Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1k0jC2OVIoslSAK_jhq6ioTEbY5Ep1MChD5UIOSRIHgA/edit?usp=sharing
#circuitpython-dev
random junco
onyx hinge
The bundle zips are created on a Linux VM so I don't know why they would have Mac specific metadata attached.
minor plume
The bundle zips are created on a Linux VM so I don't know why they would have Ma...
I think macOS is adding the extended attribute when Python downloads them.
onyx hinge
I think macOS is adding the extended attribute when Python downloads them.
I won't be fit to review or test it but if you think of tagging me in the fix PR I'm curious to find out the answer
minor plume
I won't be fit to review or test it but if you think of tagging me in the fix PR...
Will do! I’m going to take a stab at it this afternoon after work. I think writing a unit test is actually going to be harder than writing the patch in this case. LOL (The circup unit tests are ripe for someone to refactor all of those MagicMocks into some test fixtures, but that’s a bigger project than I am prepared for this week.)
onyx hinge
I do know enough to help you make the ci run on a Mac VM if that comes in handy
tame creek
downloading using curl on 10.15 doesn't seem to set the quarantine attribute. it seems weird but maybe circup is doing something explicit to set the attribute? or using a library that does?
jaunty juniper
it just writes into a new file
idle owl
@jaunty juniper and @tame creek since you weren't in our Weekly meeting today, I wanted to let you know you got a massive hug report from me for helping me out this weekend. I was trying to finish up something that's already way past its deadline, and kept running into things I didn't know how to fix. Your help was imperative. Thank you so much!
jaunty juniper
in order to have a progress bar during download it uses click.progressbar and writes into a file from a requests.get stream
jaunty juniper
<@199811382793863168> and <@824013908490059826> since you weren't in our Weekly ...
I was listening 😉
idle owl
Oh! I was busy trying to keep track of if Paul needed any help 😄
I missed you being there.
jaunty juniper
I try to follow somewhat when i can and I look at the notes 🙂
idle owl
Fair enough! I basically knew that, but like I said, was entirely focused elsewhere today.
tame creek
i don't see anywhere obvious in circup that's explicitly setting the quarantine attribute, or running anything to do the download that might set it
jaunty juniper
well no of course we don't set it, what surprises me is that it's not even downloading directly, but writing an arbitrary file, does that mean that they put that attribute to every file ???
maybe it's a beta thing
midnight ember
The reason I chose the RFM95 over the RFM69 is because the packet length is longer. Thought I read somewhere the RFM69 only does 8 characters and the RFM95 does 32. Didn't even think about it until just now that I'm working with 252 characters. I think packet length is worth noting in the product guide. I got far more characters available to me than I purchased it for, allowing something like a messenger to become a viable project.
idle owl
The reason I chose the RFM95 over the RFM69 is because the packet length is long...
Please use the "Feedback? Corrections?" link in the guide to provide your suggestion. That attaches it directly to the guide and doesn't leave it simply on Discord.
idle owl
I feel you there 😄
midnight ember
Wow so the issue that @errant grail discovered actually affected about 30 boards. Looks like lingering code from trying to remove the auto brightness stuff.
midnight ember
That little bug started out as maybe a small little thing then grew up to be a 6 headed hydra.
It's neat to see how this one finally gets resolved after many people were looking into it during help desk session. It was really cool to see others breaking out their displays and running code to help narrow it down within an hour but we obviously missed a ton of other boards it was affecting.
blissful pollen
The time slowdown was pretty small at 50KHz on other boards. A look at the code confirms (at least for ATMEL) setting the PWM duty cycle only happens at the end of a cycle. So at 10Hz that is not very often.
midnight ember
Looks like boards only with a built-in display so I would have never detected it on my featherwing tft?
blissful pollen
I am fixing this
I was going to check if you were getting progress on that. If you need a test later let me know
blissful pollen
Looks like boards only with a built-in display so I would have never detected it...
It affected any setup where you used the backlight_pin/pwm settings when creating the display. So it could affect non-built-in displays too (it did for me)
tulip sleet
the logic for pwmout_reset was wrong; it was resetting only the never-rest PWM's 🙂 . It works on all boards now except Feather ESP32-S2 TFT. I reordered display init a bit in main.c. Strangely ESP32-S3 TFT is OK. So there
midnight ember
Hmm I have a modified featherwing tft with backlight pwm and it didn't affect me.
tulip sleet
is still something wrong somewhere, but it's a lot better.
We have had weirdness on some samples of these Feather TFT's. It might be a reset timing problem.
blissful pollen
Hmm I have a modified featherwing tft with backlight pwm and it didn't affect me...
The boards where the PWM was set at 50Khz was barely noticeable too. If you set it up at 500Hz you would see it. Don't try 10Hz 🙂
jaunty juniper
the pwmout_reset bug was general on all ports ?
jaunty juniper
well maybe this ? https://github.com/adafruit/circuitpython/issues/6223
midnight ember
oof since april, that'll be super cool if it fixes that one too.
tulip sleet
could be -- first pass is here https://github.com/adafruit/circuitpython/compare/main...dhalbert:circuitpython:espressif-pwmout-reset but I am making some more cleanup changes before submitting the PR
midnight ember
if i remember correctly cgrover said he's been dealing with it since may? so april/may sounds like prime real estate for it to start rearing its head.
tulip sleet
it was stopping the PWMs no matter what
jaunty juniper
it might also be unrelated, I'll test when it's in
actually I haven't tested since, who knows if it didn't fix itself in the mean time 😛
errant grail
if i remember correctly cgrover said he's been dealing with it since may? so apr...
I’ve been working on various display and palette brightness projects since about April, but didn’t see the performance issue until August.
midnight ember
Duh that’s right and how Foamyguy tracked it to the aug 10th commit.
errant grail
Another example of how well the community works together and "shows the work." I soak in as much experience and techniques as possible. I hope to grow up to be on-par some day.
blissful pollen
I think the "best" part of it too was it was literally one line. For the slowdown at least, the brightness issue is another thing.
tulip sleet
I just found out why ESP32-S2 TFT was changing to a dark display, but ESP32-S3 TFT was not. The issue was in the .... ESP32 camera code (!), which was resetting PWM channel 0 willy-nilly without checking it really had to.
this is a second bug on top of the other one i found
blissful pollen
oh wow
tame creek
The time slowdown was pretty small at 50KHz on other boards. A look at the code ...
huh, the PWM peripherals aren't double-buffered on those devices? (or CircuitPython isn't using such when available?)
tulip sleet
i dunno, it really shouldn't have been resetting the PWM so much anyway
it is the LEDC peripheral
tame creek
i dunno, it really shouldn't have been resetting the PWM so much anyway
yeah, it should probably at least have waited for an actual display refresh period
blissful pollen
I'm not really sure how the PWM worked, just noticed it was being called every background refresh. It really only needs to be set if it changes
midnight ember
Do some or all bug fixes get back ported to 7.3.3? How would I know if a bug fix that might fix my board was added to 7.3.3, like do I need to track commits or do you wait for a 7.3.4?
manic glacierBOT
Adds support for the BananaPi BPI-Bit-S2 Boards.
Based on esp32s2 chip.
with 25 ws2812 LEDs, 1 buzzer, 2 photosensitive sensors, 1 thermosensitive sensor.
2 buttons on the front, 1 reset button and 1 boot button on the back.
thorny jay
Hi @lone axle , at one point in a stream you were checking the speed of triangle(?) of some library on the Titano(?). It could be related to the way lines are drawn. https://github.com/adafruit/Adafruit_CircuitPython_Display_Shapes/blob/main/adafruit_display_shapes/triangle.py 3 out of the 4 call to _line are for lines where the line has the same "y" and are made for filling the triangle. I did not find the implementation of _line but my assumption is that if those are continuous pixel in a memory location, there is a fast way to "memcopy" that or even with Python array slice trick. Either _line check that condition, or you may want a _horizontal_line function? I hope it can be usefull.
onyx hinge
Do some or all bug fixes get back ported to 7.3.3? How would I know if a bug fix...
We selectively backport some fixes to the stable release. You can see whether a related issue is assigned to the 7.3.x or 7.x.x milestone, or if a PR targets the 7.3.x branch. If you need this fix for your product then your best bet is to open a PR targeting the 7.3.x branch to open a concrete discussion
manic glacierBOT
idle owl
UGH. @tulip sleet Keypad isn't available for the NeoKey Trinkey?
I guess I'll sort out doing it manually. All I want is a spacebar press. And it's becoming a lot of code. 😄
tulip sleet
I guess I'll sort out doing it manually. All I want is a spacebar press. And it'...
no room on non-Express SAMD21
debouncer?
tulip sleet
or just do a little delay after detecting the keydown
idle owl
bigger delay is probably also ok
tulip sleet
that is the least code
idle owl
This is for a remote trigger for my phone camera, because I need two hands in a lot of these guide photos.
tulip sleet
50-100ms min is probably fine
idle owl
So I was hoping to bang this out quick. Which is not happening, heh.
tulip sleet
self-timer on camera phone?
jaunty juniper
just copy paste https://learn.adafruit.com/USB-foot-switch-circuit-python/software ?
idle owl
self-timer on camera phone?
There are Reasons this might not work.
tulip sleet
idle owl
And yes, but absolutely not, I am never setting up my phone to be listening to me for any reason.
idle owl
There's a toggle on my phone.
stuck elbow
mechanical?
idle owl
Oh you mean overall.
I get that it's listening. But I refuse to actually set up the software that ensures it will respond.
jaunty juniper
let's not give in to the paranoia
stuck elbow
it's not paranoia when they really are out to get you
idle owl
It's not paranoia for me, I simply hate having that feature on. Responds to anyone talking to the same software, etc.
jaunty juniper
yeah I disable that too, my point is that it's not always on, but you can enable it temporarily
stuck elbow
so it stays fresh longer
jaunty juniper
now, a Clue that listens to you and presses space via bluetooth when you say something on the other hand
idle owl
All of these examples involve more than one button/switch
I cannot figure out the flow control for one.
Either it's sending spacebar all the time
or not at all
OK, I apparently didn't even set up the switch right in the first place.
It reports True all the time, pressed or not.
How did I mess that up?
import time
import board
import digitalio
import neopixel
import usb_hid
from adafruit_hid.keyboard import Keyboard
from adafruit_hid.keyboard_layout_us import KeyboardLayoutUS
from adafruit_hid.keycode import Keycode
keyboard = Keyboard(usb_hid.devices)
keyboard_layout = KeyboardLayoutUS(keyboard)
key_switch = digitalio.DigitalInOut(board.SWITCH)
key_switch.switch_to_input(pull=digitalio.Pull.UP)
pixel = neopixel.NeoPixel(board.NEOPIXEL, 12, brightness=0.2)
while True:
print(key_switch.value)
time.sleep(0.1)```That's what I have right now.
jaunty juniper
there's only one button in the USB Foot Switch guide though
idle owl
then why does it do list or whatever to set it up like there's more
That's what I don't get, I don't want to deal with that.
jaunty juniper
right, future proofing
idle owl
Sigh. I'll copy it.
jaunty juniper
maybe we need a simpler sample code
idle owl
We do. But I guess I'll deal with this for now since I am pressed for time.
