onewireio in 8.0.0, we will gain enough space. We will also be dropping gamepadshift which will save space.
idle owl
I'm going to keep poking at it, but to my understanding the difference between t...
I'm with you on choosing the first one because that's how Dylan's been doing it overall, and I'm the one who did it the other way on two PRs. Much easier to bring everything in line with Dylan's choice than with mine.
idle owl
I have the glasses running native with displayio! Runs the same as the rgbmatrix...
Nice!
ember iris
I'm with you on choosing the first one because that's how Dylan's been doing it ...
I think both options are fine, and I still haven't found a reason one is better or worse than the other beyond style. It's such a weird edge case of style I've never once thought of it until you asked 😆
idle owl
I think both options are fine, and I still haven't found a reason one is better ...
If there's one thing I'm good at, it's finding edge cases. 😄
ember iris
Which is pretty great, because it'd be users running into them otherwise!
idle owl
I think both options are fine, and I still haven't found a reason one is better ...
https://github.com/adafruit/Adafruit_CircuitPython_BME280/pull/57 Done. That's the only one I did that involved that import style.
idle owl
Which is pretty great, because it'd be users running into them otherwise!
Valid point!
ember iris
https://github.com/adafruit/Adafruit_CircuitPython_BME280/pull/57 Done. That's t...
Since this import style has fewer characters, it seemed like the import over multiple lines might be skipped this time
idle owl
I tried 🙂
Everything fails, Black reformats it to this, and the Pylint disable has to be put back where it is.
ember iris
haha good to know
onyx hinge
Good morning / [time-zone appropriate greeting], <@&356864093652516868> ! Our weekly meeting is in about 80 minutes from now. This is shifted 1 hour from previously, as most of the US switched to standard/"winter" time yesterday. In any case, if you plan to participate or just have notes for us, please add them to the document: https://docs.google.com/document/d/13krd3FcC82ir_a0DBnJMaq0OEqJzY_bPgHg_5bhufWQ/edit -- if you will not be reading your own notes, please note [no mic], [missing meeting] etc, and I'll read them out when the time comes. I hope to see many of you in the CircuitPython voice channel!
dusk mauve
I've just started experimenting with building my own CircuitPython binaries for a Trinket M0 to make sure that only the things I need are in there and its been fun! But I've got a weird question that the docs didn't seem to have an answer for.
Even though I've added in CIRCUITPY_PIXELBUF = 1 in the mpconfigboard.mk it doesn't look like its showing up in the generated moduledefs.h or being compiled into the final binary. Is this library being legitimately removed somewhere along the line, like maybe its just not something the SAMD21 can handle?
idle owl
I've just started experimenting with building my own CircuitPython binaries for ...
We have it on two of the Trinkeys, so, no, you should be able to use it.
onyx hinge
apologies, I think I managed an off-by-one-hour error in my message above <@&356864093652516868> ! We've still got 2 hours until the meeting. @idle owl can you just confirm that I'm correct this time?
dusk mauve
We have it on two of the Trinkeys, so, no, you should be able to use it.
Ahh, yeah, I see it enabled there in the configs for the trinkeys. Interesting. Diving into the build plumbing now 😄
idle owl
Ahh, yeah, I see it enabled there in the configs for the trinkeys. Interesting. ...
I'm not sure if it's being removed elsewhere, though.
misty garnet
I'm putting .py code in the modules directory but it doesn't seem to be putting them in the build code as frozen modules
idle owl
apologies, I think I managed an off-by-one-hour error in my message above <@&356...
2 hours, yes!
onyx hinge
thanks
thorny jay
That seems right now. The time divergence between EU and US was last week and only last one week (from memory).
misty garnet
do I need to include a flag in the mpconfigboard to freeze code in the ~/modules directory?
onyx hinge
@misty garnet I don't think we use modules/ for freezing, we use frozen/ and yes each thing to be frozen is listed explicitly. e.g., ```make
Include these Python libraries in firmware.
FROZEN_MPY_DIRS += $(TOP)/frozen/Adafruit_CircuitPython_CircuitPlayground
that's from ports/atmel-samd/boards/circuitplayground_express/mpconfigboard.mk
@dusk mauve my first suggestion would be to make clean, i'm not sure the dependencies are tracked all the way from mpconfigboard.mk, so merely changing that file might not make everything rebuild as needed.
dusk mauve
<@!253692424772124673> my first suggestion would be to `make clean`, i'm not sur...
yeah, my builds are being kicked off with a make clean && make -j4 BOARD=trinket_m0 V=1 line. Did a make clean in the project root for good measure as well.
Are there other things I should know about for build-to-build cleanup?
onyx hinge
not really, make clean should get everything.
I made this change locally:
--- a/ports/atmel-samd/boards/trinket_m0/mpconfigboard.mk
+++ b/ports/atmel-samd/boards/trinket_m0/mpconfigboard.mk
@@ -9,3 +9,5 @@ CHIP_FAMILY = samd21
INTERNAL_FLASH_FILESYSTEM = 1
LONGINT_IMPL = NONE
CIRCUITPY_FULL_BUILD = 0
+
+CIRCUITPY_PIXELBUF = 1
``` and the build doesn't fit anymore so it seems to me that this had the expected effect.I've ripped out a bunch of stuff to get a max of 15,000 bytes free. I looked at the code for adafruit_pixelbuf in shared_bindings and shared_modules and it doesn't look like it requires any of the stuff I've removed, but it seems like one or more of those could be the culprit?
Here's the full mpconfigboard I'm working with. I've got tag 7.0.0 checked out too btw.
USB_VID = 0x239A
USB_PID = 0x801F
USB_PRODUCT = "Trinket M0"
USB_MANUFACTURER = "Adafruit Industries LLC"
CHIP_VARIANT = SAMD21E18A
CHIP_FAMILY = samd21
INTERNAL_FLASH_FILESYSTEM = 1
LONGINT_IMPL = NONE
CIRCUITPY_FULL_BUILD = 0
# Original Flavor: 185,928 / 2,232 free
CIRCUITPY_USB_MIDI = 0 # 183,932 / 4,228 free / 1,996 saved
CIRCUITPY_USB_HID = 0 # 180,188 / 7,972 free / 3,774 saved
CIRCUITPY_ROTARYIO = 0 # 179,036 / 9,124 free / 1,152 saved
CIRCUITPY_RTC = 0 # 177,612 / 10,548 free / 1,424 saved
CIRCUITPY_TOUCHIO = 0 # 176,292 / 11,868 free / 1,320 saved
CIRCUITPY_PWMIO = 0 # 173,796 / 14,364 free / 2,496 saved
CIRCUITPY_ONEWIREIO = 0 # 173,160 / 15,000 free / 636 saved
#
# MAX SAVED/FREE: 12,768 / 15,000
#
CIRCUITPY_PIXELBUF = 1 # no change? must already be included?
# CIRCUITPY_KEYPAD = 1 # TODO
# Uses 1,872 bytes
# FROZEN_MPY_DIRS += $(TOP)/frozen/Adafruit_CircuitPython_NeoPixel
I'm near the tip of main fwiw
dusk mauve
is main basically the develop branch?
onyx hinge
yeah, right now it's the version that will become 7.1.
dusk mauve
yeah, right now it's the version that will become 7.1.
That's helpful, I'll just continue building from there I guess.
For what its worth, when I build adafruit_slide_trinkey_m0 on 7.0.0 the pixelbuf stuff gets included as expected in the generated moduledefs.h. But not when I build trinket_m0
onyx hinge
At 7.0.0, same behavior for me. That patch above causes the build to not fit.
I mean, I see the same behavior at 7.0.0 and at main
dusk mauve
Huh interesting. What's the expected behavior if the build doesn't fit? Just doesn't get added to the generated header? I'm more used to a failure toward the end but I've mostly worked with projects build around PlatformIO
onyx hinge
```/home/jepler/.local/gcc10/gcc-arm-none-eabi-10-2020-q4-major/bin/../lib/gcc/arm-none-eabi/10.2.1/../../../../arm-none-eabi/bin/ld: build-trinket_m0/firmware.elf section .text' will not fit in region FLASH_FIRMWARE'
ooooh here's one thing ... build BOARD=trinket_m0 clean .. specify the board name when you clean
may / may not matter
I mean, it definitely changes what gets cleaned
$ make -f Makefile clean
Makefile:27: *** You must provide a BOARD parameter. Stop.
