@slender iron is there something I can do with my supervisor terminal memory allocation branch right now that will make things go more smoothly? do you want me to merge 3561 on the assumption that everything's fine or wait for the number to be seen in the CI?
#circuitpython-dev
onyx hinge
[hmmmm what if we huffman-compressed the "words" array too !? should save dozens of bits, while complicating the access to it when decompressing because now you have to recursively decompress]
slender iron
@onyx hinge sorry, I was on a run. did you get it sorted out?
thorny jay
@slender iron I am going to make a PR to add HCI _bleio to all boards for which it will fit (plain Metro M4, etc.). Right now it's only enabled for the MM4 Airlift Lite board. Would like to get this into the rc
@tulip sleet Can we have: PyPortal familly, Matrix Portal M4, ...
tulip sleet
it is going to be everything except SAMD21 and a few boards with no flash chip (e.g. kicksatsprite)
that is, it is just a regular part of the "full build"
thorny jay
Perfect.
tulip sleet
and no nRF (since they have native ble
i am not sure if the wiring on the pyportal etc, will work, or if there are some needed ESP32 pins that are not connected. I haven't checked that yet.
onyx hinge
@slender iron yeah I think it's just a matter of waiting. 3561 is still in progress in the CI, then the other PRs which were affected can be rebased or have main merged in
https://github.com/adafruit/circuitpython/pull/3561/checks?check_run_id=1260824356 feather m0 of 3561's run is done, let's see the memory usage of fr translation: 253072 bytes used, 624 bytes free in flash firmware space out of 253696 bytes (247.75kB).
@slender iron so if you want I can go ahead and merge that, then I can rebase my own PR
slender iron
thanks @onyx hinge! I had a meeting and lunch
onyx hinge
I'm looking forward to the release -- it will be the highest numbered version of circuitpython ever
slender iron
for a little while at least 🙂
manic glacierBOT
thorny jay
I am playing with a NeoTrellis connected with STEMMA.
With the STM32F405, no problem, the neotrellis_simpletest works ok any way I try.
With the QT-Py Haxpress, if main.py contain "import neotrellis_simpletest" I get
"RuntimeError: maximum recursion depth exceeded"
If I just rename neotrellis_simpletest.py into main.py, then I don't have that problem.
Maybe the QT-Py need a bit more of stack(?) love, no recursion involved, just neotrellis calling seesaw, calling bus_device.
solar whale
What is the difference between the unexpectedmaker_feathers2 and the unexpectedmaker_feathers2_prerelease? Is the "pre-release an earlier rev of the board?
idle wharf
@solar whale yes. There are differences in the pins
solar whale
Thanks, I see the pin differences, but the boards being sold now are the feathers2, correct?
I have some arriving tomorrow and just wanted to make sure I used the correct build.
idle wharf
My board says pre-release on it. I assume yours will be the final versions since Seon was just making those
Mine is also green instead of his production black
slender iron
@onyx hinge is there any way to change the weblate behavior when other changes are merged in?
onyx hinge
@slender iron mmmmaybe, I can check their documentation. How would you like to see it work?
slender iron
I always reset the CI clock on weblate PRs when I merge other PRs in
can we have it not automatically rebase?
and force push
onyx hinge
hm I'm having trouble logging in to weblate 😕
slender iron
merge would be better I think
onyx hinge
if I turn off "push on commit" I don't know how often it will push upstream
slender iron
that way it doesn't need to force push and we can easily see the CI history
onyx hinge
OK I changed merge style to merge
slender iron
thank you!
onyx hinge
I hope that helps
slender iron
we can always switch it back if it doesn't. I expect it will though
blissful pollen
@slender iron when i do make translate it includes a bunch of other translations too (probably commits ahead of me) When I push will it prompt me about those?
slender iron
I'd expect them to be included already
blissful pollen
I'm thinking maybe my local copy wasn't up to date. Maybe I'll try pushing it to my PR and see. Learning opportunity 🙂 thanks
slender iron
👍
lone axle
Does CircuitPython have support for using multiple displays? Like the two screens on the Monster M4sk both showing something at the same time with displayio?
stuck elbow
yes
though most of the time you need to compile your own firmware for this
I think Monster M4sk is the only one that has that enabled by default
orchid basinBOT
solar whale
Congratulations on 6.0.0-rc.0 
idle wharf
Woo Hoo 6.0RC ! 
Congrats to the Adafruit Team and all the CircuitPythonistas !
ornate breach
wooo
slender iron
when does the resolution of a float become worse than an int? maybe we should use ints for time.monotonic at that point
stuck elbow
around 6 digits?
slender iron
should we reset monotonic every VM restart?
tulip sleet
I thought we were already doing that? Or only on hard reset?
tulip sleet
time.monotonic is defined to be a float in the CPython specs
slender iron
@tulip sleet can you check the release notes. I may have said it was bleio_hci
I haven't posted it anywhere yet so we can correct things
tulip sleet
I'm editing it.
ok, now says: The HCI version of _bleio is available on many non-nRF boards (nRF is not included it already has native BLE)
slender iron
👍 thanks!
tulip sleet
i will add the word "since"
slender iron
np 🙂
tulip sleet
🙂
solar whale
@tulip sleet I just loaded 6.0.0-rc0 onto a metro_m4_airlift_lite and tried the ble_uart_echo_test.. It worked with older libraries, but when I updated the libraries it is now reporting "no adapter availible"
tulip sleet
i have to test it; arturo182 also had problems. so which library specifically did you update? airlift? ble? both?
solar whale
all -- from todays bundle
tulip sleet
actgually, hold on, you need a different test program
try this
it's less automatic but more flexible
you have to specifcally get the adapter and pass it to BLERadio()
atomic summit
Oooer, we hit RC !!!! Congrats everyone!
solar whale
OK -- that works!! thanks!
tulip sleet
we will need more examples and a Guide
solar whale
All in good time! Still need new Nina firmware as well...
tulip sleet
that is the big stumbling block...
solar whale
We know it is possible to generate it 😉
So far I have updated a grandcentral, metro_m4_airlift, teensy 4.0, matrixportal and esp32s2_wrover to 6.0.0-rc.0.. no anomalies to report! (after a few user errrors!)
Have to quit for today --- will run more checks tomorrow.
orchid basinBOT
fathom trellis
I'm on the organizing team for PyCascades 2021 (along with @ivory yew) and I wanted to share that the CFP is now open through Nov 10th 2020.
It's an amazing conference that loves proposals from both first time and experienced speakers. @slender iron gave a fantastic talk last year.
We'd love to see some talk submissions from CircuitPythonistas.
Submit a talk at this link: https://pretalx.com/pycascades-2021/cfp
manic glacierBOT
orchid basinBOT
idle wharf
This would be a nice thing to get a ruling on. How do all you feel about the pylint warning to use % in logging. Its failing a PR I submitted for circup.
Discussion about W1203 here: https://github.com/PyCQA/pylint/issues/2354
No rush... the PR isn't urgent. I'll be afk in a bit too.
atomic summit
I'm currently wrangling <- nice term APA support into the UF2 bootloader for esp32s2.
I'm really not the right person to be doing this, and I'm sure whatever I push is going to get shoved back at me, but well, it's an awesome procrastination opportunity!