Nope.
It doesn't work.
jaunty juniper
is the button between the pin and gnd ?
idle owl
It's not sending spacebar all the time this time, but it's still not sending it on press either.
It's a NeoKey Trinkey
The keyswitch has a pin.
That's it.
Built in.
jaunty juniper
ah the trinkey
idle owl
Is that a pulldown or something?
proven garnet
This is a documentation only PR!
idle owl
Hah!
Pull down at least makes it change state.
I'm rolling with this, even if it's wrong.
jaunty juniper
yep the trinkey needs a pulldown
idle owl
OMG it works.
(╯°□°）╯︵ ┻━┻
LED doesn't turn off when you stop pressing the key, but whatever.
Ok now it does.
So the delay, I set it to 0.5, is that too high?
I really don't expect to be spamming the phone to take photos.
And would rather not do exactly that.
Simply means it takes the LED that long to turn off as well.
ornate breach
(╯°□°）╯︵ ┻━┻
Do you have this saved somewhere to copy paste? Lol
idle owl
Oh! .. No you do /tableflip in the input box for Discord, and it resolves to that emoji.
idle owl
No!
ornate breach
Lol
manic glacierBOT
in the circuitpython repo /ports/expressif/boards there is a board definition for the ESP32-S3-BOX. See the (edited) screenshot combo below. I guess that a board definition could be added to the tinyuf2 repo, isn't it ?
idle owl
@tulip sleet Are you still here? It's totally ok if not.
Trying to figure out how to word an explanation of why the "deep sleep" numbers on the Feather V2 are more "light sleep" numbers. e.g. How much detail to get into or how to say it.
blissful pollen
If you want to run wording past someone I am around, though only have a passing knowledge of why that was the case
idle owl
It's a right-now problem, not a Feather problem. Dan's working on a fix in CircuitPython that will get it down to legit deep sleep numbers. But the fix isn't available yet.
So... I have no idea how much of that to explain.
Or maybe what I said works.
blissful pollen
I'm trying to recall was it because a pin was left on in deep sleep mode (if I recall?)
idle owl
Yeah all pins are pulled up when deep sleep starts.
Which means power pins are also pulled up and turned on.
So turning them off in software does nothing.
blissful pollen
You could also just say that "due to a known issue deep sleep current is higher on X boards. We hope to resolve this in the near future" or similar to that?
blissful pollen
No problem
idle owl
I was seriously overthinking it. My brain is kind of mush at this point. Been doing photos all day for this guide, and spent the last hour and a half putting them into the guide and rewording everything to make sense. It should have been dinner time two hours ago, heh.
blissful pollen
I completely get that feeling. At a certain point your brain cannot read it/think about it anymore objectively that day
idle owl
bah
how do I word "We hope to resolve this in the near future" without using "we"
which is counter-indicated in guides
blissful pollen
hmmm
This issue is targeted to be resolved...
This issue is planned to be resolved...
blissful pollen
Maybe "Track this issue <GH issue link here>"
idle owl
That's an excellent idea, but I'm not sure if that might confuse folks more. 😕
The issue is being investigated, and a fix is expected in the near future.
?
vague but valid?
blissful pollen
I like that too... even The issue is being actively investigated (if that is true makes it sound like it is a priority)
blissful pollen
Welcome! Let's not tell me 20 years ago his primary job would be writing documents and away from development 🙂
idle owl
😄 I feel you
tulip sleet
i agree with all this; the current issue with ESP32V2 is turning out to be even weirder, so don't bother with details
idle owl
Thanks for follow up. Glad we nailed it.
olive turret
Just checking - is this the best channel for someone with questions about the actual process of building CircuitPython from the source on github? There's someone over in #help-with-circuitpython having build issues, and I wanted to know if this is the correct place for them.
idle owl
Just checking - is this the best channel for someone with questions about the ac...
Yep, this is the best channel.
olive turret
Yep, this is the best channel.
Thank you! 👍
idle owl
@tulip sleet With the .env file removal being necessary, is that something worth noting on this page? Or might that not be the case once a fix is in and I shouldn't mention it? Or no idea with it getting weirder...
tulip sleet
I don't that's related to power consumption. it's related to thonny issues
olive turret
And ... since I'm here, I have a really weird one from my one attempt to build CircuitPython. I got it all building and running just fine on a Feather RP 2040, everything looked really peachy. I then tried importing an MPY file from the bundle, and it raised a really odd exception - it complained about incorrect byte order.
I have a suspicion I know what's going on, but have no clue whatsoever (a) if I'm right, and (b) if so how to fix it.
Many many many years ago (like the mid 1980s) I worked with the UCSD P system, that was a very early "byte-code" system. P-Code files were composed entirely of 16 bit words, but to deal with the "endianness" issue e.g. Intel vs Motorola CPUs of that era, the first word in the P-Code file was always 1. If you read the first word, and got 256, you knew the whole file was byte-swapped, and could react appropriately,. Does this sound even remotely plausible for what I had happen?
tulip sleet
And ... since I'm here, I have a really weird one from my one attempt to build C...
could you show the exact error, and which file were you importing?
olive turret
I probably can, but it may take a while. I've since flashed it with the uf2 from Adafruit, i I'd have to reflash it with my image. Which means backing up what's on it, for the sake of safety. Give me a few minutes ...
tulip sleet
@thorny dove the missing "minify" you are getting would be due to not installing some pythong packages: https://learn.adafruit.com/building-circuitpython/build-circuitpython#install-required-python-packages-3086211
tulip sleet
I probably can, but it may take a while. I've since flashed it with the uf2 fro...
what platform did you build it on?
thorny dove
From the CircuitPython repo I tried to build for a board using make BOARD=espressif_esp32s3_box . The build appeared successful ('Build files have been writtten to: ...') however: a) no .bin neither .uf2 is created; b) one error occurred: ModuleNotFoundError: No module named 'minify_html'. The CPY repo is V8.0.0-beta.0 with esp-idf version 4.4.1. The build took place from within a MS Windows 11 WSL2 (Ubuntu 20.04) session.
tulip sleet
From the CircuitPython repo I tried to build for a board using ```make BOARD=esp...
Did you see my suggestion above about doing the pip3 installs?
the minify package is installed when you do pip3 install --upgrade -r requirements-dev.txt as described in the link
thorny dove
@tulip sleet. I did all the steps in the 'Install Required Python Packages' (and everything else on that page)
I can try to repeat the pip3 install steps
tulip sleet
minify_html was added rather recently, so if you did it a while ago, it would not have been isntalled
$ pip3 list |grep minify
minify_html 0.10.0
is what I have now
thorny dove
@tulip sleet i just dit the requirements_dev one and yes... I see: Collecting minify_html Downloading minify_html-0.10.0-cp38-cp38-manylinux_2_27_x86_64.whl (699 kB) ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 699.3/699.3 kB 75.5 MB/s eta 0:00:00
olive turret
what platform did you build it on?
Sounds weird, but it was built in a Docker container based on Ubuntu:2004, hosted on a Windows system.
tulip sleet
do a make BOARD=... clean and then try make BOARD=... again
thorny dove
@tulip sleet I have to add that I did a fresh clone of the CircuitPython repo after I had various problems with the previous clone (in which I the the pip3 installs today also)
olive turret
Lemme get the container spun up first ...
tulip sleet
Sounds weird, but it was built in a Docker container based on Ubuntu:2004, hoste...
I wonder if it was something being treated as a text file when it should have been binary
thorny dove
@tulip sleet maybe I forgot to repeat the steps in the new clone
tulip sleet
the pip3 installs are not specific to the clone
thorny dove
Ok
tulip sleet
but you have to do them inside the venv created by esp-idf/export.sh
thorny dove
I just did
thorny dove
OK
@tulip sleet I saw various times during the pip3 install Requirement already satisfied: So, various requirements were already there.
tulip sleet
minify_html was added to support the web workflow, which uses HTML and javascript
thorny dove
Thanks for the elaboration
tulip sleet
i have to get off for a while, I will circle back in something over an hour
olive turret
@tulip sleet I wouldn't have thought so. I used git inside the container to do the check out, and I know from past experience that files read and written by a linux container on a Windows host respect linux conventions, not Windows.
thorny dove
@tulip sleet Now the build goes steps further as b4: ```Partition table binary generated. Contents:
ESP-IDF Partition Table
Name, Type, SubType, Offset, Size, Flags
nvs,data,nvs,0x9000,20K,
otadata,data,ota,0xe000,8K,
ota_0,app,ota_0,0x10000,2M,
ota_1,app,ota_1,0x210000,2M,
uf2,app,factory,0x410000,256K,
user_fs,data,fat,0x450000,11968K,
[8/826] Building C object esp-idf/mbedtls/mbedtls/library/CMakeFiles/mbedcrypto.dir/bignum.c.obj```
Wow, in WSL2 the time to build is way much longer as in a real Linux / Ubuntu session.
olive turret
@tulip sleet directly from the repl using my uf2:
>>> import board
>>> import digitalio
>>> import busio
>>> from adafruit_wiznet5k.adafruit_wiznet5k import WIZNET5K
>>> import adafruit_wiznet5k.adafruit_wiznet5k_socket as wiz_socket
>>> spi = busio.SPI(board.SCK, MOSI=board.MOSI, MISO=board.MISO)
>>> cs = digitalio.DigitalInOut(board.D10)
>>> eth = WIZNET5K(spi, cs)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 202, in __init__
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 222, in set_dhcp
File "adafruit_wiznet5k/adafruit_wiznet5k_dhcp.py", line 490, in request_dhcp_lease
File "adafruit_wiznet5k/adafruit_wiznet5k_dhcp.py", line 383, in _dhcp_state_machine
File "adafruit_wiznet5k/adafruit_wiznet5k_dhcp.py", line 234, in send_dhcp_message
File "adafruit_wiznet5k/adafruit_wiznet5k_socket.py", line 264, in send
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 791, in socket_write
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 872, in _get_tx_free_size
ValueError: Invalid byteorder
Whereas if I switch to the uf2 from Adafruit's web site, everything works perfectly.
crimson ferry
@olive turret is this relevant to the byte-order issue? https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/pull/64 (via https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/issues/66#issuecomment-1241211958)
manic glacierBOT
olive turret
<@270010834732711939> is this relevant to the byte-order issue? <https://github....