I would have expected you to see an error from this
dusk mauve
I'm doing make clean BOARD=trinket_m0 and it looks like its deleting the right things
yeah, got that error the first time I tried blindly 😄
onyx hinge
yeah, my builds are being kicked off with a `make clean && make -j4 BOARD=trinke...
OK above you'd pasted something else
dusk mauve
oh, yeah, I mistyped instead of copy/paste 😂 I'm using Barrier to "KVM" 4 systems including my Linux build ThinkPad and copy/paste doesn't always work coming from that system, much to my chagrin
Thanks for working with me on this @onyx hinge , I'm gonna retreat into the makefile and general build plumbing for awhile I think. Learned many years ago that it can be perilous to skip learning how that kind stuff works, even in a well-maintained project 🙂
After many years of wrangling Maven POMs this is a breath of fresh air 😂
onyx hinge
@dusk mauve best of luck! sorry I didn't have the instant answer
dusk mauve
<@!253692424772124673> best of luck! sorry I didn't have the instant answer
Don't sweat it, its a weird question/issue! I really appreciate the discussion, haven't had this kind of back-and-forth with experienced devs on other projects I've worked with/on. Makes a huge difference ❤️
onyx hinge
I have a feeling you'll come back with a solution and we'll say "oh of course"
"it's obvious in retrospect what was happening"
dusk mauve
"Its not about why the build failed, its about the friends we made along the way."
onyx hinge
oh and I should say my moduledefs.h begins
// Automatically generated by makemoduledefs.py.
#include "py/mpconfig.h"
#if (CIRCUITPY_PIXELBUF)
extern const struct _mp_obj_module_t pixelbuf_module;
#define MODULE_DEF_MP_QSTR_ADAFRUIT_PIXELBUF { MP_ROM_QSTR(MP_QSTR_adafruit_pixelbuf), MP_ROM_PTR(&pixelbuf_module) },
#else
#define MODULE_DEF_MP_QSTR_ADAFRUIT_PIXELBUF
#endif
``` (build-trinket_m0/genhdr/moduledefs.h) so by looking at that file and not seeing PIXELBUF anywhere you do seem to be looking at a relevant spotoooohhhhh another thing
@dusk mauve makefiles are awful, and you've fallen into a trap
slender iron
@dusk mauve I always do V=2 on a make command to see the final result for adding a module. the CIRCUITPY_* will be part of the long command line
onyx hinge
X = y # z and X = y are different. The content of the CIRCUITPY_PIXELBUF variable needs to be exactly 1
remove the comment and any trailing whitespace on the line
dusk mauve
<@!253692424772124673> I always do `V=2` on a make command to see the final resu...
I've done V=2 and I can definitely see it in the compile command line entries. the flag just never ends up in the generated moduledefs.h 🙁
onyx hinge
in some .mk file we compare lots of variables for exactly the content "1" and make a decision based on it
mmmmaybe that's the problem here?
dusk mauve
Of course its trailing whitespace/comments in a .mk file. 😂 Me and my inline commenting habits, wreaking havoc in an unfamiliar parser yet again.
Removed all whitespace and comments from the file. BOOM. There it is.
I'm thoroughly confused why the preceding exclusions were 'working', maybe because 0 can have trailing whitespace and 1 can't? Parsers are mysterious things sometimes though, turtles all the way down from there.
TIL: Don't add dumb stuff to .mk files, keep those notes elsewhere 😄
onyx hinge
yeah something like that. If you look for truthiness as exactly the string "1", then "0", "0 " and "0 # I like to write comments" (not even actually sure which one make gives you) all lack truthiness
you can put it on its own line though
dusk mauve
"it's obvious in retrospect what was happening"
Nailed it. Thanks again!
idle owl
@ember iris Really appreciating my inbox blowing up! Early hug report!
ember iris
No problem, I figured I could work through them while eating breakfast. Props to Dylan for getting so many of these going!
Related to the pull requests--A lot of the modules have commented out code for alternative setups (spi vs i2c, etc)
Since it's getting commented out, I imaging it's not being evaluated by pylint--in the future should we look for that kind of code, comment it out, get it to pass pylint, then re-comment it out?
idle owl
Related to the pull requests--A lot of the modules have commented out code for a...
Sounds like a good issue to file. 🙂 But yeah, probably.
You mean in example code? Or libraries as well?
ember iris
You mean in example code? Or libraries as well?
...I think the only time I've seen code written like that is either in the example code or in the docstrings. The example code needs to pass pylint as well right?
idle owl
...I think the only time I've seen code written like that is either in the examp...
Yes! I was simply making sure we were on the same page.
ember iris
Ok cool then yes, same page. I was mentally going, "Where else might I have seen it" 😅
idle owl
Ok cool then yes, same page. I was mentally going, "Where else might I have seen...
No worries, it only made sense in examples. But yeah, all the code should pass. Pylint ignores comments, so eventually we need to fix that code as well.
tidal kiln
@idle owl ping. got read the docs question.
idle owl
<@!330227457296957440> ping. got read the docs question.
Sure. What's up?
tidal kiln
do libraries with multiple .py's in a folder need anything special?
tidal kiln
the "API Reference" section is missing for some, ex:
https://circuitpython.readthedocs.io/projects/magtag/en/latest/
tidal kiln
thanks!
tulip sleet
@slender iron the current BLE HCI is only peripheral, so it would need to be fleshed out for central
idle owl
thanks!
It's failing remotely because it can't find libraries.
Hmm.
I wonder if the update also broke things. Like maybe we need to include the other requirements in requirements.txt in the docs/requirements.txt
Blergh.
tidal kiln
blergh is all i can offer as well 🙂
idle owl
Yeah I broke it.
idle owl
So, this won't be an Adabot fix. This will be a manual fix. Though it seems automatible.
@tidal kiln Basically without the other files being installed, the docs don't know what things are. They were using the main requirements.txt file, but I had to change it to include Sphinx to get RtD to run the latest Sphinx version which fixes some docs issues. But I have apparently made more in the process.
So! Dylan has another task ahead of her. 😕
manic glacierBOT
tidal kiln
@idle owl blergh indeed
solar whale
progress is not always forward 😉
tulip sleet
thanks!
@slender iron re naming of "bare metal" Rpi CPy, maybe "OS-less" or "no OS" are alternative names. There could indeed be a CPy that runs under Raspberry Pi OS, just as MicroPython has a Unix port, so need to differentiate from that
modern wing
Happy Monday y'all -- enthusiastically lurking 🙂
mental nexus
Lurking y’all. 👋
slender iron
@tulip sleet (we're in the voice channel)
Most CP installs don't have an OS though. We could call CP unix something special instead
proven garnet
Listening in (no chat)
modern wing
On top of her game 🙂
prime flower
Lurking only
tulip sleet
<@!329766224093249548> (we're in the voice channel)
Most CP installs don't have an OS though.
Yeah, I just mean emphasize that for RPi
thorny jay
Hi
idle owl
@thorny jay That ATtiny one was difficult, thanks!
lone axle
Ahhhh
slender iron
(it's their website)
this is why folks say to use camel case in hash tags KlarDotSH
thorny jay
<@!681528999520108564> That ATtiny one was difficult, thanks!
Wait for the next read out. 😉 I have an unused eeprom (because there is not CP support I believe) https://www.adafruit.com/product/5146 but in the seesaw guide there is a eeprom emulation https://learn.adafruit.com/adafruit-attiny817-seesaw/eeprom so I have hope to use that code to make some use of the 24LC32 I2C EEPROM. Not sure it will work, but the code should be a good base.
slender iron
good work!
idle owl
😄 All good!
slender iron
you two are doing a lot of good work 🙂
mental nexus
Awesome work on scalable widgets @errant grail
errant grail
Awesome work on scalable widgets <@339922377746022402>
Thanks. Learning a bunch. Think I may have sprained my coding muscle in the process.
idle owl
you two are doing a lot of good work 🙂
Thanks. 🙂
slender iron
@inland tusk probably after SD cards are going and the first version is merged in
idle owl
@proven garnet Nice to see you in the meeting! Thanks for all the work you've been doing on the libraries!
lone axle
I think I've run into this issue before as well with readthe docs using an ancient version of causing differences between what I see locally.
modern wing
Thanks everyone!
gilded cradle
Thanks everyone
modern wing
Yep, it shows "live in 23 hours" Nov 9th @ 2:30pm eastern to me
thorny jay
You can click on "remind me" in YouTube.
idle owl
@lone axle If you find the time, please remove Hacktoberfest label from the last 22 repos, most of which are on the CP core repo.
lone axle
<@!382939733107408897> If you find the time, please remove Hacktoberfest label f...