And I have the APA lighting up - w00p!
orchid basinBOT
atomic summit
Ok, I have the boards.c side of the UF2 APA support implemented and working.
starts red, then goes green when UF2 bootloader is mounted. Flashes orange-y when writing.
Can someone explain to me what the neopixel support in the bootloader_start.c is for? That's Espressif's boootloader level right?
When does this get instigated, and what should I be seeing?
lone axle
Tinkering away at it for the night I managed to get both displays running from CircuitPython! 
I think maybe I was misunderstanding the things that are on board for this device. I think now that board.RIGHT_DISPLAY was descriptively named as the masks right eye looking outward. (with the USB port above it). board.DISPLAY I think is just an alias for RIGHT_DISPLAY so that code from other devices with built-in screens carries over without needing to be changed.
the other display which would be board.LEFT_DISPLAY is not set up by the system, perhaps because it requires using some of the pins from the seesaw. I'm not sure that I could figure out how to set it up at the C layer. But I think I could make a library that initializes it and lets user code just start showing things on it.
stuck elbow
that would be nice
manic glacierBOT
atomic summit
So I just converted my testjig over from flash firmware.bin, then run tests etc to flash uf2 bootloader, then copy firmware, then run tests etc
and my test cycle has gone down from 53secs to 42secs !!!
and it looks awesome with so much flashing APA going on....
actually, sometimes it's faster and sometimes slower. weird. best is 41secs, worst is 49 secs. Same test, different board.
Video for those interested...
timber mango
Hi guys, silly question here, I am completely stuck. 😦 I need to restart my PCB to start over again, like it was when new. Also I need some advice how to get it back listening to a feed via IO. I got lost as I jumped between Arduino/Circuit Py and what not. I only want to make a wearable thing that subscribes to a feed that other people could submit input. For example like a christmas tree color changing feed that people decide over. (In the best of worlds I would get it to listen to triggers from other sources like a Zapier feed where I can get colors from a certain person tweeting a specific word or so, but for now it would be ok with just a feed via my Adafruit IO.) Just how to restart again and what language would be better to concentrate on? We talked about this earlier but I lost my thread and got busy with other stuff and forgot the lot... Best! /Åsa Maria
Also, is there any example code that you would recommend? I cannot find any just now, but I am pretty sure there is one.
solar whale
@tulip sleet FYI -- I just verfied that the ble_hci_uart_echo_test works on a PyPortal after settig the TX./RX pins ```esp32 = ESP32(tx=board.TX, rx=board.RX)
I think I saw this come up in a discussion in the past few days.
solar whale
Also works on PyPortal Titano.
bronze kernel
Guys, maybe a stupid question: how much does circuitpython differ from micropython? Are CP libraries compatible with micropython?
Please ping me if you answer
lapis hemlock
@bronze kernel Here are two documents that should clarify: https://core-electronics.com.au/tutorials/circuitpython-vs-micropython-differences.html, https://circuitpython.readthedocs.io/en/2.x/README.html#differences-from-micropython
bronze kernel
Thanks 😀
lapis hemlock
@onyx hinge I am really close now. Title of the latest git commit: restored circuitpython compatibility.
lunar crown
Is the new "FeatherS2 - ESP32-S2 Feather Development Board" F.C.C. type accepted? I don't see any labels or documentation that says it is.
ornate breach
@atomic summit ^
toxic ice
@lunar crown it's almost impossible to certify a devboard. Even if it has a certified module at the moment you can connect an antenna to it, that certification is voided...
lone axle
Is there an appropriate way to "release" a busio.SPI() object so that the script can work successfully if it's restarted. As is I'm getting an error that CSK pin is in use when the script launches for the second time.
lunar crown
@toxic ice I agree, but that does not remove the requirement if you are openly selling to the public.
ornate breach
@lunar crown adding the words “for evaluation purposes only”
That’s how companies like TI get around it
lone axle
Looks like displayio.release_displays() is what I need, but it also releases the "main" screen board.DISPLAY
thorny jay
@tulip sleet FYI -- I just verfied that the ble_hci_uart_echo_test works on a PyPortal after settig the TX./RX pins ```esp32 = ESP32(tx=board.TX, rx=board.RX)
@solar whale And what/where version of NINA should be used?
ESP firmware: bytearray(b'1.6.1\x00')
That is what I have.
solar whale
@thorny jay there is a "special" version NINA HCI BT.bin attached to this PR https://github.com/adafruit/circuitpython/pull/3310#issuecomment-678344925 -- it si different from the released 1.6.1 but still reports itself as 1.6.1 😦
thorny jay
Thanks. I'll give this a try. I need to check the limitation. Right now I just want to listen to advertisement.
solar whale
At this point, it is the only version that works. The current master of the nina-fw builds 1.7.0, but it does not work...yet.
thorny jay
I may have the right one already... because I already downloaded that... no way to tell if I installed on my PyPortal.
solar whale
@thorny jay for a quick test, this uart echo test works for me https://discordapp.com/channels/327254708534116352/327298996332658690/766763460591484950
connect with the adafruit bluefruit connect app.
crimson ferry
@lone axle What board? I thought boards are supposed to deinit everything on reload except certain reserved pins and any built-in display. But I get that error sometimes on boards that aren't fully fleshed out yet, and have to hard reset.
lone axle
@crimson ferry Monster M4sk
solar whale
@thorny jay if the correct version is not installed, you will get a "No adapter available" error
but on PyPortal make sure you change the TX/RX pins
lone axle
One of the screens gets initialized by the system as normal for built-in board.DISPLAY but the other needs to initialized "manually" with the st7789 library.
That manual one needs the release_displays() call to happen before it. But releasing also throws away the one screen that was already working. So then both have to be set up manually in order to use both.
solar whale
I'll be offline for an hour or so -- good luck!
crimson ferry
@lone axle It looks like they are on separate spi buses? But only the primary is init'd and common_hal_busio_spi_never_reset(spi);. I wonder if there's any drawback to adding the other TFT to the init.
...or maybe there's not infrastructure for that elsewhere in the core
lone axle
It seems like displayio is okay with it (on this device at least). It's a bit out of my depth but after playing with it, I suspect it could be done in the core, but would be trickier than the primary because it requires initializing the seesaw also.
crimson ferry
ah, yeah. Can both be set up manually after a .release_displays()?
lone axle
Yep that is working fine for me. And I no longer seem have problems restarting the code.py
Working on a library now that will do all of the setup and allow user code to start showing things after init()
thorny jay
Thank you @solar whale and great work to @tulip sleet . I was able to connect and the echo is working. I'll assume I tried this firmware weeks ago, when the support was missing from PyPortal code. I'll see how far I can go with Bluefruit PyPortal...
tulip sleet
@solar whale and @thorny jay Thanks for the existence proof for PyPortal! Brent spent a while trying to reproduce the "special build" but there is something different we haven't figured out yet.
solar whale
@tulip sleet glad to help. I keep meaning to try digging into the Nina build as well. May have some time this weekend... oooh -- UPS just arrived with my UMfeathers2 🙂
manic glacierBOT
solar whale
Hmm -- How do you update CP on an Unexpected Maker FeatherS2?
ah --- got it -- RESET with BOOT down then load to /dev/ttyACM0
slender iron
When does this get instigated, and what should I be seeing?