That's probably it. The line that's raising the exception in my case is:
return int.from_bytes(val, "b")
However, I'm actually using the mpy files from the bundle, I just have the bundle source that I can look at for reference. TL;DR somewhere between the downloaded uf2 getting built, and mine getting built, this error crept in. Which also has me puzzled because I'd gone out of my way to make sure I had checked out the same source as was used to build the reference uf2.
thorny dove
@tulip sleet good morning. Today I booted my desktop PC into real Linux/Ubuntu 20.04. In the home folder I cloned the CPY repo. I did all the needed steps as described in the link you sent me yesterday. Then a successful build for BOARD=espressif_esp32_box took less than a minute! After reset the CIRCUITPY drive was mounted. I was able to open and read the code.py file. So, I guess, things are looking OK. I go to reboot my PC into MS Windows 11. Then I go to try to install @lone axle's flipclock repo.
tulip sleet
<@329766224093249548> good morning. Today I booted my desktop PC into real Linux...
In the past WSL had slow file access. I thought they fixed that mostly in WSL2. Also, are you using make -j<n> ?: https://learn.adafruit.com/building-circuitpython/build-circuitpython#use-all-your-cpus-when-building-3086075
thorny dove
Yes I was using -j6
Anyhow the build did not finish. Somewhere it hang. I have made a screenshot of the last group of lines the build process spit out to the WSL2 terminal window
blissful pollen
@tulip sleet My comment about the Funhouse SPI speed was just out of curiosity why it was so slow vs other boards. Then this morning I remember there was a Funhouse display glitch that was solved by slowing SPI down. And I may have been the one to do that and had forgot 🙂
thorny dove
@tulip sleet build result in Linux/Ubuntu session for BOARD=espressif_esp32s3_box.
tulip sleet
Anyhow the build did not finish. Somewhere it hang. I have made a screenshot of ...
that is not the screenshot above, which looks successful?
thorny dove
that is not the screenshot above, which looks successful?
Yes, this image is from this morning (my time) in Linux/Ubuntu session where the build of the BOARD was successful.
Resulting in CIRCUITPY drive
midnight ember
looks like a good build to me, am I missing something?
thorny dove
After the successful BOARD build (shown above), after reset and CIRCUITPY is present the run of code.py results in a immediately reset in safe mode informing me that an 'internal wachtdog timer expired' error occurred. I don't know how to solve this one.
thorny dove
I just read this informative article: https://www.allaboutcircuits.com/technical-articles/watchdog-timers-microcontroller-timers/ I wonder how I can arrange at CPY repo BOARD build time I can disable de internal WDT.
jaunty juniper
no but it means that it crashes basically, we have currently open issues related to crashes I think
thorny dove
OK
jaunty juniper
you know what I don't actually know
thorny dove
I tried this. Seems to help in a part in the beginning
jaunty juniper
usually the internal watchdog is there to detect when some part of the code crashes
thorny dove
Yes, that I read in the article I mentioned
jaunty juniper
but maybe it's not set correctly 🤷
thorny dove
So, it is a safeguard to get control back if something goes wrong.
I repeat 'wdt.feed()' in the main loop of the program
jaunty juniper
or maybe it's crashing when the user-set watchdog triggers, causing the internal watchdog to trigger ?
thorny dove
I inserted these WDT lines after the internal WDT errors appeared immediately after a reset in an attempt to solve them
thorny dove
With this simple piece of code, no WDT error. Ergo: must be a problem in the code or in the used libraries. I go to check all of them.
Btw: this new statusline in CPY V8.0.0-beta.0 doesn't end with a linefeed, so the first print statement will be joined at the end of the CPY status line. Odd.
Here you see that the first print command is joined at the end of the CPY status line
jaunty juniper
yeah there's tweaking to be done to the status line
thorny dove
So, I did a test "by hand" in the REPL of the mu-editor: ```
import wifi as wifi
wifi.radio.connect("<ssid>","<password>")
crimson ferry
@thorny dove esp32-s3 box? possibilities: • difference in config between box and other -S3 devices (particularly that would affect runtime behavior); • differences in network environment
I believe espressif wdt is seperate and distinct from the CircuitPython alarm module watchdog, see: https://github.com/adafruit/circuitpython/commit/d42c811fb9c37ce0eda811f3e47304b75e2ae691
certain wdt errors in the core could trigger safe mode
alarm module is for user-level hangs
thorny dove
@crimson ferry Thank you. Yes that is what is happening. The internal WDT triggers
@crimson ferry the scan for networks, b4 the 'wifi.radio.connect()' worked as you can see in the screenshot
tulip sleet
After the successful BOARD build (shown above), after reset and CIRCUITPY is pre...
This is due to some other bug (something is looping or hanging), so you don't want to disable the watchdog timer)
thorny dove
This is due to some other bug (something is looping or hanging), so you don't wa...
I give up my efforts to use CPY with the ESP32-S3-BOX (for the moment). I flashed CPY 7.3.3 and then the script give a hard crash. I'll wait until I read positive news about CPY for this device.
blissful pollen
Was there a preliminary guide/instructions on getting CircuitPython installed on a C3?
idle owl
Was there a preliminary guide/instructions on getting CircuitPython installed on...
No guide that I'm aware of. Not sure about other docs.
jaunty juniper
I guess it would be the same as on an ESP32, except some C3 boards don't have a serial chip so they might need to be put in bootloader mode, I don't remember if that's required on the QT PY
this guide could be extended to the C3 https://learn.adafruit.com/circuitpython-with-esp32-quick-start/installing-circuitpython
blissful pollen
Thanks. I just was quickly looking at it as I have a C3 sitting on my desk, but then got distracted
orchid basinBOT
random junco
@proven garnet Funny you should tag this issue with documentation an hour ago, I just forked the Macropad repo a few hours ago to work on it: https://github.com/adafruit/Adafruit_CircuitPython_MacroPad/issues/40 - can you help me? I can't figure out where the documentation is - it's not in api.rst, instead I see automodule:: adafruit_macropad - I'm not sure where to go next
jaunty juniper
<@135415127879974912> Funny you should tag this issue with documentation an hour...
random junco
around here: <https://github.com/adafruit/Adafruit_CircuitPython_MacroPad/blob/5...
got it! Thanks!
jaunty juniper
you test it by pip installing requirements.txt and docs/requirements.txt (preferably in a venv personnally) and running:
cd docs
sphinx-build -E -W -b html . _build/html
and thanks again, been a couple years since I used sphinx
random junco
I ran pylint and black without issues after my update, but my commit is failing with an error when pre-commit runs. The full console output is here (I don't know why it lists every file in my venv): https://gist.github.com/prcutler/8a21b6ca3182cda4248ec04930cde4ad
The only thing I see in the summary at the end is: Unfortunately, your project is not compliant with version 3.0 of the REUSE Specification :-( If anyone has an idea, please let me know what I did wrong
jaunty juniper
ah reuse doesn't know to ignore .venv
you can add .venv after .env in gitignore, it's in the cookie cutter now, I think that will make reuse ignore it
edgy edge
i'm trying to build Circuit python using the excellent instructions at https://learn.adafruit.com/building-circuitpython. Everything builds correctly, the only issue I appear to have are several directories indicating version control issues, even though git status reports up-to-date. Any ideas? Screenshot below from VS code.
jaunty juniper
submodules are often checked out in "detached head" that might be what it's telling you, ignoring that's is the way it's supposed to be
edgy edge
jaunty juniper
python-semver on HEAD (2001c62) via 🐍 v3.9.6
❯ git status
HEAD detached at 2001c62
nothing to commit, working tree clean
because fetch-submodules will only fetch the minimum I think, if you git submodule sync --quiet --recursive && git submodule update --init it might stop saying that
random junco
you can add .venv after .env in gitignore, it's in the cookie cutter now, I thin...
Thank you, that worked! submitting my first PR now. 🙂
jaunty juniper
but then you'll be pulling in Gigabytes of broadcom SDK 😉
edgy edge
@jaunty juniper thank you. For sure the points me in the right direction
jaunty juniper
or just ignore it, that's what I do with vscode 😁
edgy edge
Doesn't that make it painful to generate pull requests?
in case anyone is interested https://www.cloudbees.com/blog/git-detached-head
jaunty juniper
as you've seen, it doesn't affect git status
it's just vscode being weird
listing submodules as if they were independent git repos I assume
edgy edge
got it. I typed the question before finding the answer ...
jaunty juniper
just close that triangle and ignore it
edgy edge
yup
proven garnet
<@135415127879974912> Funny you should tag this issue with documentation an hour...
Ha! Nice, I was just going through library issues, sorry for the late response, but looks like you're all sorted!
proven garnet
you can add .venv after .env in gitignore, it's in the cookie cutter now, I thin...
This is the intended fix, I think I have an adabot patch waiting for this actually.
jaunty juniper
yeah I was gonna ask about that 👍
proven garnet
Yup, I do: https://github.com/adafruit/adabot/pull/298
spiral dune
When I try git commit now, I encounter the problem that circuitpython.pot is automatically modified. Skipping it can indeed solve the problem, but I'm not sure whether it will have an impact after submitting the PR. I'll do it first.
I know the problem is caused by pre commit:
pre-commit run --all-files
Check Yaml...............................................................Passed
Fix End of Files.........................................................Passed
Trim Trailing Whitespace.................................................Passed
Translations.............................................................Failed
···
sed -e '/"POT-Creation-Date: /d' > locale/circuitpython.pot
···
what's the right way to handle version numbers when building Circuit python? After a build from Main it looks like the version number is set to head, causing subsequent library load to fail. You can see the build number and the failure in the screenshot, pulled from boot_out.txt and REPL
lone sandalBOT
atomic summit
Hi folks... I was putting together CP board definitions for my TinyPICO boards (ESP32) and well, my boards don't have any way of controlling IO0 and IO0 is not broken out for the user to use either, as the boards only have an auto reset circuit on them.
There is no UF2 bootloader for the ESP32, so IO0 is not needed for that, but is there anything else that's going to be a problem with not having a "boot" button possible?
spiral dune
There is no UF2 bootloader for the ESP32, so IO0 is not needed for that, but is ...
ESP32 chip?
jaunty juniper
There is no UF2 bootloader for the ESP32, so IO0 is not needed for that, but is ...
the lack of a button would make it impossible to get into safe mode I think, like the original Huzzah32
but it's supported
atomic summit
the lack of a button would make it impossible to get into safe mode I think, lik...
Ok, No way to do it from the web interface? Though I guess that won't appear if the device is stuck.
atomic summit
ESP32 chip?
TinyPICO uses a PICO-D4
jaunty juniper
yeah it would be nice if we had a solution, dunno how though
atomic summit
I'm getting this... obviously part of the new web workflow.. but no idea how to fix it.
So I have to install minify_html manually?