Will do
ember iris
thanks all!
idle owl
@lone axle Do you know how to test cookiecutter? Do I run it and specify . ?
Seems to be doing something if I do that.
Ok, it ran, but it didn't do what I need it to. I am going to need some help, I think. Trying one more thing.
Oh wait. Ugh. I think I know what I did.
lone axle
<@!382939733107408897> Do you know how to test cookiecutter? Do I run it and spe...
yep you can specify a local path instead of the repo slug. I have a directory that I run it in and usually do something like cookiecutter ../cookiecutter-adafruit-circuitpython/ if you use just . it will probably dump the result in your cookie cutter repo
onyx hinge
brrrrr the cold part of the year is coming to my basement
lone axle
I keep finding more nifty arguments for the github CLI. just realized you can filter by label with it gh issue list --label Hacktoberfest you can also use --json and specify the fields you want in your JSON objects that get returned, which makes it super easy to chain the results into something else.
idle owl
Neat!
@ember iris Are you done PRin' it up for a bit?
I don't want to clobber your work if you're still going through things.
GitHub will let everyone approve at the same time, so it doesn't keep you from duplicating work. 😄
onyx hinge
Here is the notes document for 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/1ROYIKjHd8lfuKZJEgcWpzsnfbV4AaZjyvmRipSjxKXs/edit?usp=sharing
idle owl
@lone axle @gloomy shuttle With the voice channel only allowing circuitpythonistas to talk, I think you're pretty safe remaining connected, but in the event you forgot you were, you're both still in the CircuitPython voice channel. A circuitpythonista could still surprise either of you if you weren't expecting audio. 🙂
I disconnected you, jfurcean. I don't want to ping you again though.
solar whale
@gilded cradle FYI -- just put blinka on a Raspberry Pi Zero 2W running bullseye --- blinkatest.py works !!
Next up is trying bullseye with braincraft....but probably not today ...
idle owl
@tidal kiln I fixed the MagTag library documentation for testing purposes - the docs are now building properly. We're going to apply the fix across the board tomorrow.
Here is the cookiecutter PR update to get everything in line for future libraries to match the behavior of current libraries (once patch is applied to current libraries): https://github.com/adafruit/cookiecutter-adafruit-circuitpython/pull/160
slender iron
I just got interrupts going on the zero 2w and uart and usb now work!
searches for a mini hdmi
mini hdmi works too
ember iris
<@!459119350851567626> Are you done PRin' it up for a bit?
Sorry, needed to step away and take care of some other stuff. I'm still up for working through but looks like you got a lot of them 🙂 Thank you!
idle owl
Sorry, needed to step away and take care of some other stuff. I'm still up for w...
No worries! I simply didn't want to duplicate work. There's one left that you had questions on - if you could take care of that one, I'd appreciate it.
Because I also want to know the answer to that one since we did a number of libraries that way, I think.
ember iris
No worries! I simply didn't want to duplicate work. There's one left that you ha...
That's the first one I was going to take a look at
I don't think it's an issue, I just wanted to make sure I understood 'why' it was being done
*duplicates the above comment to git*
idle owl
Ah ok, thanks! I asked her the same thing on an earlier PR (or maybe in Slack?), and thought her explanation was well thought out and figured it was good.
Good to know
@ember iris There's also this one for you if you like: https://github.com/adafruit/cookiecutter-adafruit-circuitpython/pull/160
tidal kiln
@idle owl thanks for docs fix
ember iris
No worries! I simply didn't want to duplicate work. There's one left that you ha...
Do you have the ability to test something for me? I can fake a matrix keypad, but I want to reduce the chance that I mess two things up at once
Ah it's late--I'll try to get it going on my end.
idle owl
Do you have the ability to test something for me? I can fake a matrix keypad, bu...
Yep it's dinner time here 😄 if you can't get it sorted, I can take a look tomorrow. I am pretty sure I have hardware for it.
ember iris
Yep it's dinner time here 😄 if you can't get it sorted, I can take a look tomor...
I think I got my test working, so I'm actually good! (I got it printing button presses less than a minute ago too--awesome timing!)
idle owl
Nice!
ember iris
Can I request you test the change I'm going to request tomorrow just for my sanity's sake?
idle owl
Can I request you test the change I'm going to request tomorrow just for my sani...
Sure, tag me on it and if I can find the hardware, I'll test it. (Should be able to - I think I can picture where it is.)
solar whale
@gilded cradle hmm - trying to setup braincraft under bullseye, but bullseye does not appear to support raspistill or raspivid....
gilded cradle
<@!478616919721771019> hmm - trying to setup braincraft under bullseye, but bull...
Thanks, good to know. Currently I'm testing PiTFT on latest Raspberry Pi OS with mixed results.
Do you think you need to manually install them or is there an error?
solar whale
not sure yet .. there is a new camera utility call libcamera ... not sure how to use it yet.
solar whale
from the bullseye announcement ```New camera driver
The driver used by Raspberry Pi to access camera modules has now been replaced with libcamera, a standard Linux API. As with the video driver change, this means less closed-source proprietary code, and makes it easier for third parties to develop new camera hardware and software.
For anyone who has written camera-based applications in the past, this is a very significant change, and is too large to be covered here. There will be a blog post dedicated to the new camera driver published soon, so watch this space!```
gilded cradle
That's probably good news since it may mean this could work on non-raspi boards
Of course it will probably mean updating guides
solar whale
gilded cradle
Thanks
solar whale
I tried libcamera-still and it does not display to the braincraft hat screen. I think I'll need to try this first on a full desktop .
gilded cradle
Ok
solar whale
the good news is libcamera-still -o test.jpg captures an image just fine ...
gilded cradle
Nice. I think the problem I'm dealing with is more complicated than I thought...
solar whale
I uninstalled pitft for this test
gilded cradle
I don't think the fbcp stuff can work without a rewrite on the desktop version of pi.
solar whale
I have lite on my braincraft. the pitft install did not error . What did you test it with?
solar whale
ok but you can't get anything to display can you?
gilded cradle
i do in lite, just not on desktop
solar whale
waht are you using to display in lite?
gilded cradle
Same hardware,
solar whale
what program?
solar whale
oh yes, I see the console
gilded cradle
Just whatever it would display on hdmi
solar whale
but no way to get camera to display.
gilded cradle
I haven't gotten that far
solar whale
have you tried tensorflow lite demo?
gilded cradle
No. I was just seeing the extent of the breakage of PiTFT. Spent most of my time waiting for SD cards to be written.
solar whale
I assume it needs raspistill... so likely broken
gilded cradle
I'm not sure, I haven't tried. You're way ahead of me there.
solar whale
Good to know we are seeing the same thing ... good luck.
solar whale
this is not a good sign ... sigh ERROR: tensorflow-2.3.1-cp37-none-linux_armv7l.whl is not a supported wheel on this platform.
hoary moat
Pi zero 2w arrived. I ❤️ Adafruit. (The other things in the box are also fun)
hoary moat
No. I was just seeing the extent of the breakage of PiTFT. Spent most of my time...
SanDisk Extreme has the fastest write speeds I’ve found so far. Not extreme plus. Not ultimate. Be careful when buying. Also 64 GB seems to be a sweet spot of write speed.
spiral elk
I've been using Sandisk A1 Ultra pretty much exclusively for everything since Samsung stopped making A1-rated cards
hidden oxide
this is not a good sign ... sigh `ERROR: tensorflow-2.3.1-cp37-none-linux_armv7l...
I've had trouble with the Tensorflow wheel as well
I do believe my problems were with CUDA dependencies though
slender iron
anyone want to do a 7.1.0-beta.0?
ornate breach
Sure
idle owl
@ember iris I am pretty sure I know the location of a 1x4, can I fake it for the 4x4 test? Or will my "fake" test be only as good as your "fake" test? 🙂
ember iris
<@!459119350851567626> I am pretty sure I know the location of a 1x4, can I fake...
I think it should work. I'm pretty confident that two enumerates can be used while toggling the pin values without issue. I'm just not 100% and didn't want to make a big, "Make a change" while missing something obvious
idle owl
I could drag Dan into it 😆 - I know he has multiple keypads.
ember iris
but all of the test code snippets seemed to work
idle owl
Fair enough!
ember iris
To the readthedocs yaml--it looks good, but I don't know anything about the sphinx/readthedocs build process so I'm not sure what to look for. Anything specifically I should keep an eye out for?
idle owl
To the readthedocs yaml--it looks good, but I don't know anything about the sphi...