@atomic summit In the second stage bootloader its purple while waiting for the second button press.
@toxic ice I agree, but that does not remove the requirement if you are openly selling to the public.
@lunar crown The normal way is to sell it as a component of a larger system. That leaves the need for certification to the final assembler.
thorny jay
Hi, @tulip sleet am I unlucky and scanning for BLE advertisement does not work with HCI _bleio? Still on the PyPortal where the "echo" was working, but I may have done mistake. Too bad for my Corona application scanner.
Traceback (most recent call last):
File "main.py", line 213, in <module>
File "adafruit_ble/init.py", line 267, in start_scan
NotImplementedError:
And this is my line 213:
for ad in ble.start_scan(minimum_rssi=-127, timeout=scan_time_s):
@thorny jay hci bleio currently ONLY works as a peripheral, not as a central. We wanted to get something working as soon as possible, so we chose peripheral first (with the idea of being able to do BLE UART). I will be working on central, but not immediately.
There's nothing to prevent it from working, but there's a lot of impl to do
thorny jay
Ok, perfect. No emergency, I have my scanner working on CLUE and CPB. But I figured it would be fun on a PyPortal too. My app is only checking for advertisement.
thorny jay
And preparing for that, I connected NeoPixel and NeoTrellis for the first time to my PyPortal. So this is all good.
idle wharf
I'm working on a fix for circup and I'm stuck at the pythonic way to format the loggging message which also makes pylint happy.
logger.info("### Started {}", datetime.now()) This doesn't work.
These two work, but pylint doesn't like f-string or format interpolation
logger.info("### Started {}".format(datetime.now()))```
What's the correct way?pylint should be ok with f-strings. Maybe you just need to update pylint? Or change its minimum python version. Do you think this code is going to run on a version that's pre f-strings?
what doesn't pylint like about the .format version?
idle wharf
This the f-string warning W1203 discussion here: https://github.com/PyCQA/pylint/issues/2354
.format warning is Use lazy % formatting in logging functions
Actually they're both the same really
pylint 2.6.0
and the pylint failure is in the CI for the PR 😉
tulip sleet
ok, I see, the issue is that the logger may do nothing if the .info level is not enabled, but the .format is going to generate a value which will then be discarded. So it's a wasted computation, which is why they suggest using % instead. In the latter case you pass the arguments to the logger function, and it only does the interpolation if that logging level is enabled.
example:
import logging
logging.warning('%s before you %s', 'Look', 'leap!')
example from python docs
idle wharf
OK, I just always stick with F-strings and never learned the more confusing string stuff. I'll give that a shot. Remove the {} in the logging message and replacing with that format.
tulip sleet
also, the logging module already does timestamps if you want it to, so normally you wouldn't include a timestamp explicitly. So in the above, if the .info level is not enabled you are still calling datetime.now(), whose value would be discarded
idle wharf
This is not my orignal code, I'm just tweaking circup so it runs on Python 3.9
tulip sleet
that's what I was just thinking: the original code should not really be including a timestamp, or should not be using the logging module. It doesn't really matter, it's cheap, so anyway just do
logger.info("###Started %s", datetime.now())
idle wharf
and now pylint: "Too many arguments for logging format string"
(this is one of the reasons, I abandoned this style when f-strings was released)
I have to run and work on a home project (shelves don't paint themselves... )
You put me on the right path and I'll sit down this evening and work my way through it. I just need to put in the time and it will be good to learn. Thanks @tulip sleet
tulip sleet
ugh - checking on it
np
not sure what the issue is. this works for me:
import logging
>>> logger = logging.Logger("x")
>>> logger.info("hi there")
>>> logger.error("hi there")
hi there
>>> logger.error("hi there %s", 1+3)
hi there 4
but anyway, have a fun time with paint 🙂 🖌️
manic glacierBOT
tulip sleet
@idle wharf maybe it's just a typo, because the above passes pylint when I put it in a file
idle wharf
@tulip sleet With datetime.now() ?
lapis hemlock
@onyx hinge Jeff, I am sure this is a trivial error, but I just don't see what it is. Here is your test script ```try:
import ulab as np
except:
import numpy as np
for num in range(1,4):
for start in range(-num, num+1):
for end in range(-num, num+1):
for stride in (-3, -2, -1, 1, 2, 3):
l = list(range(num))
a = np.array(l, dtype=np.int8)
sl = l[start:end:stride]
ll = len(sl)
try:
sa = list(a[start:end:stride])
la = len(sa)
except IndexError as e:
sa = str(e)
la = -1
print("%2d [% d:% d:% d] %-24r %-24r%s" % (
num, start, end, stride, sl, sa, " ***" if sa != sl else ""))
a[start:end:stride] = np.ones(len(sl)) * -1
print("%2d [% d:% d:% d] %r" % (
num, start, end, stride, list(a)))``` and it fails on line 15, `sa = list(a[start:end:stride])` with the error message `TypeError: 'int' object isn't iterable`. This happens, when `a[start:end:stride]` is a single value. But I still don't get it. That statement is in the `try` clause, so what the heck is going on here.
By the way, this is the very last outstanding issue. All other tests pass now.
onyx hinge
hmmm
lapis hemlock
Even if I hard-code sa = list(0), it still fails.
onyx hinge
>>> a[0:1:1]
array([1])
On the other hand, micropython is OK with the naked statement try: list(0) except: print('failed')
onyx hinge
shouldn't the result of slicing a be an iterable thing, even if the slice refers to a single element?
desktop python3 rejects list(0), so I would expect CircuitPython to reject it too.
atomic summit
@slender iron Ok, got it. The neopixel implementation between board.c and bootloader_start.c are very different. Are you ok with my APA implementation and am I ok to use it again in bootloader_start.c ?
tulip sleet
@idle wharf ok, it's datetime.datetime.now()
solar whale
@atomic summit do you have a time for a strange issue with a feathers2 board?
lapis hemlock
shouldn't the result of slicing
abe an iterable thing, even if the slice refers to a single element?
@onyx hinge That's right.
atomic summit
I don't want to implement it and then find out it's not what you are after.
@solar whale Sure! What's up?
lapis hemlock
desktop python3 rejects
list(0), so I would expect CircuitPython to reject it too.
@onyx hinge I get that, but why does it hang in atry-exceptstatement?
onyx hinge
The code you pasted only catches IndexError, but you said the type of exception thrown is TypeError.
the code you said you tested in micropython was a bare except: so it would catch a TypeError
solar whale
@atomic summit I just received 2 boards -- one is fine -- the other is acticg very oddly. It will not boot unless I probe one of several I/O pins with my multimeter. So far IO7 IO38 SCL SDA seem to make it boot.. ONce it boots it seem to work until I press RESET, then it does not rebott until I probe the pin again.. Probing voltage from Ground to the Pin)
lapis hemlock
The code you pasted only catches
IndexError, but you said the type of exception thrown isTypeError.
@onyx hinge How stupid of me! Thanks for pointing that out!
solar whale
I also cannot get it to go to the bootloader unless I hold BOOT when applying power -- RESET idoes not work. I can then flash it OK
atomic summit
Hmm... That's super strange. Ive never seen that before. I wonder if it took a knock during shipping.