File "../../tools/gen_web_workflow_static.py", line 7, in <module>
import minify_html
ModuleNotFoundError: No module named 'minify_html'
make: *** [build-unexpectedmaker_tinypico/autogen_web_workflow_static.c] Error 1
make: *** Waiting for unfinished jobs....```Surely not? Shouldn't the build system take care of any dependancies etc ?
jaunty juniper
it's in requirements-dev.txt ?
atomic summit
Are there build docs specific to this?
jaunty juniper
python requirements are installed manually
hmmmmm
maybe it's missing that you need to install them in the ESP-IDF environement ?
atomic summit
I've never had to install any manual python dependancies before when working with CP. Maybe I've just been lucky?
jaunty juniper
that could be done automatically
I have my own script that does all the steps, I don't remember what you really need to do
the guide says to install them at the start https://learn.adafruit.com/building-circuitpython/build-circuitpython#install-required-python-packages-3086211
sometimes you need to redo it to update
but do you need to do it after esp-idf/export.sh ?
atomic summit
not sure... trying now
jaunty juniper
yeah I do it in my script
atomic summit
I installed req and got past that error., but now get new ones ! hahaha
jaunty juniper
oh yeah it says:
You will need to repeat the pip3 install commands described here so that they take effect inside the Python environment set up by source esp-idf/export.sh. But you will need to do this only once.
atomic summit
Maybe I should go to bed and sleep and try again tomorow!
yeh, this is going badly...
from ../../shared-module/displayio/Bitmap.h:33,
from ../../shared-bindings/displayio/Bitmap.h:30,
from build-unexpectedmaker_tinypico/autogen_display_resources-en_US.c:2:
esp-idf/components/newlib/platform_include/assert.h:20:10: fatal error: sdkconfig.h: No such file or directory
#include <sdkconfig.h>
^~~~~~~~~~~~~
compilation terminated.
make: *** [build-unexpectedmaker_tinypico/autogen_display_resources-en_US.o] Error 1
make: *** Waiting for unfinished jobs....
so I'm going to bed!
onyx hinge
what's the right way to handle version numbers when building Circuit python? Aft...
We depend on your clone having the "tags" present (and that the correct tag be findable via the "first parent"). If you used a technique like a shallow clone, these tags might not be present. If your internet bandwidth is not at a premium, doing a fresh full clone is the way I'd recommend proceeding. Otherwise, after creating a shallow clone here's how our CI ensures it has the tags it needs (it's a fragment of the python script tools/ci_fetch_deps.py: ```py
run(
"Fetch back to the start of 2021 to get tag history",
'git fetch --tags --recurse-submodules=no --shallow-since="2021-07-01" https://github.com/adafruit/circuitpython HEAD',
)
run(
"Fetch back to the start of 2021 to get commit history",
f'git fetch --recurse-submodules=no --shallow-since="2021-07-01" origin {ref}',
)
@onyx hinge i used "make fetch-submodules" so would expect to have a shallow clone, per the build instructions. Does ci_fetch_deps.py get involved by make BOARD=.... Or do I need to run it manually first?
onyx hinge
@edgy edge hum these instructions https://learn.adafruit.com/building-circuitpython/build-circuitpython should result in the main repo being complete including tags. In the top level of circuitpython, what does git tag --list "8*" print?
edgy edge
@onyx hinge interesting. When I follow the link you just sent the building circuit python the instructions I see do not include the git submodule update step. See Screenshot. I'll give that a try
onyx hinge
the top code block hasgit clone https://github.com/adafruit/circuitpython.git cd circuitpython make fetch-submodulesand the following text states that the fetch-submodules step functions a substitute for running git submodule type commands.
The initial problem, in which your CircuitPython build is not marked with a version like "8.0.0-..." is probably not related to submodules, because the version number is associated with the circuitpython repository itself.
Initially making a shallow clone of circuitpython itself with --depth=1 reproduces the problem where the description of the git version is just a commit ref string: ```jepler@bert:/tmp$ git clone https://github.com/adafruit/circuitpython.git --depth=1
Cloning into 'circuitpython'...
…
jepler@bert:/tmp$ cd circuitpython/
jepler@bert:/tmp/circuitpython$ tools/describe
14fc4a0
I agree 100% with your assessment, and so does git tag --list "8*" which returns null. What process change is needed to get the version number correctly included?
onyx hinge
If you used git clone https://github.com/adafruit/circuitpython.git (without any --depth argument) as your initial clone command then we need to pursue what element of your git configuration led to you having a shallow clone so that we can update the text for other people who may have a similar configuration.
In any case, the command used during CI appears to fetch the needed tag info: git fetch --tags --recurse-submodules=no --shallow-since="2021-07-01" https://github.com/adafruit/circuitpython HEAD
Does your command history show you what initial git clone command you used?
edgy edge
cut/paste git clone git@github.com:aulitech/circuitpython.git
after creating a fork from main
jaunty juniper
oh from your fork
edgy edge
… Just following directions
onyx hinge
OK, and github recently started changing the default settings for a fork, making it so that it does not copy tags (and calls it "copy the (name) branch only")
onyx hinge
… Just following directions<:adafruit:454328690638061570>
OK, this text will need to be revised:
But if you want to make changes, you might want to "fork" that repo on GitHub to make a copy for yourself, and clone from there.
because github has made "fork and clone" not work in the way we require.
@edgy edge Repeating this so you can try it and let me know if it gets you going:
In any case, the command used during CI appears to fetch the needed tag info:
git fetch --tags --recurse-submodules=no --shallow-since="2021-07-01" https://github.com/adafruit/circuitpython HEAD
edgy edge
thank you!!! Tremendously appreciate the quick response. Not sure how I could've missed something so totally obvious
onyx hinge
@tulip sleet ^^^ you'll want to read back through this later
@edgy edge Thanks for taking the time to troubleshoot with me, we'll make it better for the next person with this information.
edgy edge
building now ...
edgy edge
looks like a winner: Adafruit CircuitPython 8.0.0-beta.0-50-g505b8f5a6e on 2022-09-15; Aulitech Cato with nRF52840
Board ID:aulitech
UID:E3031BF92BB05686
edgy edge
sorry for the duplicate Post. Discord seems to be devouring them
is there an easy way to change the name circuit python to something else per your branding recommendation
onyx hinge
No, we'd rather you upstream your board.
edgy edge
this is just a trivial port of Seeed BLE Sense to optimize for running Neuton ML models with IMU, BLE and battery management. Happy to upstream but I feel like it's an edge case
onyx hinge
Only 77 of the 329 supported boards are from Adafruit, we love adding other makers' boards. Well, we love it when other makers add their boards, to be fair.
edgy edge
OK will do. Hope you don't mind putting up with rookie mistakes. 30 years of embedded experience, but brand new to python and contributing to open source projects
onyx hinge
You're doing great so far! Happy you joined us.
edgy edge
are there any guides for building libraries, cooperative threading, monitoring memory and performance? My goal isn't to get the last bit of performance - it's just to make sure we don't overload the system causing reliability problems. Our products Will be used by disabled people, they need to be really stable.
onyx hinge
for making new modules in C we have https://learn.adafruit.com/extending-circuitpython/overview though I think it's a little stale.
We don't have a multithreading model, though some platforms (espressif) do have freertoos. in python code, we have asyncio.
https://learn.adafruit.com/Memory-saving-tips-for-CircuitPython talks about monitoring and conserving memory from the point of view of python code.
and of course as a last resort you may want to have a watchdog so that your device restarts automatically when it's not running properly https://docs.circuitpython.org/en/latest/shared-bindings/watchdog/index.html
You might want to chat with @tough flax if you're working in the assistive technology realm
edgy edge
The watchdog is number one on my dev list. Thanks for the reference to @tough flax. Our intent is to open source a variety of assistive technology products including hardware and software. We have a roadmap that includes trainable IMU and audio models that can drive BLE input devices ranging from TV remote control to wheelchairs and robotic feeding arms. If you can twitch a body part you can control your environment. We would love to be working with the QT PY RP 2040 but really need integrated IMU and microphone in a wearable form factor so stuck with Seeed
tulip sleet
OK, and github recently started changing the default settings for a fork, making...
So I/we need to update the initial forking instructions, sounds like
lone sandalBOT
manic glacierBOT
I tried this obvious change, and confirmed the duty cycle is set to 100 internally. However, the waveform does not become uniform.
This is before the change:
This is after the change (changing divisor to 0xffff):
The RMT documentation does not explain this, and ...
jaunty juniper
@tulip sleet the NRF uart issue was found on the forum by someone trying to run this guide: https://learn.adafruit.com/uart-communication-between-two-circuitpython-boards/code
maybe the guide code should be updated to test if byte_read instead of if byte_read is None since the fix is not gonna be in immediately ?
tulip sleet
<@329766224093249548> the NRF uart issue was found on the forum by someone tryin...
if not byte_read, yeah, if you make a PR to the Learn guide I will approve it right now.
since the fix is only for 8.0.0 anyway
I will edit the pasted in part of the code in the guide
jaunty juniper
lemme check that I don't forget the not and I'll push
tulip sleet
I just changed the pasted page example
ember iris
@jaunty juniper I am running with about two brain cells, and one of them is about to tap out. But would you be able to explain to me what you did to isolate the IDF update? Specifically the workflow involved to track it down?
jaunty juniper
<@199811382793863168> I am running with about two brain cells, and one of them i...
I usually do it manually because downloading a file is faster than compiling. A bisect is when you use a binary search algorithm to find the first commit that has an error (or anything)
with git you use the git bisect command
get to a commit that is "bad", git bisect start, then tell it it's bad with git bisect bad
then checkout a "good" commit, and git bisect good
ember iris
I usually do it manually because downloading a file is faster than compiling. A ...
Ok! I always mentally think of it as way more techinally nitty-gritty than that. So hearing that makes it feel more doable to me.
How did you decide how far back to look at the beginning of the bisect?
jaunty juniper
then bisect will automatically checkout a commit in the middle (hence the binary search)
make sure it's a compilable commit, make fetchsubmodules check that git status is clean
compile it, install it on a board
then git bisect good or git bisect bad depending on the result of your test
and then back to compiling the new checkout
I stopped when there was 2 commits remaining to test because it was pretty obvious by the git log where I was
I then tested the PR artifacts to be sure
tulip sleet
I usually use my alias alias gitsubupdate='git submodule sync --quiet --recursive && git submodule update --init' instead of make fetch-submodules, which is kinda slow
jaunty juniper
doesn't that fetch a Gigaton of broadcom now ?
tulip sleet
not all the time, if there's no change
jaunty juniper
I don't build the broadcom port so I never fetch it
jaunty juniper
Ok! I always mentally think of it as way more techinally nitty-gritty than that....