Nope, that one is simple. Add a requirements file to docs, and tell Read the Docs to use it.
We're updating cookiecutter to do the same so all libs have the same files.
ember iris
That's convenient!
idle owl
@tulip sleet Can I bring you in on the MatrixKeypad linting PR (it's not ready yet, I would tag you when it is so you don't have to pay attention for updates)? Keith suggested a change that we want to test, but I don't know where my 4x4 is (probably in the US), and we need the matrixkeypad_4.4.py example tested. I know you have a bunch of matrix keypads.... If you're too busy etc, I can fake it with my 1x4 keypad that I think I have here. But I figured I would ask.
idle owl
To the readthedocs yaml--it looks good, but I don't know anything about the sphi...
It's actually good that you waited, because I had forgotten the .txt file. 😄 It's ready now though.
tulip sleet
<@!329766224093249548> Can I bring you in on the MatrixKeypad linting PR (it's n...
sure, I have several of the telephone-style ones
idle owl
sure, I have several of the telephone-style ones
Perfect. I'll tag you when it's ready for review.
tulip sleet
anyone want to do a 7.1.0-beta.0?
I can work up the release notes; it should only take an hour or so
@ornate breach or if you want to do the release notes, just copy a sample beta one from a previous release and do it up in the same style in markdown, and I will finish it off
main thing to do is to find the last PR after the 7.0.0 release and then summarize all the PR's until the present
ornate breach
<@728676724522352733> or if you want to do the release notes, just copy a sample...
I can take a gander this evening
tulip sleet
are you interested in this? 🙂 not try to push you. The PR titles are not necessarily the best summary, so there's a bit of rewriting of the goal of a PR
viral mist
hey, @slender iron recommended me this server/channel, because i asked for some help on the pi pico discord.
im trying to port my project (https://github.com/freeyourstream) to the pi pico and i am struggling with the cdc-serial/hid composite device tinyusb configuration. It's my first time go this "low level" with usb on a microcontroller.
Im trying to use the circuitpython code as an example because @slender iron told me it contains tinyusb composite device code. now i cant find it. I had no problem finding the corresponding files in the micropython repo, so what am i missing?
Ps: anybody here willing to help me/guide me with my tinyusb problem more directly?
Thx and have a nice evening 🙂
idle owl
@tulip sleet https://github.com/adafruit/Adafruit_CircuitPython_MatrixKeypad/pull/14 is ready for testing, when you have a moment.
slender iron
if anyone wants to try pyleap, here is the testflight link: https://testflight.apple.com/join/cUiVF4iy
hidden oxide
I've had trouble with the Tensorflow wheel as well
NVM, that was pytorch
I also had problems with the Tensorflow wheel though
solar whale
if anyone wants to try pyleap, here is the testflight link: https://testflight.a...
once installing pyleap and the latest build of CP, do I need to install anything on the CPB, My iPhone is not seeing it...
nevermind -- got it...I missed the RESET while blue step...
very cool -- 'blink' works
as does the rainbow.... woohoo!
Is the glasses test also only for the CPB? or will it work on the glasses driver board?
slender iron
@solar whale I'm not sure. I don't have a driver board
I don't think pyleap actually checks but the code.py might have specific pins in it
solar whale
It just worked fine on the driver board!!
slender iron
great!
the other app, file glider, should be available through test flight this week too
slender iron
The app side is all trevor 🙂
solar whale
I don't think pyleap actually checks but the code.py might have specific pins in...
That demo is just I2C. Very cool.
onyx hinge
anyone want to do a 7.1.0-beta.0?
I'd really like to have https://github.com/adafruit/circuitpython/pull/5557 in the next release, it'd be unfortunate to go out without a working safe mode
slender iron
I was letting @tulip sleet do the review
onyx hinge
ok, I was worried it was being overlooked
manic glacierBOT
ornate breach
are you interested in this? 🙂 not try to push you. The PR titles are not necess...
Do I need to be part of the circuitpython librarians group?
I’ve never formally asked to be added/nor has it been brought up but I’m willing to lend a hand when I can
idle owl
Do I need to be part of the circuitpython librarians group?
If it makes sense, we can add you directly to the CircuitPython core repo. The CPLibians team is designed to provide review access to the libraries.
ornate breach
Okay, whatever I need to be helpful 🙂
Not sure what I’d need to help prep a beta release
idle owl
Not sure what I’d need to help prep a beta release
I'm not certain either, but if it involves access to the core, we'll get it going. Dan will be back later and can provide more information.
ornate breach
Sounds good 🙂 I have another 2.5hr at work so no huge rush
tulip sleet
Okay, whatever I need to be helpful 🙂
To write up the release notes, you don't need any privileges. I was going to do the actual release, because there are some things I need to do that require other privileges, not even on GitHub.
tulip sleet
👍🏻
I mean, just pass off the release notes to me when you're done. You can write them or paste them in a gist or something.
hidden oxide
Is it all right to put a new CP supporting board announcement in the Community News section of the CP meeting notes for next Monday?
onyx hinge
@hidden oxide the community news section is excerpted from the e-mail newsletter, so the thing to do would be to submit it there. That doesn't guarantee it gets picked up to be read during the meeting, though. Status updates are read during the meeting. https://github.com/adafruit/circuitpython-weekly-newsletter#contribute -- either contribute via github pull request, or e-mail cpnews@adafruit.com
idle owl
For the newsletter? I don't think Anne has that in her workflow... It would likely get missed.
onyx hinge
@idle owl did we ever specifically cover new boards in the meeting and/or newsletter?
we cover new libraries specifically, and maybe new blinka boards when they used to come along weekly-ish
ornate breach
I’ve seen some of my boards covered
idle owl
<@!330227457296957440> did we ever specifically cover new boards in the meeting ...
Not really, it's a section much further down in the newsletter that covers new boards.
ornate breach
But it’s rare
hidden oxide
Community newsletter for Python on Microcontrollers. Please feel free to put your current news, projects, etc. as a PR or Issue.
I just saw the "Issue" thing
idle owl
Oh hmm.
hidden oxide
It's this board - https://www.kickstarter.com/projects/udoo/udoo-key-the-4-ai-platform/
idle owl
We should fix that.
idle owl
`Community newsletter for Python on Microcontrollers. Please feel free to put yo...
Is this in the notes doc?
We can fix the template.
onyx hinge
@idle owl no, the description of the repo on github
hidden oxide
No, it's in the Github repo
idle owl
<@!330227457296957440> no, the description of the repo on github
Let me check with Anne then.
Maybe she does look for issues and I didn't know it.
onyx hinge
@hidden oxide the udoo key was in the newsletter that went out today fwiw https://github.com/adafruit/circuitpython-weekly-newsletter/blob/cf992d87f37f3344dd626268335550d597c63d8d/_drafts/2021-11-09-draft.md
The UDOO KEY is a fully programmable Raspberry Pi RP2040 + ESP32 microcontroller with Wi-Fi, Bluetooth and Bluetooth Low Energy. It also gives you access to an AI platform, Clea, with pre-trained AI models. Both microcontrollers can be programmed using different environments, such as TinyML, TensorFlow Lite, MicroPython, C/C++, and more - Kickstarter.
idle owl
No worries! At least we can find out if filing an issue works for the newsletter and update either GitHub or my brain.
idle owl
I'll let you know when I find out.
onyx hinge
If Anne prefers not to work via issues, we can update that text but also turn off the issues tab altogether
hidden oxide
OK, great
idle owl
If Anne prefers _not_ to work via issues, we can update that text but also turn ...
Right. That's what I was thinking.
onyx hinge
I'm going to step away, it's time to start cooking soup!
idle owl
I just opened a tab in Firefox, clicked back here to reply to you, and forgot why I opened the tab.
Ok, have a delicious one!
onyx hinge
or maybe it's more of a stew, there will be big chunks of veggies
onyx hinge
@hidden oxide we do appreciate you bringing a potential newsletter topic! keep 'em coming in the future too
ornate breach
We're having Vietnamese.
Pho way! That’s cool 🙂
idle owl
Do I remember we started suggesting using bitmap_label over label? Or was it the other way around... FoamyGuy would know, but he's busy tonight.
idle owl
If Anne prefers _not_ to work via issues, we can update that text but also turn ...
@hidden oxide Anne does check the issues, so that's on me for not being aware. Thanks again for keeping an eye out for things!
crimson ferry
what does this do: MP_STATE_VM(some_singleton)?
tulip sleet
what does this do: `MP_STATE_VM(some_singleton)`?