DM me and I'll organise to send you a replacement.
solar whale
definately odd -- i don't see any obvious issues on the board.
atomic summit
Do you have a hot air gun ?
solar whale
Yes
atomic summit
try giving it a good heat-up with the gun. no hotter than 330deg C though. it's been ultrasonically cleaned so much of the flux is gone, but you might still see a bit come out around the S2.
solar whale
OK -- I'll give that a try.
atomic summit
All I can think of is it got a knock and maybe a bad solder joint from it, or something that might have already been on edge might have gone bad. though, super unlikely.
When you say it wont boot, you mean no USB comes up? but it powers up?
solar whale
Power ligth come on and the 3V is there, bit nothing registers on the USB
solar whale
all better now -- a little hot air worked wonders!
modest hazel
HI! I'm building a CiruitPython IDE that's browser-based and designed especially for use on Chromebooks. Even though it's easy to write code in a text-editor and drop the file on the a plugged in-module, there are still steps in involved in that can be tricky to follow correctly and slow especially when you are iterating your code a lot. The IDE I made loads code temporarily right through the REPL, so as long as the device is "tethered", you can just keep loading/iterating. I also make it really easy to load in any of Adafruit's example code.
I'd love some help beta testing it and seeing how people use it.
You can try it out at https://slappy.io/
ornate breach
Hey @modest hazel try to keep your posts to one channel so it’s not spammy
modest hazel
@ornate breach ok sorry
solar whale
FYI -- I've run into an issue the esp32s2 I2C when using a bno08x IMU -- it works for the "simpletest" but fails (just hangs in read_packet) when trying the more complex "more_reports" test. It works normally on a grandcentral_m4... If anyone has the hardware and time to try to reproduce this on an esp32s2 it would be good to know. My realt question is, would you suggest opening an issue in CP on the esp32s2 or in the bno08x librray?
idle wharf
If anyone had been following along at home when I was talking with @tulip sleet about logging and pylint yesterday. After spending a few enjoyable hours trying different formatting options, I believe the proper syntax is logger.info("The reading is: %s", value_read) and that the pylint warning about it having too many parameters is actually a bug which they fixed previously and either there is a regression or Python 3.9 broke it. https://github.com/PyCQA/pylint/issues/3362
@solar whale that is so timely, I was just last night hooking up a BME680 to my FeatherS2 and the BME680 example doesn't work, so I was going down into bus_device to try the simple test. https://circuitpython.readthedocs.io/projects/busdevice/en/latest/examples.html
I'm curious, how did you know the device address for your board ?
solar whale
The default address in the driver is correct for the bno08x
the har part was getting the SCL/SDA pins correct?
idle wharf
are they not board.SCL, board.SDA ?
solar whale
yes -- they are for the FeatherS2, but not defined for the saola_wrover -- I tried it on both -- same result
I have not tried a bme680 yet -- I think i have one here somewhere -- i'll try that after the dog walk -- priority interrupt 😉
solar whale
I have run several other i2C devices on both the feathers2 and the wrover -- aht20, shtc3 and a few more I've forgotten...
idle wharf
Helpful to hear you're getting further than I did, maybe its the pre-release boards pinouts... 🤔
solar whale
@idle wharf I had to make a small change to the bme680_simpletest -- now it works on my feathers2 - before the fix, it threw an error converting the temperature_offset in the print statement.
I'll check this on an M4 and the put in a PR,
idle wharf
huh... I can't get passed device = I2CDevice(comm_port, addr) .... I get a value error for every possible address.
I should mention.. CP6.0 built locally as of last night's main. Not a released build.
solar whale
Me too -- build today! what code are you running -- I don't see that in the example?
idle wharf
I'm going down the layers to the basic connection (ruling out the samples above)
import busio
import board
from adafruit_bus_device.i2c_device import I2CDevice
DEVICE_ADDRESS = 0x68 # device address of DS3231 board
A_DEVICE_REGISTER = 0x0E # device id register on the DS3231 board
# The follow is for I2C communications
comm_port = busio.I2C(board.SCL, board.SDA)
device = I2CDevice(comm_port, DEVICE_ADDRESS)
for every device address from 0x00, to 0xFF (0-255) it fails
solar whale
just a sec -- need to look up the bme680 address
idle wharf
And 77 is hardcoded into adafruit_bme680
spare prawn
when doing a custom build of circuitpython to enable the bluetooth REPL how does one change the default to enable it? it does not seem to be indicated in the doc.
idle wharf
solar whale
works for me ```Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 6.0.0-rc.0-4-g3c05a8c68 on 2020-10-17; FeatherS2 with ESP32S2
import busio
import board
from adafruit_bus_device.i2c_device import I2CDevice
DEVICE_ADDRESS = 0x77
comm_port = busio.I2C(board.SCL, board.SDA)
device = I2CDevice(comm_port, DEVICE_ADDRESS)
ah -- yeseterdays build...
only transaltions since then
@spare prawn not ignoring you, just don't know the answer...
idle wharf
newp... same ValueError. ValueError: No I2C device at address: 77
Thanks for the help @solar whale. I'll push my forehead against this later.
Could be the library version.... I'll have to check that too.
I've not used the BLE REPL either @spare prawn
solar whale
@idle wharf you are using 0x77, correct. not 77
idle wharf
>>> DEVICE_ADDRESS = 0x77
>>> DEVICE_ADDRESS
119
>>> comm_port = busio.I2C(b
board busio
>>> comm_port = busio.I2C(board.SCL, board.SDA)
>>> device = I2CDevice(comm_port, DEVICE_ADDRESS)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "adafruit_bus_device/i2c_device.py", line 68, in __init__
File "adafruit_bus_device/i2c_device.py", line 182, in __probe_for_device
File "adafruit_bus_device/i2c_device.py", line 180, in __probe_for_device
ValueError: No I2C device at address: 77
have you been able to use any I2C devices?
idle wharf
This was my first go... :-)
I have a production CP Playground Express I'm going to try that too. Things to rule out: my build, pre-release FeatherS2, library version, not enough caffeine...
solar whale
the pre_release pinouts for SCL/SDA are different from the released version.
Do you have a schematic for the Pre-release version?
crimson ferry
Not sure if this is helpful, but I have a pre-release FeatherS2, and I2C works (Qwiic connector) with this setup, I don't have a BN***:```FeatherS2 PreRelease with ESP32S2
6.0.0-rc.0 on 2020-10-16
adafruit-circuitpython-bundle-6.x-mpy-20201017
i2c = busio.I2C(board.SDA, board.SCL) # REVERSED pins
pins.c says SDA is 38, SCL is 33. The product card says SCL is 38, SDA is 33.
solar whale
That would make a difference 😉
atomic summit
@idle wharf I think I might have SDA/SCL backwards on the pre-release board pin definitions. Try switching them around, and I'll also check on a board today as well.