I start with the builds available on S3, so I tested 7.3.3 then I tested 8 alpha
tulip sleet
How did you decide how far back to look at the beginning of the bisect?
you find a version that works, and a version that doesn't. sometimes the working version is reported.
jaunty juniper
then I tested the oldest artifact on S3 that is not a tag
if was bad
so I started with those 2
ember iris
Ok so finding how far back to look is as simple as that. For some reason I just assumed there was 'magic knowledge ™️' that dictated where to look. That's helpful to know it's that easy
jaunty juniper
this is a PR old enough that the artifacts are no longer on S3, but when the starting good and bad PR are on S3 I prefer to do it manually by downloading the builds and approximating the middle
much faster than compiling
tulip sleet
I try to avoid bisecting whne it straddles a micropython merge update. There are tons of commits in there, and many of them don't build. You can do git bisect skip to skip bad ones
jaunty juniper
I actually started by searching the PRs that were related to web workflow, tried one and its previous PR but that was bad too, if I had thought of looking for an ESP-IDF update I wouldn't have needed to bisect probably
ember iris
Ok, now I have another question--given it's tied to an ESP-IDF update, what is the way to proceed to find a bug fix? Where do you begin looking?
ember iris
And thank you both so much for describing the process!
jaunty juniper
ah that is whole other can of worms, not my domain. I would probably search current and closed issues in the IDF repo with a choice of keywords
ember iris
Ok that's helpful to know. I don't even know where to begin so I couldn't tell if it was something I should have familiarity with, or if it was that large of a can of worms
jaunty juniper
I assume it could be a new bug in the IDF or a change that requires us to do a thing we didn't do before (revealing a bug that was silent on our side) so it could be hard to find
jaunty juniper
maybe the fact that it seems to fix it when the web workflow is on indicates that CP doesn't close wifi properly on reload or something
tulip sleet
we try to get the minimal CircuitPython example. If we can get a stack trace we can decode the addresses. I have tracked down some things by commenting out chunks of code inside the IDF and then gradually putting things back, also by logging deep inside until I track down the failed line.
ember iris
Ok, I think I'll come back to this with a fresher mind tomorrow to see if the above still makes sense. Thank you both so much for walking me through how you isolate these kinds of issues, and pointing out where you look next to solve it. I really appreciate it
jaunty juniper
and if like me you are easily amused by saying to yourself "fire bad, git bisect good", don't worry, that's... normal. I think.
ember iris
Normal enough in the present company at least.
One of those two brain cells from earlier tapped out and the other one is currently day dreaming of tapping out as well, so I am very amused by saying that to myself
halcyon path
Did you end up getting far with porting libs to circuitpython for Wio Terminal WiFi/BLE support? I asked Seeed in their Discord channels and they said they won’t be doing anything for CP or micropython. Seems a wasted opportunity to me for such a great platform.
stuck elbow
I still think that naming it code.py and not autoexec.py is a missed opportunity.
analog bridge
@tulip sleet I was able to get esp-idf v5.0 working:
https://github.com/adafruit/circuitpython/compare/main...microdev1:update-idf
Can we move to it now or still post CP8?
cobalt grail
I got one of those odd translation strings again.
%q ``must`` be a bytearray or array of type 'h', 'H', 'b' or 'B'
Not sure about the "must" part? if I translate that literally the quotes makes it sound like it is not a must, more like a recommendation.
https://github.com/adafruit/circuitpython/blob/main/shared-bindings/analogbufio/BufferedIn.c#L101
stuck elbow
scare quotes
note those are backticks, I suppose someone tried to emphasize the word by typesetting it with a monospaced font?
cobalt grail
In a c file? 🙂
stuck elbow
¯_(ツ)_/¯
analog bridge
its from here:
// Bytes Per Sample
if (bufinfo.typecode == 'h' || bufinfo.typecode == 'H') {
bytes_per_sample = 2;
} else if (bufinfo.typecode != 'b' && bufinfo.typecode != 'B' && bufinfo.typecode != BYTEARRAY_TYPECODE) {
mp_raise_ValueError_varg(translate("%q ``must`` be a bytearray or array of type 'h', 'H', 'b' or 'B'"), MP_QSTR_buffer);
}
More of an emphasize, I believe.
jaunty juniper
trying to emphasize with sphinx docs syntax in an error message
cobalt grail
That would be my suspicion yes. Point being - I don't work in the CP repo at all, so I don't fix error messages in code.
stuck elbow
maybe we should ask the author
stuck elbow
I got one of those odd translation strings again.
```%q ``must`` be a bytearray ...
@hidden rain why did you use backticks on the "must" here?
cobalt grail
Weblate does not show the author so I wouldn't know who to ask. Thanks for doing it for me. 🙂
tulip sleet
<@329766224093249548> I was able to get esp-idf v5.0 working:
https://github.com...
Thank you for doing the work! I have been reluctant to move to 5.0 because it may introduce new issues in wifi or I2C, etc. If you submit it as a draft PR we could test the builds. There are many changes and it hasn't been shaken out that much. We could do it for 8.1. We still have a lot of things to fix or look at (or push forward) for 8.0.0.
I am hoping for S3 I2C fixes but I am not optimistic, because I tried to test master independently and I still had problems.
analog bridge
a draft PR sounds good
jaunty juniper
@tulip sleet would it be possible to consolidate the error message with ?
shared-bindings/audiocore/RawSample.c
91: mp_raise_ValueError(translate("sample_source buffer must be a bytearray or array of type 'h', 'H', 'b' or 'B'"));
that's an excellent idea, I will do that
tulip sleet
<@329766224093249548> would it be possible to consolidate the error message with...
force pushed a change
cobalt grail
Thanks guys for being so fast handling this! 😎
tulip sleet
Thanks guys for being so fast handling this! 😎
We very much appreciate your noticing these things; you move fast too!
jaunty juniper
circuitpython_community_library_list.md in the Community Bundle is still edited manually right ?
tulip sleet
`circuitpython_community_library_list.md` in the Community Bundle is still edite...
Yes, I think all those lists in all bundles are manual (since they require categorization..
tulip sleet
@onyx hinge Is it the fork or the clone that is set up wrong? Do existing forks that were not set to just copy the main branch work OK still? I'm confused whether this is a github or a git problem
we should add text to the github-mentioning guides that say to uncheck that box
CircuitPython must be built from a git clone with tags.
If you cloned from a fork, fetch the tags from adafruit/circuitpython as follows:
git fetch --tags --recurse-submodules=no --shallow-since="2021-07-01" https://github.com/adafruit/circuitpython HEAD"""
"If you cloned from a fork that copied only the main branch"??
onyx hinge
@tulip sleet yes the most obvious and severe manifestation of this is the new github behavior, where a fork doesn't get tags. It used to get all the tags and branches.
But in thinking about it, there's also a problem when a user's fork has been around for awhile. You can click in the github UI to get your main branch updated with ours, but neither your repo nor your local clone will get tags just by following that procedure. So, once 8.0.0 (or any release) is tagged, we need to make sure the tag is visible during the build process or the build will be identified wrong.
I wouldn't experience this because in my personal workflow, adafruit is always "origin" and my own fork is always "jepler" but that's just one of many ways to do it, and not what the guide suggests (because dealing with multiple remotes is advanced and in many cases can be avoided?)
tulip sleet
I wouldn't experience this because in my personal workflow, adafruit is always "...
I always do git fetch adafruit and git merge adafruit/main to update, and I seem to get all the tags.
I am kind of dissatisfied with make fetch-submodules. It's slow, and not necessary all the time. If we could get rid of the broadcom firmware fetching, I feel like we could go back to the old way
onyx hinge
I mean that using e.g., the "sync fork" button won't do anything with tags (to the best of my knowledge) and that's what a large number of folks are going to do unless/even if a guide advises them otherwise
stuck elbow
git pull --tags
tulip sleet
git pull --tags
but pull from the upstream repo?
onyx hinge
I don't really know how others use git, there are many many ways to do it, but I assume many people use these graphical / webby things
stuck elbow
but pull from the upstream repo?
git pull --tags adafruit main
that's what I do, anyways
tulip sleet
I think git fetch does --tags by default ... yes, it does
onyx hinge
also fetched; the effect is to fetch tags that point at branches that
you are interested in. This default behavior can be changed by using
the --tags or --no-tags options or by configuring remote.<name>.tagOpt.
cool
tulip sleet
I try to keep the adafruit/circuitpython repo free of stasle branches, etc., so it's not necessary to ignore that stuff locally
onyx hinge
Deleted tag '8.0.0-beta.0' (was 43e18f171)
jepler@bert:/tmp/circuitpython$ git fetch origin
jepler@bert:/tmp/circuitpython$ git fetch --tags origin
From https://github.com/adafruit/circuitpython
* [new tag] 8.0.0-beta.0 -> 8.0.0-beta.0
```Despite whatever the docs say, "git fetch <remote>" on its own doesn't fetch all tags. I don't think my personal global git configuration is affecting that.I think it fetches them at the time it fetches the commits that are tagged. If it doesn't need to fetch the commits again it will not check for new tags
onyx hinge
yes I think the more nuanced explanation is something like that
tulip sleet
the broadcom repo is an abuse of git; they could use LFS support, but they don't, I think
we might consider making our own cleaner version somehow, though i'm not sure how without destroying history
analog bridge
```jepler@bert:/tmp/circuitpython$ git tag -d 8.0.0-beta.0
Deleted tag '8.0.0-b...
For me git fetch <remote> does the job:
mdev@PC:~/code/cpy-update-idf$ git tag -d 8.0.0-beta.0
Deleted tag '8.0.0-beta.0' (was 43e18f171)
mdev@PC:~/code/cpy-update-idf$ git fetch origin
From https://github.com/adafruit/circuitpython
* [new tag] 8.0.0-beta.0 -> 8.0.0-beta.0
maybe you two can compare git versions
analog bridge
Mine is git version 2.25.1 Ubuntu 20.04 WSL
onyx hinge
git version 2.30.2 (debian bullseye)
I tried HOME=/ git fetch origin to eliminate my ~/.gitconfig settings as a possible cause/culprit but it didn't change the behavior
onyx hinge
and origin is origin https://github.com/adafruit/circuitpython (fetch)
tulip sleet
halbert@tuna:~/repos/1circuitpython$ git tag -d 8.0.0-beta.0
Deleted tag '8.0.0-beta.0' (was 43e18f171)
halbert@tuna:~/repos/1circuitpython$ git fetch adafruit
From https://github.com/adafruit/circuitpython
* [new tag] 8.0.0-beta.0 -> 8.0.0-beta.0
halbert@tuna:~/repos/1circuitpython$ git --version
git version 2.34.1
so mine is working like MicroDev's
onyx hinge
Deleted tag '8.0.0-beta.0' (was 43e18f171)
jepler@bert:/tmp/circuitpython$ git fetch ssh-origin
From github.com:adafruit/circuitpython
* [new tag] 8.0.0-beta.0 -> 8.0.0-beta.0
``` ssh and http behave differentlyI'm getting more confused by this
here, it's repeatable that origin and ssh-origin behave differently 
analog bridge
mine is https as well
origin https://github.com/adafruit/circuitpython (fetch)
onyx hinge
weird. 2.34.1 from bullseye-backports behaves the same for me as 2.30.2 did.
url = https://github.com/adafruit/circuitpython
tagOpt = --no-tags
fetch = +refs/heads/*:refs/remotes/origin/*
```OK my local .git/config was sabotaged somehowmy error, please proceed
analog bridge
well, git fetch --tags origin works in both cases
Its better to be explicit about it.
tulip sleet
@cobalt grail I cancelled the https://github.com/adafruit/circuitpython/pull/6910 build with your translation done before https://github.com/adafruit/circuitpython/pull/6911, and we'll await the touched-up translation. Thanks.
thorny dove
just following a couple of minutes this interesting conversation . My git version 2.37.3 in MS Windows 11 WSL2 Ubuntu
jaunty juniper
@tulip sleet the guide doesn't mention circuitpython_community_library_list.md https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/sharing-in-the-community-bundle#updating-the-drivers-dot-rst-library-list-3019623
that's because it describes the Adafruit Bundle in that section. And in fact
I noticed that the community bundle and the Adafruit bundle don't have the same structure (the second one has a docs directory)
I think the guide could mention the difference, and maybe we should change the community bundle to the same structure ?
tulip sleet
<@329766224093249548> the guide doesn't mention `circuitpython_community_library...