There is a struct of per-VM storage: that's a macro that's a reference into that struct.
crimson ferry
so you always get your correct object in and out of it
but it gets automnatically cleared sometimes?
(going into REPL, reload)
tulip sleet
that struct is traversed during garbage collection, to make sure that anything in it is marked as not garbage (and anything that transitively points from there)
when the VM is restarted, it starts fresh
there is an MP_STATE_THREAD() also, but we don't have threads turned on
or more precisely, there is only one thread
crimson ferry
ok, thanks, I'm trying to narrow down a hardfault, and it happens around or soon after that reference... sometimes. but still gathering info
and no issue if VM has been restarted, and the some code refers to that macro on the (non-existent) object?
hidden oxide
<@!746175873203437652> we do appreciate you bringing a potential newsletter topi...
Great - thanks!
hidden oxide
<@!746175873203437652> Anne does check the issues, so that's on me for not being...
Really! OK, great. Thanks for looking
tulip sleet
and no issue if VM has been restarted, and the some code refers to that macro on...
I think that struct will get cleared when the VM restarts
in order to to clear the unset fields
onyx hinge
This is weird, there are a lot more missing-prototypes diagnostics in the raspberrypi port. did I make an error in #5561 that made most of the files not actually built with that flag? Or is there a reason it could differ from samd to raspberrypi
I've looked at the Makefile a couple of times but didn't spot my error yet
tulip sleet
This is weird, there are a lot more missing-prototypes diagnostics in the raspbe...
due to -flto?
which is not enabled on raspberrypi
onyx hinge
ah, no, I got the makefile wrong as I initially suspected (once I started seeing more problems on raspi)
-$(patsubst $(SRC_ASF),%.c,%.o): CFLAGS += -Wno-missing-prototypes
+$(patsubst %.c,%.o,$(SRC_ASF)): CFLAGS += -Wno-missing-prototypes
The way I wrote it, the result of $(patsubst) was %.o, so it tacked that on for all object files, not just ones in SRC_ASF. ah, well, it means there's more to do, but it doesn't mean the stuff I fixed was wrong in any way.
time to call it a night!
manic glacierBOT
tulip sleet
@idle owl @trim elm I am working up this library: https://github.com/adafruit/Adafruit_CircuitPython_asyncio, which is not yet public, but will be soon. It is Python code that is a near copy of some code in the micropython repo (they freeze it in). The code is not formatted with black and does not come close to passing pylint, and probably never will, unless the MPy folks take that as a task. I would like to keep the code mostly the same so it's easy to merge changes.
To prevent failing pre-commit and CI checks, I would need to turn off black and pylint. Can I do this permanently, so that mass updates of the yaml files don't turn them back on? Do you have any other library exceptions like this? Thanks.
analog bridge
Is there an easy way to save about 24 bytes (specifically for ja language)?
tulip sleet
I think my comment above is the easiest
CIRCUITPY_ONEWIREIO = 0
right now there are two ways to get OneWire, and we can turn one off for now
it was meant to be a staged change, but for these boards, we can avoid that
analog bridge
After doing some more changes to the PR code, only arduino_nano_33_iot ja is overflowing by 24 bytes.
With other languages on this board there is room for at least another 216 bytes.
tulip sleet
there'll probably only be more increases (slight hopefully) pre-8.0.0, so turning off onewireio is still fine.
this board has a lot of pin names
trim elm
<@!330227457296957440> <@!362023826105827340> I am working up this library: <h...
@tulip sleet Yeah you should be able to. I think just removing the pre-commit section from build.yml should be enough
analog bridge
Thanks @tulip sleet, I will turn off onewireio on it for now.
tulip sleet
<@!329766224093249548> Yeah you should be able to. I think just removing the pre...
i was just worrying that when you do your periodic update sweeps that it would get put back. But if they are done as PR's maybe that would not happen: the merge would keep the removal
trim elm
huh interesting
tulip sleet
i don't care about the public/private bit, it's more that this library needs not to use pylint or black for the indefinite future
so i can (hand) merge from upstream as necessary
tulip sleet
(upstream here is not even a separate repo)
tulip sleet
so if a sweep breaks again we will just remoe them again
also the .pre-commit-config.yaml ?
tulip sleet
that would be fine, and there aren't any right now 🙂
tulip sleet
👍 thanks
trim elm
Opened the PR, never really done this before, so there's a significant chance I've forgotten something, but we'll see
No problem
@tulip sleet I'm assuming you still want to include licensing info, right? I disabled pylint for the library code but it's currently failing on the spdx stuff
(however if you don't we can always remove it)
tulip sleet
<@!329766224093249548> I'm assuming you still want to include licensing info, ri...
yes, I do want the licensing
tulip sleet
<@!329766224093249548> I'm assuming you still want to include licensing info, ri...
I'll add licensing
idle owl
<@!330227457296957440> <@!362023826105827340> I am working up this library: <h...
Scanning through what Dylan suggested, I'm not certain I agree.
Give me a minute to read.
I feel you should leave the CI files intact.
The library's infrastructure should not be altered.
Hmm.
tulip sleet
I could turn off balck and pylint completely with inline # format off etc. The main thing is preserving the original source so I can do merges in the future
idle owl
So here's an option - add # pylint: skip-file at the beginning of each file, and the # fmt: off at the beginning of each file as well.
I caution against having differing CI setups on any library, regardless of whether they're needed.
tulip sleet
do we have any other such library?
idle owl
Not exactly. datetime is close, but I think we managed to lint it and Black didn't do terrible things to it.
There are no libraries where we are preserving the lib for merging though.
tulip sleet
there is a whole page of lint for this: bad variable names, many other issues
idle owl
I'm all for disabling it. But I don't like the idea of removing parts of the CI.
tulip sleet
I didn't realize the pylint disabling was so easy.
I will revert the PR and add the disabling
idle owl
Disable with a comment explaining why, please.
Alright, thank you.
@tulip sleet I haven't verified that skip-file thing, I only found it. So if it fails, let me know and I'll look into the proper way to do it. Also, if it gets out of hand, there are ways to alter the CI files, but leave them in place that will also cause it to skip things.
Remember to remove the adafruit_asyncio.py file as well 🙂
@slender iron Shouldn't the auto-response regarding CI failure have triggered here? https://github.com/adafruit/Adafruit_CircuitPython_PCF8523/pull/25
idle owl
Hmm ok.
I'm trying to figure out if this is even right. The comment definitely isn't.
Was going to try to fix it myself because I think they submitted via GitHub web UI.
idle owl
@tidal kiln You seem to know hex values really well.... are 0Fh and 0x0F the same thing? I've never encountered the 0Fh version before. I'm assuming they are the same, but google is giving me binary versions instead of equating them for me.
tidal kiln
yep
idle owl
Thanks
tidal kiln
just different ways of writing it
idle owl
Fair enough
stoic rain
So, I just found out that atmel-samd board I added before 7.0.0 (Sparkfun SAMD51 MicroMod) has a design flaw that prevents the hardware SPI tied to the external flash from working.It seems that bitbang, software SPI can be made to work, though. Someone got that working on Arduino.
So, question is: would be be fairly straightforward to use software SPI in this board’s external flash config? And if so, is that a bad idea? 😂
In the meantime, I’ve got a PR that switches to using the internal flash for now.
onyx hinge
So, I just found out that atmel-samd board I added before 7.0.0 (Sparkfun SAMD51...
that is a painful design flaw. As far as I know we have never had to use bitbangio spi for a flash chip before, so it's probably not a simple matter of setting some defines to use it.
Given that the schematic on sparkfun's site is dated a year ago (10/28/2020) it feels unlikely that they'll fix it with a speedy board revision. The closest I've found to a sparkfun statement about the flash chip is from November 2020: https://forum.sparkfun.com/viewtopic.php?f=182&t=54214&p=220455&hilit=samd51+flash#p220447
The flash chip is tied to the bootloader so it's not available for general use unless you remove the bootloader and replace it with your own custom firmware.
https://github.com/sparkfun/MicroMod_Processor_Board-SAMD51/issues/2
Thanks for the issue, a fix is on the way. I'll update in a few weeks when the new version is ready. When that happens feel free to reach out to SparkFun's customer service to get a replacement.
though this from September 15 does point at them issuing a revision...
manic glacierBOT
onyx hinge
I don't understand the comment on the sparkfun forums about the flash chip being for use of the bootloader only, do you?
stoic rain
Yeah, I was thinking the same thing is likely @onyx hinge.