Though technically it should not matter, as any pins can be I2C
low sentinel
making a basic context manager for async/await experimentation on 6.0 turned out to be straightforward: https://github.com/WarriorOfWire/tasko#you-can-use-a-context-manager-that-wraps-your-spi-bus
I don't have a current circuitpython project that shares a communication bus but this may be handy for those who do (and are interested in concurrency).
idle wharf
@crimson ferry Ding ding !! You win the prize! That was it.
manic glacierBOT
idle wharf
@atomic summit PR above for UMFeatherS2 PreRelease ^^
atomic summit
Ahh, ok, I wont commit my change then! Thanks @idle wharf 🙂
idle wharf
lol 😉
atomic summit
sorry about that goof-up!
idle wharf
Totally not a problem... i learned more about the whole by having to look under the covers
All good
manic glacierBOT
manic glacierBOT
manic glacierBOT
blissful palm
Is there a reason why in CP SPI cannot be constructed with a CS IO and instead relies on doing that Chip Select externally to the SPI?
lone axle
I think its so that the bus can be re-used with a different CS pin to "talk" to a different device.
blissful palm
Thanks @lone axle but surely that would be optional? I.E. in that case you would construct the SPI without the CS
lone axle
Exerpt from this page: https://learn.adafruit.com/circuitpython-basics-i2c-and-spi/spi-devices.
Like I2C multiple devices can share the same SPI bus, however a big difference is that each device typically requires its own unique CS line. Remember the CS/chip select line is what tells a chip that it should listen for SPI traffic. As a result for each SPI device you connect to your board it can share the clock, MOSI, MISO lines but must have its own CS line (typically connected to any free digital I/O pin).
I think theoretically It could be setup to allow you to pass in a CS pin when you initialize it. I think it could lead to confusion that way as well though. Because if you initialize with a CS pin then it would be difficult to initialize the second device I think, since the bus is already in use.
blissful palm
Well if you pass in the CS you would expect it to have control of it, if you didn't you would expect to manage it yourself
Maybe I am missing something python specific here
lone axle
Truthfully I believe that SPIDevice is actually the recomended API for user code to implement: https://learn.adafruit.com/circuitpython-basics-i2c-and-spi/spi-devices#spidevice-library-2837757-21 In that one it does initialize a SPIDevice object with the CS pin
blissful palm
Yes but that still uses CS as a digitalInOut doesn't it?
lone axle
It does
It has to be a DigitalInOut so that it's able to toggle it high and low to indicate when there is data for the device.
blissful palm
Most uCs can do that as part of the SPI device
and avoid the terrible bitbang latency
It is particularly expensive with small SPI transfers
very poor latency
lone axle
DigitalInOut is used only on the CS pin I believe, not the data pins. I have relatively limited experience with it though to be honest. I've hooked up several devices that use SPI with CircuitPython, but not delved any deeper into it.
blissful palm
Yeah I was just testing something and it's behaviour was rather surprising to me so I took a look on the analyser
lone axle
If the hardware SPI lines are unavailable or already in use or something. You can make use of bitbangio module to use arbitrary other pins. But they won't be as fast as the main ones.
stuck elbow
there are lots of hardware spi pins, though
even the smallest samd21 has like 4 pairs
blissful palm
If you are doing large SPI transfers/transactions it's effect is relatively small but when you are performing lots of small ones it consumes the majority of the transfer time, not to mention all the cycles it steals from the CPU to do it.
I know that CP doesn't favour particular chip optimisations by principle but pretty much all the chips can optimise such transactions AFAIK
For example a 1 byte transfer will take about 45uSec and the SPI clocked part will only be about 0.5uSec at 20Mbaud, somewhat disproportional!
stuck elbow
if I only have one device on the bus, I tend to hold the cs at the beginning and never let it go
this way you don't waste any time on wiggling it
tulip sleet
@blissful palm Feel free to open an issue if you would like to propose some improvements to the SPI API, whether in busio or in bus_device
blissful palm
@stuck elbow that's fine for some devices but others look for NE and PE on the CS for transaction purposes
@tulip sleet I'm looking at the CP source right now trying to get a grip on what it does underneath there seems to be quite a bit of exception processing for certain existing Adafruit devices and pinouts etc.. that may make it difficult..
I think bus_device just wraps busio AFAIK
tulip sleet
that is correct: it ensures that transactions are atomic on an SPI or I2C bus by using the locking mechanism.
blissful palm
In my particular use case (a lot of individual register writes) it gets punished severely, but that may not be the case more generally, It might make sense to write a more specific driver for this sort of purpose, I may even go to QSPI in order to reduce then latency further..
It's the transactional latency that's killing me here..
tulip sleet
If you are using busdevice, then there's a lot of latency. You can just avoid busdevice and toggle the CS line yourself, and maybe combine multiple register writes into a single CS transaction.
busdevice is a convenience but there's certainly no requirement to use it
if you are not sharing the SPI data lines with another device, then you of course can do whatever you want
blissful palm
@tulip sleet Yup I am using busio directly, and this device needs the CS transactions (I might be able to change that later..). Is there a way to take an SPI device of the table and use it directly in my own driver?
tulip sleet
your own C driver or Python driver?
blissful palm
I'm afraid it might be claimed when I'm using it!
It would have to be C inorder to automate the CS
tulip sleet
you can allocate it in busio and then just grab the lock and hold it
so it sounds like you want to pass an optional CS pin and have the reads and writes toggle CS as appropriate ... I'm just looking at the SAMD21 data sheet ...
you want hardware-based SS, CTRLB.MSSEN
blissful palm
Yeah that sort of thing, although it's different on other boards/platforms
blissful palm
Most Uc's support it
Many use dedicated pins
I think some can use any pin
In my case I am using ESP32-S2
which can mux it to any pin AFAIK
stuck elbow
any chip with DMA SPI is going to have it
otherwise you can't really do SPI transactions in the background
stuck elbow
it doesn't
blissful palm
presumeably only for things like flash and SPI RAM in CP
stuck elbow
yes, but by the time the code gets back to Python, the transaction is finished and you can safely toggle the pin from Python
tulip sleet
the model CPy uses is the same as Arduino, which does not provide hw support for a CS pin (if I'm reading the Arduino library doc correctly)
blissful palm
I have no idea on Arduino I am afraid as haven't really used it
tulip sleet
I'm just thinking about the historical background for our choice. I don't know whether AVR chips have HW CS support.
stuck elbow
depends on the chip, attiny doesn't for sure
it doesn't even have fully hardware spi
blissful palm
I used AVRs many many moons ago, from memory the tiny devices often had cut down serial hardware units compared to their bigger more endowed brothers..
tulip sleet
it looks like MicroPython does not support hw CS either, either in machine.SPI or pyb.SPI. pyb.SPI takes an NSS pin but does nothing with it.
blissful palm
Yup you are right @tulip sleet http://docs.micropython.org/en/latest/library/machine.SPI.html#machine-spi
even the hardware version
"Each device should have a separate, 4th signal, SS (Slave Select), to select a particular device on a bus with which communication takes place. Management of an SS signal should happen in user code (via machine.Pin class)."