@idle owl see above re community bundle? (I know you are afk)
cobalt grail
<@519101478435291146> I cancelled the <https://github.com/adafruit/circuitpython...
Yeah, the build stuff takes way too long to finish before it ends up in weblate so I moved on the doing the autumn harvest in my kitchen window (hydroponics, ping me in another channel if that interest you!)
Translation done now anyways.
onyx hinge
``` I just sent an ICMP PING packet with CircuitPython on a PiCow
digital shoreBOT
onyx hinge
ooooh and now I received the response! ```>>> [wifi.radio.ping(addr) for _ in range(20)]
[0.021, 0.004, 0.008, 0.004, 0.003, 0.002, 0.004, 0.002, 0.002, 0.002, 0.003, 0.002, 0.009, 0.003, 0.002, 0.004, 0.002, 0.016, 0.003, 0.004]
min(), max()
(0.002, 0.021)
manic glacierBOT
tulip sleet
@jaunty juniper @proven garnet I am seeing these errors when installing a particular library with circup:
$ circup install adafruit_ble_heart_rate
[...]
Searching for dependencies for: ['adafruit_ble_heart_rate']
WARNING:
adafruit-blinka-bleio~ is not a known CircuitPython library.
WARNING:
typing-extensions~ is not a known CircuitPython library.
Ready to install: ['adafruit_ble', 'adafruit_ble_heart_rate']
Installed 'adafruit_ble'.
Installed 'adafruit_ble_heart_rate'.
it still works, but I thought the ~ issue was fixed? I can't remember if either of you know about this
jaunty juniper
funny you should mention it: https://github.com/adafruit/circup/pull/153
tulip sleet
okee dokee, I will merge, release, and test!
proven garnet
The fastest fix I've ever seen! 😄
stoic rain
Did you end up getting far with porting libs to circuitpython for Wio Terminal W...
I made some progress (I was able to get the underlying eRPC to generate messages) but I stalled on getting the comms to the Realtek chip working. The pin documentation is a little unclear. Also, the eRPC library is pretty large, so I'm going to have to cut it down to fit alongside CP.
I plan to get back to it now that summer is over. I'd really like to have WiFi/BLE support in CP on the WIO Terminal.
atomic summit
Got it!
atomic summit
So what do we do about the VID/PID for the board definitions for ESP32 (no native USB) ? They obviously don't do anything, so can we just remove them?
manic glacierBOT
atomic summit
yep, instead we use creator ID and creation ID
Oh, interesting. So how do I apply for a Creator ID? Is there a form or something I need to fill in?
jaunty juniper
just a PR I think
atomic summit
hmm, ok, but how do I decide what my creator ID would be? They don't seem to be sequential?
jaunty juniper
dunno, it's still rather new and unused and undocumented, it looks like it's first come first served
somebody already took 1337_1337 but nobody took DEAD_BEEF yet 😬
atomic summit
Ok, I did a PR for 0xB0B0_0000 🙂
Though I'm not sure I understand hoe CREATOR_ID and CREATION_ID correlate....
I can just use any 32 Bit number for my creations now? so long as they are with my 0xB0B0000 creator id?
jaunty juniper
it's like VID and PID but twice the bytes
jaunty juniper
yeah
atomic summit
Ok, cool
jaunty juniper
my understanding is that "community managed" creation IDs are in the repo, other can go anywhere they want, like your own repo
manic glacierBOT
jaunty juniper
I like how the community managed Espressif list uses 0032_**** for ESP32, 00C3_**** for C3 boards and such :p
atomic summit
my understanding is that "community managed" creation IDs are in the repo, other...
Thanks for your help !
hmm, is there no mem_info() in cpy ?
jaunty juniper
yeah I don't think it's enabled
manic glacierBOT
atomic summit
So no way of checking available memory from inside CPY?
There must be as I check that on my test jigs! I'll look at what I'm doing for that when I'm back in the office.
atomic summit
Yeah, that's prob it, ta !
proven garnet
Dependabot should now support compatibility release notation properly! (~=)
I can look back into that, before it was a bust and all uprevs created PRs
manic glacierBOT
manic glacierBOT
manic glacierBOT
atomic summit
Oh, do we still need to add our new boards to a build script? Been a while since I've added one, and I can't seem to find the file to add them to.
ornate breach
Oh, do we still need to add our new boards to a build script? Been a while since...
It used to be build.yaml but I believe it automatically adds boards now
atomic summit
Yeah, I couldn't find any yaml files to add my boards in - thanks mate!
ornate breach
Np
orchid basinBOT
edgy edge
I need to bind a third-party cpp library to CrcuitPython. This doc looks dated https://learn.adafruit.com/extending-circuitpython and getting pybind11 integrated into an nrf build hasn't worked for me yet. Any pointers to current documentation or examples would be appreciated. I prefer a Library implementation, not native. I did find the examples in the repository and assume usercmodule is the right starting point, but need the noob info from the readme that isn't there.
stuck elbow
circuitpython doesn't support c libraries that are not compiled in
what hasn't worked exactly? what did you try and what error did you get?
manic glacierBOT
onyx hinge
@edgy edge at the moment there is no C++ code (as opposed to C code) in CircuitPython, so depending which language you mean that's going to be an additional hurdle for you to pass.
Besides the guide which admits right up front it's out of date, looking at the whole content of an actual wrapped C lib turned into a Python module might be instructive to see more about the changes involved. https://github.com/adafruit/circuitpython/pull/2337 is one such example, it adds an external mp3 library written in C then creates the necessary wrapping. It'll let you see things like adding the actual library code as a submodule, adding lists of files to the build, etc.
edgy edge
@onyx hinge - thank you that's exactly what I was looking for. @stuck elbow I tried using pybind11, but could not find instructions for cross compilation and ended up with x86 executables. I also tried following the guide, at the end the instructions to the makefile appear incorrect - it looks like SRC_PATTERNS gets set in circuitpy-defns.mk based on a new #define that would go in either nrf or my board mkconfigport.mk
onyx hinge
I have not seen any indication that pybind11 will work with circuitpython or micropython. "Python 3.6+, and PyPy3 7.3 are supported" - https://pybind11.readthedocs.io/en/stable/index.html
circuitpython's C API is unrelated to CPython's C API
edgy edge
I understand that now. Learning by doing. I really appreciate the great help you provide and the great information in the guides. The commit in the example you provided makes things very clear.
ornate breach
Has anyone attempted to put circuitpython on the SAM4 from microchip?
analog bridge
Can someone with write-access to Adafruit's github page, fork https://github.com/espressif/esp-iot-solution and create a branch named circuitpython.
This is needed for esp-idf v5.0 upgrade. Above espressif repo is a submodule but it needs some customization to work with CP.
onyx hinge
Is there a "usb storage disable", or just read-only?
stuck elbow
Is there a "usb storage disable", or just read-only?
you can disable the msc entirely
tulip sleet
Is there a "usb storage disable", or just read-only?
storage.disable_usb_drive()
onyx hinge
ah I was looking at all the "usb_" modules
does that make the drive writable by code.py too?
stuck elbow
no, you still have to remount it
tulip sleet
does that make the drive writable by code.py too?
it could be, I thought maybe it could check. I think there was some discussion about this but I don't remember the details.
stuck elbow
explicit is better than implicit
tulip sleet
Can someone with write-access to Adafruit's github page, fork `https://github.co...
@onyx hinge I can do the above
onyx hinge
thanks dan
tulip sleet
@analog bridge should be done now
analog bridge
One more thing, can you create circuitpython branch on https://github.com/adafruit/esp32-camera
onyx hinge
OK this seems to do what I want ... ```py
import storage
import usb_cdc
import usb_midi
import microcontroller
from code import TandyKeyboard
tandy = TandyKeyboard()
tandy.ready()
held = (tandy.read_timeout(.5) or [0])[0]
print("held", held)
if held == 1: # Escape key
print("Safe-ish mode: Leaving USB storage & CDC enabled")
elif held == 87: # Keypad enter key
print("Entering bootloader")
microcontroller.on_next_reset(microcontroller.RunMode.UF2)
microcontroller.reset()
else:
print("Disabling USB storage & CDC")
usb_midi.disable()
usb_cdc.disable()
storage.disable_usb_drive()
I can hold one key at power on to enter bootloader, another key to not disable USB features, otherwise the device shows up as a keyboard only
it adds another ~800ms before I can send a first keypress but that's not super important to me
tulip sleet
One more thing, can you create `circuitpython` branch on https://github.com/adaf...
it already has a branch named pass-bus-to-sccb, which was created by @onyx hinge . Jeff, do you want to rename this to circuitpython? I already did a PR to that branch (and it took me a bit of time to figure out which one to PR against)
onyx hinge
esp-iot-solution does?
tulip sleet
no, esp32-camera
onyx hinge
oh yeah sure rename it to circuitpython if that's helpful.
analog bridge
it already has a branch named `pass-bus-to-sccb`, which was created by <@4262005...
Since, we are still supporting v4.4... I need v5.0 support in a new branch on adafruit/esp32-camera
tulip sleet
let's call the existing branch circuitpython-esp-idf-v4.4 and make a new branch from that called circuitpython-esp-idf-v5.0 ? (or drop the "esp-idf" inside??
does that make sense?