Nah, that response made no sense, which is why I kept looking at it. My guess is that something got messed up with they switched from QSPI to SPI in the board design.
stoic rain
I did find that the HOLD and WP pins also need to be pulled high. I got an equivalent external flash chip working on an Arduino MKR Zero that I “haxpressed”. But if MOSI/MISO don’t work, it’s kinda for naught.
idle owl
@gilded cradle Is it a safe assumption that https://www.adafruit.com/product/5297 will work on Raspberry Pi with Blinka? Or is that not a safe assumption to make...?
gilded cradle
<@!478616919721771019> Is it a safe assumption that https://www.adafruit.com/pro...
Probably not a safe assumption. I think lesamouraipourpre was adding SH1107 support to Blinka Displayio, but I don't think a PR was actually made.
idle owl
Probably not a safe assumption. I think lesamouraipourpre was adding SH1107 supp...
OK, I don't have one to test yet, it's not in the shop yet. I'm doing the guide and was wondering whether to include Pi. I will not for now. Thanks!
gilded cradle
OK, I don't have one to test yet, it's not in the shop yet. I'm doing the guide ...
Sounds good.
proven garnet
Not sure who to bring it up to, but I noticed a repo with a different Github Action workflow than the rest, so I submitted an issue: https://github.com/adafruit/Adafruit_CircuitPython_PCF8563/issues/2 Could be others, but I at least didn't want to go messing around with the workflows here without some assistance.
idle owl
Not sure who to bring it up to, but I noticed a repo with a different Github Act...
That might explain the issue earlier!
It means it got missed in the update.
ember iris
Not sure who to bring it up to, but I noticed a repo with a different Github Act...
Nice catch!
proven garnet
Oh hey, I guess I'm late to the party haha
idle owl
Not sure who to bring it up to, but I noticed a repo with a different Github Act...
If you're up for it, you can find a library that has pylint-2.11.1 in the .pre-commit-config.yaml file, and update this one to match. I bet you'll have to lint some things - I did a PR earlier to this one and didn't notice the issue. Good catch!
@proven garnet Here is a library that is passing, and has been updated. https://github.com/adafruit/Adafruit_CircuitPython_LIS3DH
proven garnet
If you're up for it, you can find a library that has pylint-2.11.1 in the `.pre-...
Thanks! Glad I wasn't too late to this! Sure I'll pull a .pre-commit-config.yaml from a there and yoink it on over there!
idle owl
Oh hey, I guess I'm late to the party haha
Nah, you're not late. There was an issue earlier with the PR I fixed up, and it is probably related.
idle owl
Thanks! Glad I wasn't too late to this! Sure I'll pull a `.pre-commit-config.y...
There were other changes made. Hold on.
This commit shows what the patch changed.
But if you look at the next few commits as well, you'll see there are a few more changes added in: https://github.com/adafruit/Adafruit_CircuitPython_LIS3DH/commits/main
@proven garnet Dylan is supposed to be working on the last libraries that got missed by the patch (the patch skipped a number of libraries, as always). But she apparently hasn't gotten to this one yet.
proven garnet
<@!135415127879974912> Dylan is supposed to be working on the last libraries tha...
So pull the .yaml file, and make the missing changes like the Patch, linting, etc., right?
manic glacierBOT
idle owl
So pull the .yaml file, and make the missing changes like the Patch, linting, et...
Yep!
That patch turned out to be kind of meandering, so there's a lot to it.
idle owl
Thanks so much!
slender iron
here is the testflight link for glider: https://testflight.apple.com/join/DNccPz6G
solar whale
Glider connected right up to my glasses driver board...so far just using the explorer.
Very nice!
proven garnet
Thanks so much!
Hokay, submitted it. I was hoping I could cherry pick commits but I ended up doing it manually.
idle owl
Hokay, submitted it. I was hoping I could cherry pick commits but I ended up do...
Link please!
idle owl
https://github.com/adafruit/Adafruit_CircuitPython_PCF8563/pull/3
Ooh I think that's good to go! Asking Dylan for the final review.
@proven garnet Thanks so much! For the record, that does explain the anomaly earlier with the PR submitted by another community member. Good catch! I completely missed it.
Actually, I'm wrong, it's a slightly different repo altogether 😄
oops.
But still!
proven garnet
My pleasure! Glad I was able to help! 
idle owl
So there's no explanation for the anomaly earlier after all. Oh well! 🤷🏻♀️
proven garnet
I'll stay on the lookout for others
idle owl
@proven garnet Are you interested in helping out with this? I can get you a short list if you'd like.
Basically it'd be pretty much what you just did, although some of the changes might already be present.
proven garnet
<@!135415127879974912> Are you interested in helping out with this? I can get yo...
Yup! Happy to help out with that!
idle owl
Yup! Happy to help out with that!
Alright, great! Give me a moment.
idle owl
Yup! Happy to help out with that!
Dylan needs to pull together a list. It'll be about 20 minutes. I'll get back to you then!
proven garnet
Sounds good! I'll work off of that when it's ready! 🙂
idle owl
Thanks so much! Really appreciate it. The cleanup after running a patch like that can be tedious, and Dylan's the only one working on it. So this is immensely helpful.
trim elm
@idle owl @proven garnet Ok, here's a gist of the files. https://gist.github.com/dherrada/9d8b7abd23c3c7a512615508b6258c8b
file_has_none are (mostly) repositories that aren't fully set up and some repositories that aren't circuitpython libraries, although I'll have to look into gps and st7565
file_has_text_1 are libraries that didn't have the initial patch applied to them
file_has_text_2 are libraries that did have the initial patch applied to them
idle owl
OK, asyncio is still WIP. So skip that one.
trim elm
Let me know if y'all have any questions or anything is confusing
proven garnet
Will do!
idle owl
Let me know if y'all have any questions or anything is confusing
It's a little unclear to me, but it's the file_has_text_1 (43) "pylint-2.7.1" that need to be updated?
trim elm
Yes
idle owl
OK.
trim elm
text 2 are the ones that don't need to be
and none are the ones that don't have either and need to be looked at
idle owl
@proven garnet Start at the bottom of the file_has_text_1 (43) "pylint-2.7.1" part of the list, please. Then Dylan can start at the top of the list.
trim elm
Works for me!
proven garnet
Sounds good!
trim elm
Thanks @proven garnet
idle owl
@proven garnet But as I said, skip asyncio.
proven garnet
@idle owl got it! And no problem @trim elm !
idle owl
@lone axle Do I remember we started suggesting using bitmap_label over label? Or what am I remembering w/r/t that?
proven garnet
How should I handle this error with the pylint_tests hook? Other libraries didn't have it, I think, but this repo does have a tests folder: https://github.com/tekktrik/Adafruit_CircuitPython_SimpleMath/runs/4171367246?check_suite_focus=true
idle owl
How should I handle this error with the pylint_tests hook? Other libraries didn...
Hmm. That's a new one.
onyx hinge
Looks like the old 'local' step of pylint didn't get removed from the yml file
that's a guess bu tI notice there's (test code) and (tests code)
lone axle
<@!382939733107408897> Do I remember we started suggesting using `bitmap_label` ...
Yep, that is correct. I do generally recommend folks use bitmap_label instead of label It tends to use a bit less RAM in the cases that I have tested, and it allows for some additional tricks with colors that are not possible with label The API for both should be identical so swapping from one to the other generally should just be a matter of changing the import and possibly initializer call.
onyx hinge
in the list of steps
idle owl
Oh.... ok
proven garnet
that's a guess bu tI notice there's (test code) and (tests code)
So I should remove that hook?
idle owl
Yep, that is correct. I do generally recommend folks use `bitmap_label` instead ...
Keen, thanks. I'm updating an example for a new board, and figured I wanted to do it right.
idle owl
So I should remove that hook?
Yes, Jeff is right, remove - repo: local hooks: - id: pylint_tests name: pylint (tests code) description: Run pylint rules on "tests/*.py" files entry: /usr/bin/env bash -c args: ['([[ ! -d "tests" ]] || for test in $(find . -path "./tests/*.py"); do pylint --disable=missing-docstring,consider-using-f-string $test; done)'] language: system
That might fix it.
proven garnet
Youuuuu got it!
I think at least the ones I've seen so far have needed this fix, I can keep applying it if that's the fix
idle owl
I think at least the ones I've seen so far have needed this fix, I can keep appl...
Sounds good.
proven garnet
I'm also noticing at least on the same repo it's running a workflow called Run Tests that I haven't seen before, is that outdated? I can remove that workflow if its a relic, here and on any others I encounter
PR for reference: https://github.com/adafruit/Adafruit_CircuitPython_SimpleMath/pull/5
idle owl
I'm not sure.