I.E. assumes multiple devices being connected
tulip sleet
if your are going to be around for the 2pm ET weekly CPy audio meeting, we could discuss this in the "in the weeds" section. I think it's plausible to propose an addition to busio.SPI that would take a CS pin and use the HW CS capability of the SERCOM's. We certainly have seen SPI devices that need vigorous toggling of the CS pin to work. It would not necessarily be a high priority, but I think the changes are not large. The main issue is implementing it across all the ports.
blissful palm
It could be tricky to implement in situations where there are few hardware CS/NS pins available on a board (i.e. generally)
@tulip sleet is there a way to claim an SPI device that can be particular to a given board? (I.E take it of the table completely)
(The actual device not at python level)
tulip sleet
you can just not set the flags in the pin table that say those pins can be used for SPI
but it would require a custom build.
blissful palm
WEll no I mean the device itself not the pins (in this case the device can be mapped to any pins)
tulip sleet
the busio.SPI constructor looks for a device that will work on the MOSI/MISO/SCK pins you give. If it can't find one, it will give up
i'm not sure I understand: do you mean an external device or do you mean an on-chip SPI peripheral
blissful palm
on chip peripheral
tulip sleet
why do you want to grab that peripheral?
blissful palm
In the case of the ESP32-S2 there are from memory 3 spi devices that can be muxed to any pins (1 is already taken as are its pins)
tulip sleet
so why do you want to reserve another device? If you just busio.SPI() create it, you will have grabbed one of the ones that will work? If the API were enhanced with hw CS, does it matter which of two you get?
I know ESP32S2 has been tricky to implement for SPI; Scott is the best person to comment on this
blissful palm
Yes it does matter actually in this case
tulip sleet
you mean they are not identical?
spare prawn
when doing a custom build of circuitpython to enable the bluetooth REPL how does one change the default to enable it? it does not seem to be indicated in the doc.
blissful palm
the 3 different SPis have different capabilities
tulip sleet
@spare prawn we are not really supporting the bluetooth REPL at the moment. If it works it's an accident
spare prawn
ah ok, i saw it on the nrf52 build page documentation
tulip sleet
I think that's old, unfortunately. We have not tried to maintain it; I think we got it working as an existence proof, but it could easily have broken since then. THere is also file transfer over BLE, which right now I think is not enabled, or at least not maintained. We have longer-term plans
spare prawn
ok, thank for the clarification
tulip sleet
sorry!
blissful palm
@tulip sleet What I am trying to do will be board specific, I think I will need to do it differently, however there may be a benefit to offering CS SPI control generally, but looking at the CP code it will require a rethink as much of it is based on the assumption of external CS control and multiple devices as standard.
spare prawn
i just built file transfer method over bluetooth...my implementation is a bit complicated needs testing and i was thinking that an alternative would be repl based...might be simpler to conceive with different layering of functions
tulip sleet
@spare prawn you mean like ampy? possibly, but you mean you write file transfer? I thought we had that too. We'd be quite interested in what you're doing. It also requires some app to know how to do it
@blissful palm I will ping @slender iron here to review the above, but I would encourage you to join the audio discussion if possible, or write this up as an issue with your use case
spare prawn
@tulip sleet i am happy to discuss what i have done, its not pretty, i have a pythonista app on iphone as a receiver to upload data to ML pipeline in cloud....i am not sure how to organize nor familiar with the mechanisms for audio...do we need to find a common time to meet?
i also still have a pending open item re synchronous capture of i2c a2d data to disk...we discussed in past and i have not had time to try implementing your (or maybe it wasa @slender iron)'s suggestion to implement as a PR.
blissful palm
@stuck elbow I'm a bit confused about the SPI DMA, does CP intercept the DigitalInOut calls for the CS line?
spare prawn
i think the use case i would describe as a module to faciliate timeseries capture with an all python pipeliine...over phone and ble with using file upload of files post capture.
tulip sleet
@spare prawn we have a weekly audio meeting on Mondays (usually) at 2pm Eastern time, with status reports. At the end of the meeting we devote some time to discussing some issues, in the "in the weeds section" . It's not meant for a really long meeting, but you can get quick feedback or communicate what you're doing. And/or, in the status section, you could describe your project in a few sentences.
blissful palm
@tulip sleet Thank you, not sure I can make the discussion as am on UK time I'm afraid (all depends)
spare prawn
is that happening today?
tulip sleet
So it's today, just join the "CircuitPython" voice channel.
spare prawn
ok, i guess i will sort out how to use discord 🙂
tulip sleet
in the discord settings, there is audio setup, and also some tests so you can make sure your microphone is working
spare prawn
ok thanks, is the circuitpython audio channel associated with this channel?
tulip sleet
click the "pin" icon at the top for more details about the meeting. ... not explicitly, look at the "VOICE" section in the left sidebar. but we do text in this channel at the same time
onyx hinge
That seems like as good a time as any to say <@&356864093652516868> The weekly meeting is coming up at 2PM ET today, which is about 3 hours from now. here's the notes document for the next meeting: https://docs.google.com/document/d/1YnwzXSsa2jI8DdzguNvYpXXwG7-MEip9TyDt3PmvwvM/edit?usp=sharing -- If you'll be speaking, or if you have notes for us to read, please add them to the document beforehand. If you want us to read your notes aloud, put "(text only)" "(missing meeting)" etc next to your notes. Even if you don't have time to write up your notes, please add at least your name so we know to give you a turn during the round robin sections. (and if you plan to talk but can't add your name to the notes doc for whatever reason, let me know and I'll add it for you)
@spare prawn I have added you to the "circuitpythonistas" role, which will allow you to speak in the voice channel and will also send you a small number of notifications each week, mostly about the weekly meeting schedule.
tulip sleet
@onyx hinge could you add @blissful palm also? tnx
onyx hinge
Done!
spare prawn
ok thank you...could you add me to the appropriate place in the meeting timeline? What would be most helpful to discuss?
tulip sleet
edit the google doc above and either add something to the Status Reports section or "In the weeds". There are some entries already so you can get an idea of what kinds of things are list
turbid radish
?serverinfo
digital shoreBOT
adafruit
tulip sleet
you could describe your project in a couple of sentences in the status section, and if you have questions or discussion topics, then put that in the weeds. The Status section does not usually have any discussion: it's just peopel saying what they are doing
spare prawn
@tulip sleet ok, thank you for the pointers!
tulip sleet
FYI, "Hug reports" is a section where people thank others for various things. It comes before "Status Updates".
idle owl
@spare prawn Please add your notes in alphabetical order 🙂
That's how we read them off during the meeting.
blissful palm
BTW @tulip sleet I think for a ESP32-S2 Board specific route it would be relatively easy to claim, I just change the SPI.C common_hal_busio_spi_construct and remove SPI2_HOST branch logic..
(That way it could only offer SPI3 or soft versions..)
It would be nice to be able to do that in the board specific files though rather than the port specific ones..
@stuck elbow Looking at the code I think the DMA transactions are happening inside (and irrespective of ) the external SPI transaction boundaries.
spare prawn
@idle owl thanks for the feedback. OK! I hope I have been a good citizen...I look forward to the meeting! I really appreciate all of this. Thank you!
blissful palm
So from a latency POV it makes little real difference for these sorts of transactions
(DMA)
stuck elbow
as I said, the call waits for the transaction to finish
currently
but if you wanted to take advantage of dma and have python running while things are still being sent, then you would need a hardware CS
onyx hinge
(it arguably has to wait because there's no other general way to preserve the data from modification in circuitpython)
blissful palm
Yeah the real benefit is delivered when transferring a number of bytes that fills the DMA buffer, that's the most efficient usage..
for individual register byte writes there is no advantage
idle owl
@spare prawn By the way, if no one explained it, Hug Reports are a chance to recognise individual folks or the whole community for something positive. It sounds like you already have a hug report for the community. Feel free to add it under "Hug Reports" alphabetically, like the others in the list!
blissful palm
What might be possible is the idea of register burst writes, where registers that need updating could be done autonomously if they reside in successive addresses perhaps..