@onyx hinge ?
onyx hinge
I am fine with whatever you decide
analog bridge
for adafruit/esp-idf, we are using circuitpython-v5.0 and circuitpython-v4.4
tulip sleet
i think that's ok, unless esp32-camera has its own version numbers, which maybe it does
onyx hinge
they do have version numbers, 2.0.2 is the last release number
tulip sleet
it is 2.0.2, it doesn't say which ESP-IDF it supports 😕
@analog bridge does esp32-camera not yet support esp-idf v5.0?
they are going to have the same problem in their own repo
analog bridge
it does support it but there are still some customizations of our own
tulip sleet
that would be different between idf v4.4 and v5.0, you are saying?
can you use the same techniques they are using to support v4.4 and v5.0 instead of two branches?
are they using #if, etc.?
manic glacierBOT
analog bridge
are they using `#if`, etc.?
they are using #if so a single branch should work...
tulip sleet
they are using `#if` so a single branch should work...
ok, I renamed the branch @onyx hinge made to circuitpython
I should PR the fix I made to upstream also
analog bridge
Thansk! 🙂 Also, lookout for PRs coming to esp-iot-solution and esp32-camera forks
tulip sleet
Reminder of today'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/1k0jC2OVIoslSAK_jhq6ioTEbY5Ep1MChD5UIOSRIHgA/edit?usp=sharing
idle owl
<@330227457296957440> see above re community bundle? (I know you are afk)
I'm not sure I follow what you want to happen here.
thorny jay
Do you know about "the CircuitPython Online IDE" by Twitter user River Wang
@River___Wang ? https://urfdvw.github.io/CircuitPython-online-IDE/
onyx hinge
I wonder what this "~" is doing here. ```
//| def bind(self, address: Tuple[str, int]) -> None:
//| """Bind a socket to an address
//|
//| :param ~tuple address: tuple of (remote_address, remote_port)"""
@proven garnet I see you typing .. I would love your thoughts about https://github.com/adafruit/Adafruit_CircuitPython_PIOASM/pull/51
proven garnet
It's supposed to tell sphinx to boil down the annotation to the name only, like ~module_a.submodule_b.ClassName gets printed by sphinx as ClassName instead of module_a.submodule_b.ClassName.
But I guess it doesn't work with builtins which makes sense
proven garnet
<@135415127879974912> I see you typing .. I would love your thoughts about https...
I saw this a while ago and like the idea. It would also help us start to detect faulty typing as we get closer to the Missing Type Annotations = 0 (!!). My only concern, and I'd have to look into it more, is that something mypy has issues with things that are done very intentionally for memory savings or other reasons, and I have no idea how frequently that is or what the fix would be other than disabling or refactoring.
But on the whole I'm very pro the idea and think it's worth doing.
onyx hinge
shared-bindings/socketpool/Socket.c://| :param ~tuple address: tuple of (remote_address, remote_port)"""
shared-bindings/socketpool/Socket.c://| :param ~tuple address: tuple of (remote_address, remote_port)"""
shared-bindings/socketpool/Socket.c://| :param ~int backlog: length of backlog queue for waiting connections"""
shared-bindings/socketpool/Socket.c://| :param ~bytes bytes: some bytes to send"""
shared-bindings/socketpool/Socket.c://| :param ~bytes bytes: some bytes to send
shared-bindings/socketpool/Socket.c://| :param ~tuple address: tuple of (remote_address, remote_port)"""
shared-bindings/socketpool/Socket.c://| :param ~bool flag: False means non-blocking, True means block indefinitely."""
shared-bindings/socketpool/Socket.c://| :param ~int value: timeout in seconds. 0 means non-blocking. None means block indefinitely."""
shared-bindings/socketpool/SocketPool.c://| :param ~int family: AF_INET or AF_INET6
shared-bindings/socketpool/SocketPool.c://| :param ~int type: SOCK_STREAM, SOCK_DGRAM or SOCK_RAW
shared-bindings/ssl/SSLSocket.c://| :param ~tuple address: tuple of (remote_address, remote_port)"""
shared-bindings/ssl/SSLSocket.c://| :param ~tuple address: tuple of (remote_address, remote_port)"""
shared-bindings/ssl/SSLSocket.c://| :param ~int backlog: length of backlog queue for waiting connetions"""
shared-bindings/ssl/SSLSocket.c://| :param ~bytes bytes: some bytes to send"""
shared-bindings/ssl/SSLSocket.c://| :param ~int value: timeout in seconds. 0 means non-blocking. None means block indefinitely."""
shared-bindings/ssl/SSLSocket.c://| :param ~bool flag: False means non-blocking, True means block indefinitely."""
```looks like these might benefit from cleaning up, I think my grep maybe found most of themI'll make a gist with these and make issues out of them, fantastic Hacktoberfest issues!
onyx hinge
I saw this a while ago and like the idea. It would also help us start to detect...
two thoughts to that: I could see adding a pyproject.toml setting to SKIP mypy altogether, or for severe cases we can use separate .pyi files to lower the type annotation overhead to zero
jaunty juniper
I'm not sure I follow what you want to happen here.
the learn guide has instructions on how to modify docs/drivers.rst, but this does not apply to the community bundle where the equivalent file is circuitpython_community_library_list.md. I suggest 2 things:
- for now add the mention of
circuitpython_community_library_list.mdto the guide - in the future change the community bundle to use a
docsdirectory, sphinx, RTD anddocs/drivers.rstso that both bundles work the same, with the same instructions
tulip sleet
I'm not sure I follow what you want to happen here.
Neradoc said:
And in fact
I noticed that the community bundle and the Adafruit bundle don't have the same structure (the second one has a docs directory)
I think the guide could mention the difference, and maybe we should change the community bundle to the same structure ?
So the q is whether the bundle repos should have more parallel structure, that's all that i meant
proven garnet
two thoughts to that: I could see adding a pyproject.toml setting to SKIP mypy a...
I think these are really good solutions!
tulip sleet
ok, I crossed with Neradoc's comments
onyx hinge
One other "hmmm" moment I had was wondering if we should switch to having cookie cutter put stuff in src/, src/modulename.py or src/packagename/__init__.py. Because I had to put this new python script in a new folder to not cause trouble
☝️ @proven garnet
proven garnet
@onyx hinge Interesting, you mean the run_mypy.py script?
jaunty juniper
yeah the tooling at different levels relies on explicitly excluding setup.py and example, etc. directories, or having a single python file (other than setup) in the root (that one is for the bundle, not required for frozen libraries)
onyx hinge
yeah the tooling at different levels relies on explicitly excluding setup.py and...
right that's exactly what I encountered
there are other things it'd make easier, like the way we exclude pylint examples, tests, and the main code all differently
onyx hinge
I'm not sure what that means, "a wrench"
proven garnet
Sorry, I meant that throws a wrench in the simplicity. I'm not very good at idioms haha, I'm not even sure I got that one right.
Does using a mypy pre-commit action give us more granularity?
Oh, that's your comment
onyx hinge
I liked pre-commit because (sample of 1) it makes it just happen, and devs should be able to run it locally with ease .. in reality, who knows how it'll work out
proven garnet
I agree, also since it hopefully means it's fixed locally before being pushed, rather than working through multiple CI runs like sphinx sometimes requires.
onyx hinge
I think I linked to a blog post by the person who devised the technique I followed, if you take a look at that blog post you can see other possible trade-offs.
proven garnet
I'll definitely take a look tonight, thanks for bringing it up!
onyx hinge
I appreciate your thoughtfulness about stuff like this
modern wing
Happily lurking 🙂
onyx hinge
Hummmmm now ONLY the 'no milestone' milestone is listed in the adabot report ```
- 5 active milestones
- 5 issues not assigned a milestone
Also @onyx hinge would it be better for me to create three issues for the tilde issue (one per module) or one big one? More issues means it can be broken up between people, but also understand if that's more cumbersome.
idle owl
Hummmmm now ONLY the 'no milestone' milestone is listed in the adabot report ```...
Was that way last week too.
proven garnet
Was that way last week too.
Preemptive oops, possibly. I can dig around if that helps.
idle owl
Preemptive oops, possibly. I can dig around if that helps.
Please do!
onyx hinge
I added 8.0.0 and 7.3.x manually just now
tulip sleet
ty!
modern wing
I hear both of you, loud and clear.
idle owl
<@329766224093249548> I don't hear you if you're talking.
Apologies, Discord was set to go through headphones.
modern wing
yay, audio routing.
onyx hinge
@lone axle are you getting the newsletter links today?
lone axle
yep, I'll get them.
onyx hinge
ty
midnight ember
synthesizer 🐿️
onyx hinge
Becky makes cool youtube videos apart from digikey too. she had a series where she would do a watercolor of a hand tool, that was fun.
midnight ember
i learned how to use resistor values for led's thanks to that episode
lone axle
midnight ember
Congrats Lee! 🎉
I love how cgrover found a massive bug affecting 30 boards and he was so timid about raising his hand again to have people look into it again.
errant grail
I love how cgrover found a massive bug affecting 30 boards and he was so timid a...
Yeah, so shy and reserved.
onyx hinge
The "C" is silent I guess
midnight ember
I recommend referencing it as Csharp python. Using C#Python kids these days might get that confused with a hashtag. :/
thorny jay
I would like a vote here on naming for "the Feature of a CP board to not have a Display/Screen but be designed with the intent/possibility to connect one" like matrix portal (issue: https://github.com/adafruit/circuitpython-org/issues/1032)
1 
Display interface
4 External Display
1 
External Display Connector
0 
Intended for External Display
0 
Optional Display
0 
Display Friendly
2 
Display Ready
^ final result
turbid radish
@onyx hinge I'll look to post a blog post on it
proven garnet
@idle owl The real friends are the bugs we made along the way 😂
proven garnet
<@330227457296957440> The real friends are the bugs we made along the way 😂
And that's the royal we of course haha
analog bridge
Sorry, a bit out of alphabetical order, its been a while... 😅
idle owl
Sorry, a bit out of alphabetical order, its been a while... 😅
No worries, we simply follow the document. So in the end, it's not a big deal. 🙂
midnight ember
can move it after the meeting, no worries
idle owl
Nah! Don't move it, timestamps are in order.
onyx hinge
@proven garnet I love to start things I don't have the time or energy to see through to the end so you're welcome
midnight ember
ah, neat.
honestly contributing to the meeting is more important than being in order if you get it in, in time.
at least i think so, always cool to hear from everyone
idle owl
@tulip sleet 😂
midnight ember
carpet tools are wicked looking things. leather gloves? nah.. pot holders got this.
errant grail
Relationship between the tolerance parameter and possible number of matching color shade values.
onyx hinge
I would like a vote here on naming for "the Feature of a CP board to not have a ...
this post
errant grail
carpet tools are wicked looking things. leather gloves? nah.. pot holders got th...
Cooking with Carpet.
midnight ember
tulip sleet
ooh, pretty, I like the switch colors
errant grail
Such a beautiful PCB.
proven garnet
<@135415127879974912> I love to start things I don't have the time or energy to ...
It was me!
I love typer!
minor plume
@onyx hinge I’m curious now why/how you learned about emojis in branch names. Sounds like there’s a story in there. 😂
midnight ember
minor plume
<@353373147199832074> https://github.com/jepler/circuitpython/tree/%F0%9F%A5%A7%...