Everything passed though, I see.
@proven garnet Can you title the PRs "Pylint and RTD update patch, and other fixes" or something a little more descriptive? I'm going to tag the review team on it and, while Dylan, you and I all know what you're going for here, the review team may not. Thanks!
This one is fine, but moving forward.
onyx hinge
Yes, I think so.
trim elm
Oh hah yeah I should probably title them that too
idle owl
I'm also noticing at least on the same repo it's running a workflow called Run T...
I think this is outdated, yes. Give me a moment to check with Dan.
@tulip sleet Did we remove this from the libs? Or why is this only in this library? https://github.com/adafruit/Adafruit_CircuitPython_SimpleMath/blob/main/.github/workflows/test.yml
trim elm
@idle owl @proven garnet I'm pretty sure that is for libraries that have a tests/ directory. I don't think we removed it, but I can't say for sure
idle owl
<@!330227457296957440> <@!135415127879974912> I'm pretty sure that is for librar...
datetime doesn't have it.
I thought maybe that was the case too.
trim elm
Hmm
idle owl
But checked.
@proven garnet Let's leave the tests CI alone for now. Your PR is fine as-is. We'll revisit if necessary at a later time.
proven garnet
<@!135415127879974912> Can you title the PRs "Pylint and RTD update patch, and o...
Sounds like a plan! I'll title them as such.
idle owl
Sounds like a plan! I'll title them as such.
Thank you!
manic glacierBOT
CircuitPython version
latest
Code/REPL
bitmaptools.alphablend(dest, src1, src1, colorspace, factor1=0)
Behavior
Result is incorrectly mixed
Description
No response
Additional information
CircuitPython version
latest
Code/REPL
bitmaptools.dither(dest, src, colorspace, bitmaptools.DitherAlgorithm.FloydStenberg)
Behavior
Some stray dots are seen
Description
No response
Additional information
vocal kindle
I'm poking at Blinka/PlatformDetect on MicroPython, and hitting some ignorance on using the functions, or perhaps limitations on the available functions. I want to identify my specific board, which I've built a MicroPython firmware for that identifies it on bootup.
MicroPython v1.17-123-g7f1434442 on 2021-11-10; M5Stack STAMP-PICO with ESP32-PICO-D4
But when I use the functions from learn.ada, it only lists the generic platform...
print(detector.chip.id)
ESP32
since the integrated function only responds with ESP32, not the string from the firmware as listed above
>>>import sys
>>>print(sys.platform)
esp32
Is there a better function I could/should use to get it to return either of the strings below:
M5Stack STAMP-PICO with ESP32-PICO-D4
oh...this. Now how to make use of that in PlatformDetect
>>> print(os.uname().machine)
M5Stack STAMP-PICO with ESP32-PICO-D4
Had to modify my google query a few more times to find that answer
proven garnet
@idle owl we should only disable duplicate code as necessary right? Wait for the CI action to fail? I've been turning them on in my forks to check ahead of time, so I can keep the PRs cleaner if this is the case.
idle owl
<@!330227457296957440> we should only disable duplicate code as necessary right?...
Yes please. Only disable if it fails on duplicate code check. That way we have granular control over disabling it, and a very easy way to check which libraries required the disable, and figure out which ones should be refactored to avoid it at a later time. That is the plan for some - to refactor things into a base class and subclass it etc. But that's a task for a later date.
proven garnet
@idle owl sounds good!
slender iron
SD card CSD
bf168000 a96db7ff 325b5a83 00002f00
lexar 16G
800a4000 00774d7f 325b5900 00400e00
sandisk 8G
800a4040 003b377f 325b5900 00400e00
samsung 64G sdxc
800a4040 01dcff7f 325b5900 00400e00
bit 1 is always 1
which is bit 1?
onyx hinge
@slender iron https://github.com/adafruit/Adafruit_CircuitPython_SD/blob/main/adafruit_sdcard.py#L143
I think this code is right, at least for cards under 32GB capacity. I think there's a bug that says it's wrong for 32GB and up cards
slender iron
I don't think I've shifted in bit 1
the numbers would be right if all shifted one more byte I think
vocal kindle
under PlatformDetect, chip.py, in this function, is it cluttering the code too much to distinguish the variants of the ESP32 (i.e: ESP32-S2, ESP32-C3, ESP32-PICO-D4 in my usage), versus simply keeping the variants identified simply as "ESP32"
https://github.com/adafruit/Adafruit_Python_PlatformDetect/blob/main/adafruit_platformdetect/chip.py#L153:L170
obviously that doesn't currently exist, but as I'm adding the ability to do so, eg:
if platform == "esp32":
self._chip_id = chips.ESP32
return self._chip_id
I could use the output of the below function to distinguish a particular binary, or output ESP32 if it's not distinguished
>>>print(os.uname().machine)
M5Stack STAMP-PICO with ESP32-PICO-D4
I didn't add the test.yml; lesamouaripourpre added it because they wrote some tests (which are pretty rare in libraries; maybe this is the only one with tests)
idle owl
I didn't add the test.yml; lesamouaripourpre added it because they wrote some te...
This isn't the only one with /tests/ but it is the only one with that workflow.
I'm leaving it for now, but I have no idea why it got added.
proven garnet
For anyone who knows the macropad library, should we be using the context manager as suggested by pylint? https://github.com/tekktrik/Adafruit_CircuitPython_MacroPad/runs/4173391976?check_suite_focus=true
onyx hinge
@proven garnet in this case, yes, it would be a safe change to do so, because the use of the mp3 / wave file is done by the time the function returns.
896 with audiopwmio.PWMAudioOut(
897 board.SPEAKER
898 ) as audio: # pylint: disable=not-callable
899 mp3file = audiomp3.MP3Decoder(open(file_name, "rb"))
900 audio.play(mp3file)
901 while audio.playing:
902 pass
``` this is line 899 in my copyyeah there are two similar instances, one for wave files and one for mp3 files
I think you have to change it around a bit so only change it if you're going to test it. (because the result of 'open' is passed to a function, not put in a named variable right now)
ignoring it is also fine with me, using a pylint:disable
proven garnet
with audiopwmio.PWMAudioOut(
board.SPEAKER
) as audio: # pylint: disable=not-callable
with open(file_name, "rb") as audio_file:
wavefile = audiocore.WaveFile(audio_file)
audio.play(wavefile)
while audio.playing:
pass
That should be sufficient, right?
onyx hinge
yeah that looks right
proven garnet
Yup yup! Appreciate the late response!
bitter bronze
Is this the right place to ask about import problems using Mu and Visual Studio Code when trying to import adafruit_pyportal?
Also get import error with import board
My end goal is getting WipperSnapper running on PyPortal..
analog bridge
Will allocating an object using m_new_ll_obj keep it alive across multiple VM runs?
stuck elbow
no, it will just allocate it at the end of the free memory, where all the objects that are not expected to be released during the lifetime of the program go
the long-live objects is just an optimization to prevent memory fragmentation
analog bridge
so the only option is static allocation?
stuck elbow
only option for what?
analog bridge
for keeping the object alive across multiple VM run (like the board.BUS) singletons
stuck elbow
you probably want something like this: https://github.com/adafruit/circuitpython/pull/3695
tulip sleet
for keeping the object alive across multiple VM run (like the `board.BUS`) singl...
if size is known, then static is fine. the root pointer for the object should be in MP_STATE_VM, in the root pointers list, or should be marked during gc by a module-specific gc routine, e.g. shared-module/usb_hid/__init__.c:313:void usb_hid_gc_collect(void)
ornate breach
Hey Dan, I’m hoping to finish release notes for beta today. I haven’t had much time the last two days to work on it.
I underestimated my day job work load lol..
tulip sleet
for keeping the object alive across multiple VM run (like the `board.BUS`) singl...
I use the "supervisor allocations" to save some data about HID devices between boot.py and the second VM instantiation (for code.py). The supervisor allocations can only be done while the VM is not running, which is a bit confusing. The last supervisor allocation allocates the vm heap with the remaining storage, so there is no more available after the vm starts.
The usb_hid that needs to remember some variable-sized data between VM's. I create a table in the boot.py VM, and pass a pointer to it the calling VM. That routine copies the data onto the stack temporarily. When the boot.py VM finishes, it uses supervisor allocation to allocate space for that and copies it from the stack into the new supervisor allocation block, and then copies from the stack into there. This "data hoisting" up and down was the easiest way I could see to do this for a variable sized block (a few hundred bytes).
tulip sleet
Hey Dan, I’m hoping to finish release notes for beta today. I haven’t had much t...