That would mean a multi-byte transfer and DMA could help there
(like memory burst writes..)
(likewise for reads)
Well I now have plenty to think about..
slender iron
@blissful palm I think the approach I'd take is to reimplement bus device as a native module
then you can bitbang CS from C instead of Python and maybe later add a special case for the CS pin that matches the transaction
wild nebula
Does anyone know whether it is possible to have a SetTimeout or SetInterval-kind of function, the way the Smart/SimpleTimer lib in Arduino would work? I know already about NonBlockingTimer but it's not what I am looking for. Preferably it's a library that makes writing timing functions easy and fun
blissful palm
@slender iron At a board level is there a way of taking an SPI device of the table (Assuming I want to use it permanently) (ESP32-S2 SPI device) or do I have to claim it in Python?
(My native module would need to assume sole use, or claim at run time before anything else)
Also if I went the other route and did it at runtime how would my native module claim from the spi module? Can that be done simply?
(My guess is that you must have something like a common pool of these resources for uCs that have universal devices that handle say I2C and SPI)
wispy badger
Hey guys, I just published a simple light-meter project using CircuitPython: https://www.youtube.com/watch?v=zjhoc0t7I_0 I would love to hear your feedback. Thanks
slender iron
@blissful palm if a board wants something it can set it up in board_init
thanks @wispy badger !
tulip sleet
@blissful palm we are starting an internal meeting before the public one (in one hour)
blissful palm
@slender iron didn't know about that, that could be useful for a bunch of stuff.. 😀
@tulip sleet If I am back by then (I have to go pick up my daughter) I will try to join
tulip sleet
we typically run for 1-1.5 hours
idle owl
@blissful palm Please add your notes in the meeting notes document in alphabetical order. That's how we go through things in the meeting. Thanks! 🙂
@blissful palm Ah I see you stepped away. I'll update it.
In the weeds is whatever order folks add things. Hug Reports and Status Updates are alphabetical.
viral minnow
Can someone help with a quick audio check before the weekly meet? Last time my audio was horrible.
solar whale
sure -- go to chat
viral minnow
No permission to chat it says
lone axle
You will need the circuitpythonistas role
solar whale
I just added it
viral minnow
maybe I need to logout and in again
lone axle
Maybe try refreshing the page if you are in the browser app.
modern wing
Happily lurking, notes doc updated -- hope y'all are having an excellent day.
solar whale
progress -- you are unmuted
low sentinel
Lurking; @ me if you want me, I'm in another meeting!
serene warren
Lurking
distant shadow
Lurking
onyx hinge
We adafruit people are finishing an internal meeting, we'll be with you and get the Discord meeting started shortly! Thanks to everyone who has added their notes already. If you plan to speak, please add your notes if possible and otherwise add your name to the document, https://docs.google.com/document/d/1YnwzXSsa2jI8DdzguNvYpXXwG7-MEip9TyDt3PmvwvM/edit?usp=sharing -- if you haven't spoken before, make sure you ask us to add you to the Circuitpythonistas role.
balmy stirrup
yes
serene warren
yes
trim elm
lurking
onyx hinge
@viral minnow did you get a chance to do the mic check?
viral minnow
@onyx hinge yes thanks! worked well
turbid radish
lurking
spare prawn
sorry for the basic question do i have to do something explicit to join audio?
onyx hinge
@spare prawn on the left, scroll down to "voice"
open the list if necessary,then click circuitpython
(I don't know right from left)
spare prawn
thx
lapis hemlock
I am lurking only.
timber mango
Lurking
thorny jay
Lurking, notes in the stock.
blissful palm
I made it..
serene warren
Got mine on preorder at Cankit 🙂
lone axle
blissful pollen
My credit card was charged so guessing its real soon now!
solar whale
UPS says mine has shipped
onyx hinge
"An exclusive hint - the contents can be used in CircuitPython" shock
thorny jay
Nope for Adabox in Belgium. 😦
onyx hinge
https://github.com/adafruit/circuitpython-weekly-newsletter/tree/gh-pages/_drafts twitter AT anne_engineer
turbid radish
Contact Anne Twitter @anne_engineer, email anneb(at)adafruit(dot)com
onyx hinge
If you want Hacktoberfest credit for a CircuitPython PR, mention it in the issue; we will add the tag that is needed to make the pull request valid for hacktoberfest
ionic elk
Still out this week but wanted to ping @lone axle that I'd love to talk about the Monster M4sk this week
serene warren
Compute module 4 wish list - PCIE to stemma adapter.
idle owl
😊 You're welcome!
onyx hinge
@slender iron what's the plan for branching?
(if it's not quick maybe in the weeds it?)
slender iron
want to discuss more?
stuck elbow
lurking
blissful palm
Alloy Feather (Rev A) for anyone interested in what I'm working on https://github.com/folknology/alloy
lone axle
🎃
slender iron
nice!
thorny jay
RPM calculator are the best... if only I could remember what I was able to do on HP48!
onyx hinge
It was so tempting to add a built in programming language to it, but I was sternly encouraged to wrap up the project. The last time it had almost been finished, I decided to add an arbitrary precision decimal arithmetic library to it 🙂
idle owl
It's always clock issues!
viral minnow
thorny jay
It was so tempting to add a built in programming language to it, but I was sternly encouraged to wrap up the project. The last time it had almost been finished, I decided to add an arbitrary precision decimal arithmetic library to it 🙂
@onyx hinge This week-end, I investigated TI-Python ADAPTER that is an add-on to Ti-83 to permit running Python that was added to the math program in France. They took a Trinked with special firmware. Until I figured that I don't have a Ti-83, so I gave up. But Ti has a few Python hardware that could be CircuitPython. Also, I failed to convince my kid that I will acquire a numworks for them, but I will be the one using it... and for school they need a exam authorised calculator.