I love it! I’ll bet typing the emojis at the command line is a pain, though.
midnight ember
same, no worries
tulip sleet
mailbox photo is great!
midnight ember
mailbox is adorable!
minor plume
That mailbox graphic is sooooo cute!
midnight ember
haven't listened to paul cutlers latest podcast. going to do that right after this. synths!
minor plume
That's a photo 😄
It’s really cute!
midnight ember
still a great image, cute!
jaunty juniper
Kids' toy.
in the shop when ? 😁
midnight ember
linode, haven't heard that name in a while
onyx hinge
seems to be this amazon item, if you really want one: https://www.amazon.com/-/dp/B004TTGXCM/
idle owl
Came with little plastic mail, and letter stickers to decorate it.
minor plume
linode, haven't heard that name in a while
I didn’t know linode was even still around. I thought the whole world had moved to AWS and Azure.
midnight ember
it comes with a blinka sticker? alright i'm sold. 🙂
thorny jay
Kids' toy.
I would have bet a 3D printing from Neo and Pedro. 🙂
idle owl
it comes with a blinka sticker? alright i'm sold. 🙂
BYO Blinka sticker 😉
But you can spell her name with the ones that come with it!
midnight ember
yeah i was thinking more of a 3d print from Liz, has that Liz vibe to it.
does the flag have a motor attached to it? needs servo.
proven garnet
We could always fork the repo and revert it backwards to test, worst case
midnight ember
so is that battery the part you couldn't talk about last week?
midnight ember
yes
idle owl
Last week is a blur, I genuinely don't remember what you're referring to 😄
midnight ember
said there was new hardware for the project that you couldn't mention until the guide came out.
idle owl
said there was new hardware for the project that you couldn't mention until the ...
Oh! No, that'll be for a new eventual guide.
We may have a new LoRa board coming out. But it's not out yet, don't ask!
thorny jay
Yeah, but will it work on EU frequency, or a variant for EU frequency?
ornate breach
Well now I’m gonna ask
idle owl
Yeah, but will it work on EU frequency, or a variant for EU frequency?
I don't have any idea. It's not out yet, don't ask 😉
onyx hinge
putting the "skerr" in "scourge" today I see
midnight ember
idle owl
Ah right.
midnight ember
ah ok thought it might have been the battery, nvm then shhh
analog bridge
@onyx hinge @tulip sleet we are using espressif repos as submodules with modifications
midnight ember
because the battery is new and HUGE compared to the previous 6600mah triple 18650. i use that one for most of my projects. getting some of the new 10000 mah ones too.
thorny jay
I would like a vote here on naming for "the Feature of a CP board to not have a ...
Last? chance to vote (but I think we have a winner with "External Display").. I would have voted for Display Ready, but I cannot vote because I am the organizer and I had to vote for all of those. 🙂
midnight ember
^ integrity
jaunty juniper
I believe we don't always use a fork of espressif if we can avoid it ?
analog bridge
I'll try and give this a shot on my fork.
analog bridge
I believe we don't always use a fork of espressif if we can avoid it ?
yes, avoiding it is the best way
ornate breach
putting the "skerr" in "scourge" today I see
Mayhaps 🙂
analog bridge
we only have like a couple of files or just a single function that needs changing on espressif side
yep
stack overflow
midnight ember
🥩 overflow
analog bridge
it has QnA with up-votes and answers can be choosen
idle owl
Hmm.
onyx hinge
idle owl
I also worry about diluting the field with more communication options.
But we can definitely look into it.
minor plume
🥩 overflow
I worry that the more communication channels we have, the harder it is for a newbie to know where to go for help.
(Oops. That shouldn’t have been a reply)
onyx hinge
Just to muddy the waters, but Discord ALSO added a "forum" feature 🤯 I don't know what it does, exactly
thorny jay
I did not knew #help-with-circuitpython exist... for a long time.
midnight ember
a valid concern is github starting to charge for that stuff eventually and then the entire community gets held for randsom basically
blissful pollen
I already find that most of the reddit comments I see end up maybe a quick reply from a couple people and a direction to the Adafruit forums/discord
thorny jay
And you can make vote in Discord... with a lot of imagination. 🙂
onyx hinge
screenshot of forum in qmk discord
ember iris
Just to muddy the waters, but Discord _ALSO_ added a "forum" feature 🤯 I don't ...
It's kind of reddit like, they python discord staff wasn't thrilled with it. There's a fairly limited number of tags which reduced it's use there (they were exploring it for a category question's type form on the discord)
analog bridge
Thanks! 🙂
proven garnet
Thanks!
midnight ember
👋
errant grail
Thanks!
minor plume
Thanks everyone! 👋
minor plume
🥩 overflow
🥩 overflow could be a good thing, depending how much it overflows. 🙂
onyx hinge
I'm gonna disconnect from the voice chat. have a good week y'all
idle owl
I need to drop off as well. Have a lovely week!
thorny jay
Oups, my avatar here is a picture of me! And on GitHub it is their default... I already have problem with various names on various places (for historical reason).
ember iris
Thanks all, have a good one!
idle owl
I also would have gone for Display Ready. But it wouldn't have won anyway. 🤷🏻♀️
hidden rain
Is there a live meeting going on now?
blissful pollen
Is there a live meeting going on now?
It finished about 15 minutes ago, started at 2pm EST
hidden rain
ok.. thanks. I'll set my alarm for next week.
lone axle
@tulip sleet here is the backup recording of today's meeting: https://www.dropbox.com/s/9d5me0407etcz16/circuitpython_weekly_meeting_9_19_22.mkv?dl=0
tulip sleet
<@329766224093249548> here is the backup recording of today's meeting: https://w...
tyvm!
hidden rain
So thanks again to all that helped with analogbufio (see hug-reposts). @tulip sleet: Thanks for finishing up. (I was away from work for a week), Is there anything remaining that needs my attention? Also will it make a future release?
tulip sleet
So thanks again to all that helped with analogbufio (see hug-reposts). <@3297662...
It's in good shape! I didn't leave anything dangling that I know of. It is merged into main -- it will be in the next 8.0.0-beta release!
thanks again!
spark summit
Is there any documentation that describes the hardware/peripherals a particular CircuitPython module uses when running, on a per MCU basis? Or do I need to look through the each port's source code to figure that out?
In the future I'd like to contribute to writing CircuitPython modules in C, but I believe I'd need to know what registers/peripherals were currently in use to ovoid conflicts.
stuck elbow
you have to look at the source, also, it may be different for each port
usually they will use a function that finds the first suitable peripheral, and raises an error if all are taken
blissful pollen
You can also ask here and usually someone is around who may have a vague idea on it.
I know most modules that may share (countio and pwmio in rp2040 I think) share a "in use" array
onyx hinge
Yes it's all pretty ad-hoc.
spark summit
@blissful pollen @stuck elbow Thank you so much! That helps a lot.
spark summit
@onyx hinge Gotcha, thanks! But it does seem like there are some safety mechanisms built in to prevent modules from actively commendearing peripherals from one another. Otherwise it would be a free for all 😂 .
tulip sleet
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1eqMwdLfb53MKA5T1Ck5D1lJQCDSs7naD7hZwnAid0-c/edit?usp=sharing
thorny jay
I also would have gone for Display Ready. But it wouldn't have won anyway. 🤷🏻...
Yeah, I was for "Display Ready" too, that sound so much like an Intel or Microsoft marketing message. I guess that is democracy? But the PR is not accepted yet and MakerMelissa did not even had a chance to vote.
idle owl
Yeah, I was for "Display Ready" too, that sound so much like an Intel or Microso...
Well I voted late on your Discord post.
manic glacierBOT
proven garnet
@onyx hinge do repositories need to have a .pytyped (I think that's what it is) file for mypy to work properly?
I realize many libraries are missing those but I can figure out a way to patch them in easily enough if needed
onyx hinge
a file "py.typed" goes in a package
to make a single file module type'able by mypy you need a "-stubs" version which is kinda a pain
https://github.com/adafruit/Adafruit_Blinka/pull/610 it ends up being more like this
so maybe you have to go with .pyi for mypy with single-file modules? I'm not sure.
proven garnet
Ahhhhhh thank you for explaining, I've been rereading PEP 561 over and over again because I didn't quite understanding but you're right.
onyx hinge
looking at stubgen/stubtest https://mypy.readthedocs.io/en/stable/stubgen.html might be instructive too. I'm still learning all this myself, it is REALLY hard to find some of this info, and once found it's hard to be confident it's really the best practice
proven garnet
I'd be really interested in doing the stubgen process, and it's something could replenish the Good First Issues for typing libraries once those are burned down (or mostly).
onyx hinge
Processed 1 modules
Generated colorsys-stubs/colorsys.pyi
jepler@bert:~/src/bundle/libraries/helpers/colorsys$ cat colorsys-stubs/colorsys.pyi
``````py
from typing import Tuple
def hls_to_rgb(hue: float, light: float, sat: float) -> Tuple[float, float, float]: ...
def hsv_to_rgb(hue: float, sat: float, val: float) -> Tuple[float, float, float]: ...
``` if we ever DO need to separate out the existing stubs, stubgen looks good for thatthen you have to manually strip them out of the .py file I think 😕
proven garnet
then you have to manually strip them out of the .py file I think 😕
Manually strip the annotations?
onyx hinge
yes, and mostly the extra imports & tests needed for the annotations
proven garnet
Oh interesting I didn't realize it wouldn't use the annotations. I wonder if you could get the ast module or libcst to works in a tool to assist in some way.
Oh it does look like it does it for function/method arguments based on this PR though: https://github.com/python/mypy/pull/3169
Seems like it has issues with things member variables though
I haven't run it yet though so I might be misunderstanding
onyx hinge
anything I say may be a misunderstanding as well. ttyl, calling it a night here.
proven garnet
Thanks!
edgy edge
I'm trying to add my own custom C module to a circuit python build following instructions from @onyx hinge and an example commit he sent me using an mp3 library. With make V=2 I can see the source modules get built and the .o's linked, but an import of my module fails - not found. Any quick thought on a step I may have missed?
manic glacierBOT
analog bridge
In espressif port, is there anyway to know in Makefile, if a board has psram or not?
tulip sleet
looking...
edgy edge
re-posting a question from last night - I'm trying to add my own custom C module to a circuit python build following instructions from @onyx hinge (he/they) and an example commit he sent me using an mp3 library. With make V=2 I can see the source modules get built and the .o's linked, but an import of my module fails - not found. Any quick thought on a step I may have missed?
tulip sleet
In `espressif` port, is there anyway to know in Makefile, if a board has `psram`...
how about CONFIG_SPIRAM and related stuff from sdkconfig*
analog bridge
how about `CONFIG_SPIRAM` and related stuff from sdkconfig*
how to access it in makefile?
tulip sleet
ah yes, sorry, I was thinking about .h files
analog bridge
I was thinking of having a common sdkconfig for psram as we have for flash and then the psram size can be defined in mpconfigboard.mk
tulip sleet
I think that is a good idea. The current per-board sdkconfig's are not consistent and should be as minimal as possible.
they are partly made from ports/espressif/tools/generate_all_sdkconfig.py
analog bridge
there is a new component in idf v5.0 (esp_psram) which needs to be conditionally added depending on CONFIG_SPIRAM
tulip sleet
also tools/update_sdkconfig.py, not sure when that is used