Thanks. But don't push yourself on this; it sounds like you could use some rest. I can take over if you have something to pass to me.
ornate breach
Thanks. But don't push yourself on this; it sounds like you could use some rest....
I was having a hard time deciphering between what was new, was a fix, etc..
but i think i captured all the new boards
and at least the asyncio initial support
crimson ferry
what's the difference between bytes and ReadableBuffer in readthedocs / shared-bindings?
Oh, bytes is an example. Is it preferred in parameter docs to call it out as readableBuffer? Or is it possible some forms of ReadableBuffer wouldn't work in a given situation? Should existing bytes references be replaced with ReadableBuffer?
e.g., in Espressif wifi.radio, SSID & password are ReadableBuffer but it would be strange to feed in rgbmatrix.RGBMatrix, which is also an example of ReadableBuffer
idle owl
Yup yup! Appreciate the late response!
Hello! There's an even better way to do nested with. Would you like to see?
idle owl
Is this the right place to ask about import problems using Mu and Visual Studio...
Uncertain whether you got a response. #help-with-circuitpython is probably a better place for that sort of thing. This channel is more for development discussion.
onyx hinge
@crimson ferry bytes is a specific type and ReadableBuffer is shorthand for all types that can be treated in a certain way by the core C code, including bytes. Not all ReadableBuffers will make sense in all contexts, but I tend think about type hints as saying what is possible, not what is sensible. This may be inflexible thinking and I'm open to other viewpoints.
crimson ferry
in theory, I could set an SSID with a rgbmatrix.RGBMatrix? (technically, SSID can be any bytes, there are no restrictions in the spec)
that seems to be the decider for me... if I do it, will it break?
if so, maybe I should call it out as what will not break
onyx hinge
I'm not sure. An RGBMatrix, when treated as a ReadableBuffer, will probably be more bytes than an ssid should have. (but the type system can't say "it should be a ReadableBuffer and no more than 16 bytes", or whatever the requirement is)
crimson ferry
OK, I'll live with the ambiguity 😉
onyx hinge
additionally it looks like the length isn't verified to be valid, when it comes specifically to the implementation for espressif MCUs, so you will be able to make a HardFault if you try it
crimson ferry
oh, interesting
onyx hinge
memcpy(&config->ap.ssid, ssid, ssid_len);
config->ap.ssid[ssid_len] = 0;
``` I don't see any validation of ssid_len, which in C is the length of the ReadableBuffer. Hence I speculate that you might be able to get a HardFault or other ill behavior if you (say) try to set `ssid=bytes([0] * 4096)`lol oops, we should probably fix that... I'm in there now, I'll see if I can
onyx hinge
thanks!
similarly for all the other parameters passed in as a pointer + length, there's also password & bssid
crimson ferry
I'll research limits on password, bssid is fixed length (6 Bytes)
crimson ferry
64, boom, thx
onyx hinge
So, untested, if you write ssid_len = MIN(ssid_len, sizeof(config->ap.ssid)) you would take at most 32 bytes of ssid data, though then the assignment of config->ap.ssid[ssid_len] = 0 because it would potentially assign ap.ssid[32] which is out of bounds. memset(config->ap.ssid, 0, sizeof(config->ap.ssid)); memcpy(...) would ensure that all non-used bytes are zeroed out. [sorry if I'm lecturing you about C you already know, for me this is stuff that I need to work through every time to get right so I'm just thinking out loud and trying to help. let me know if it is not helpful to you.]
crimson ferry
"C you already know" lol, not much, so thanks
idle owl
I'm using a PDM microphone. I'm getting values back. What do these values mean? What can I expect to be a max value? Datasheet is not helping me here. It's the one on the Circuit Playground Bluefruit (also CPX), and we have it on breakouts.
Is it 0-65535 or whatever that common number is?
I am trying to make a NeoPixel sound meter, but I want to make sure the max value I use is reasonable. It seems to be currently based on testing, but if I understood what the values meant, I could make a more informed choice.
onyx hinge
@idle owl it looks like it's a theoretical 0-65535 range when you specify bit_depth=16 and a theoretical 0-255 range when you specify bit_depth=8 (the default)
idle owl
<@!330227457296957440> it looks like it's a theoretical 0-65535 range when you s...
Oh hmm ok.
onyx hinge
but whether the maximum you can shout will really go to 65535 or 255 who can say. It helps that you have loud loud construction going on I suppose
idle owl
I guess I could have looked at the code.
Yeah, it was bad enough to register.
But you have to be close to it to make it work.
Tapping the mic with my finger I got it to 4k or something.
I don't want folks to have to shout. And we're planning on making it so someone could change something in the example, this is a good thing to be able to change - the threshold.
But I need a solid default.
It's 2k right now.
onyx hinge
It seems a bit like asking "which number should I set my mic's volume to, so that it's never too quiet & never clips", there simply is no iron clad answer.
idle owl
Also valid. I'm making notes in the example about what the valid options are, but to choose based on testing I guess.
tulip sleet
Here you go: https://gist.github.com/skerr92/9f0547647ebcbe03ccefe37fd2a83595
Thanks for starting on this, it's quite helpful.
analog bridge
Thanks @stuck elbow and @tulip sleet. Looks like I will have to use static allocation as supervisor allocations can only be done while the VM is not running.
idle owl
@proven garnet I missed something you were doing in your PRs. You're going to have to go back and fix it in the PRs I merged because I didn't notice until now. I explain in the FunHouse PR - please review that comment and apply it to the rest of the repos you've updated. Sorry about that!
stuck elbow
Thanks <@!351136009922019331> and <@!329766224093249548>. Looks like I will have...
But that pull request lets you do a normal allocation, and have it moved to static when the VM finishes?
Or did I misunderstand it?
tulip sleet
But that pull request lets you do a normal allocation, and have it moved to stat...
hmm, I think you are right, though I thought I tried this and could not use it for some reason:
// ... If movable is true, memory will be
// taken from either outside or inside the GC heap, and when the VM exits, will be moved outside.
// The ptr of the returned supervisor_allocation will change at that point. If you need to be
// notified of that, add your own callback function at the designated place near the end of
// supervisor_move_memory().
the semantics of movable being true or false are quite different
@analog bridge ^^
proven garnet
<@!135415127879974912> I missed something you were doing in your PRs. You're goi...
No worries, thanks for catching it! Submitting PRs for those that were already approved now so I don't forget later, and fixing any that weren't approved yet.
idle owl
No worries, thanks for catching it! Submitting PRs for those that were already ...
Thanks so much! I really appreciate everything you're doing. Again, apologies for not being clearer and missing it in the first place.
crimson ferry
```c
memcpy(&config->ap.ssid, ssid, ssid_len);
config->ap.ssid[ssid_len]...
indeed, a Very long SSID triggered a reset. oddly with microcontroller.ResetReason.WATCHDOG (no safe mode) ...but ill behavior to be sure, I'll add a check
proven garnet
Thanks so much! I really appreciate everything you're doing. Again, apologies fo...
No worries, honestly just happy to help 
idle owl
No worries, honestly just happy to help <a:parrot:413347389055827989>
Oh! Did you want to see the better way to handle nested with statements?
idle owl
Black makes this look a little wonky, but it's with self._audio_out(board.SPEAKER) as audio, audiomp3.MP3Decoder(open(file_name, "rb")) as mp3file: https://github.com/adafruit/Adafruit_CircuitPython_CircuitPlayground/blob/e0a76a22ad49d1ce7cf33acf2ab24be05a6a1442/adafruit_circuitplayground/bluefruit.py#L161
I only learned this last week. 🙂
It's basically with first_item as first_variable, second_item as second_variable:
proven garnet
Oh! Did you want to see the better way to handle nested `with` statements?
Ooo yes please
idle owl
Black makes this look a little wonky, but it's ```with self._audio_out(board.SPE...
@proven garnet Check out this message and the ones after it!
tulip sleet
answered in #help-with-circuitpython . Please don't crosspost in multiple channels (or delete the original post if you move it).
proven garnet
<@!135415127879974912> Check out this message and the ones after it!
Nice! Never knew you could chain, thanks!
idle owl
Nice! Never knew you could chain, thanks!
I didn't either until last week 🙂 Happy to share new knowledge, helps me remember it! You're welcome.
manic glacierBOT
idle owl
@onyx hinge https://github.com/adafruit/Adafruit_CircuitPython_CircuitPlayground/pull/112 If you don't mind...