@idle owl I wanted to accumulate QT-Py project... copy cat of another such repo: https://github.com/dglaude/my_qtpy_trick
idle owl
@thorny jay It'll be in the Learn system this week or next.
thorny jay
@idle owl and I guess you have seen this one: https://github.com/todbot/qtpy-tricks
idle owl
I have, thank you!
serene warren
Been remodeling kitchen, dining rm and laundry. Can sympathize.
idle owl
@gilded cradle Hurrah!
mental nexus
Had a water heater leak and drain pan leak so had to redo some Sheetrock so took the time to add more shelves at the same time. Unfortunately even a small room takes similar amount of work as a big one!
gilded cradle
Thanks @idle owl
mental nexus
But avoiding going to the laundromat is great motivation to get the work done.
inland tusk
@spare prawn speak up your voice is low
lapis hemlock
I've got to take off now. 👋
serene warren
See yens later
lone axle
Yep, better now
solar whale
better
onyx hinge
Unless your code is written extremely carefully (to avoid any memory allocations on the heap), CircuitPython will have to pause for garbage collection from time to time. It's pretty easy to engineer a scenario where the GC pause is substantially over 1ms.
idle owl
@lone axle For your ItW topic, what would that library involve? (I'm getting ahead of things.) I feel like we'd need further (read: Limor's) input on whether to add it to the main bundle.
lone axle
It's essentially a "Device Helper" library that initializes the built-in hardware. I think of it like the Circuit Playground libary is for those devices, though this one has far less built-in so it's smaller in scope
solar whale
👍
idle owl
Alright, I'll look into it.
thorny jay
No but you can trigger it more frequenty!
thorny jay
Frequently check gc.mem_free() to see what is happening.
gilded cradle
Gonna get going. Thanks everyone. 👋
onyx hinge
blissful pollen
I should have them, I've read them a ton lately
solar whale
~20 sec after reboot it will reboot again
idle owl
@tulip sleet You're not muted
solar whale
I vote for Main Bundle
onyx hinge
@blissful palm we cannot hear you
solar whale
Have to go offline -- Have a great week 👋
blissful palm
That's weird it works on the test
tulip sleet
try switching from the app to the browser or vice versa
onyx hinge
neotrellis uses trellis.pixels[nnn] = CYAN
tulip sleet
do you have manual PTT on?
onyx hinge
@thorny jay are you familiar with the term "duck typing"? It is the term for the way you can create a class which implements the required operations and substitute it for an object of some other type. So you'd look at the operations that you use on pixelbuf objects, and then implement that in a way that talks to trellis or displayio...
lone axle
@idle owl I found "Transfer Ownership" in the repo settings and tried to transfer to "adafruit" but it I got this error: You don’t have the permission to create public repositories on adafruit
idle owl
@lone axle Hmm ok. Looking into it.
thorny jay
neotrellis uses
trellis.pixels[nnn] = CYAN
@onyx hinge I made i work on the Trellis, however I had to reduce the update frequency as I2C is maybe slower than NeoPixel. But I already had the code for that.
stuck elbow
@onyx hinge like this? https://i.pinimg.com/736x/6a/28/35/6a2835e623eb78617c9cea52721e5ae3.jpg
idle owl
@lone axle That's weird because we would have to accept it, so I have no idea why it won't let you....
onyx hinge
idle owl
To transfer a repository that you own to an organization, you must have permission to create a repository in the target organization.
Ok...
Transfer it to me.
tulip sleet
could you transfer it to an individual who works for adafruit, then we could transfer it to the organization?
what kattni said
idle owl
@lone axle Transfer it to me. And then I'll transfer to Adafruit.
low sentinel
onyx hinge
[that's when the magic happens]
idle owl
Transferring now.
lone axle
That transfer seems to have been successful Repository transfer to kattni requested
thorny jay
@thorny jay are you familiar with the term "duck typing"?
@onyx hinge No, looking into that. My idea is a "sandwich" API that hide the difference between the NeoPixel and Trellis and whatever else I create. I just need to figure out how to "replace" the real pixelbuf by a fake copy.
modern wing
Thanks y'all
onyx hinge
Thanks, so happy to meet new folks!
thorny jay
Thank you.
idle owl
@lone axle Please fork and continue working on this: https://github.com/adafruit/Adafruit_CircuitPython_MonsterM4sk
Also see if Actions will work now
viral minnow
Great call, nice to learn from you all! You'll find me around working on LoRa stuff. We have a few surprises.
slender iron
thanks @viral minnow excited to see RAK get involved
mental nexus
Is this the latest instructions for setting up the Saola for CircuitPython? https://gist.github.com/askpatrickw/0a276c7e2d4f54e442b2cb6eaa0d32ea
solar whale
@mental nexus much is now in here https://github.com/adafruit/circuitpython/tree/main/ports/esp32s2
mental nexus
Thanks @solar whale.
slender iron
Here is the notes document for Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/18_xIL_UX_ojVDne1Huz6cv13ZzVIgN2Wap_SgbzU3w4/edit?usp=sharing
spare prawn
thank you everyone, sorry i had to jump off, there was an emergency.
solar whale
Hope it was resolved.
onyx hinge
It's been requested. Here it is with EXTREMELY rough edges: Compile C++ sources into CircuitPython.
https://github.com/adafruit/circuitpython/commit/3b7e40076919a20efb578cb5182d5b3d7330d9d6
Proof of concept: Let files be built as C++
.. this requires changing all the port Makefiles, so it's only been
done for samd so far.Trickery is needed to permit the CMSIS headers be included.
All headers need the MP_BEGIN_DECLS / MP_END_DECLS macro added.
The <stdatomic.h> header is not compatible with the C++ compiler, so
it is hacked out.The file built as C++ is shared-modules/util.cc and it doesn't actually
use any real C++ features such as classes or templates.runtime type inference and exceptions are disabled.
onyx hinge
it's not even in pull-requestable state but if someone's interested in taking it further please be my guest
slender iron
@onyx hinge I think there is an open PR to micropython for C++ too
did you see it?
@gilded cradle are you seeing the A4 in use still?
gilded cradle
I haven't checked since RC0 came out.
slender iron
this person found it on a completely different board
I can't replicate it by simply initing the bus and then reloading
gilded cradle
Yeah, it's still happening @slender iron
I was able to boil it down to a pretty basic bit of code that only uses digitalio and board.
I don't have a CPB on me though.
slender iron
happening on matrix portal? I have a board now. can you send me the repro code?
gilded cradle
@slender iron repro code is in the issue at https://github.com/adafruit/circuitpython/issues/3552
slender iron
👍 thanks
atomic summit
@slender iron who's looking after the esp32s2 UF2 repo?
atomic summit
Ok, all good. I'll bug him in a few days is no feedback.
atomic summit
Any idea why board.c is using a nice neopixel helper, and the bootloader_start is just using bitbanging?
slender iron
no, he'd know though
atomic summit
ok, cool, i'll chat with him about it, cheers 🙂
slender iron
np
rare geode
minor QoL fix for language-circuitpython on Atom, tab completion support added https://github.com/jb3/language-circuitpython/pull/11
lone axle
That is awesome. I didn't realize there even was an atom plugin.
barren peak
Hello all. Do you know where I can find the bootloader .uf2 for the QT-PY? I need to reflash the board as its not cooperating trying to load CP on it. thanks.
I did not see it here BTW https://github.com/adafruit/uf2-samdx1/releases/tag/v3.10.0
slender iron
@wise shard it's not your fault the metro m0 build keeps failing it's just really tight on space
wise shard
Ha yeah I figured.
Thanks for looking into it everyone 😄
If there is anything else I can do I am happy to!
slender iron
we'll see what Jeff says. He just changed a setting to make builds faster at the expense of being larger.
slender iron
🙂 thanks!
manic glacierBOT
idle wharf
I want to mention @atomic summit's favorite dual load of code.py. I've been trying to figure out why I can only sometimes get a wifi connection. I just pulled main and flashed and I got an IP, then it loads again and does not.
tio -b 115200 /dev/tty.usbmodem7CDFA1008C2E1
[tio 01:24:28] tio v1.32
[tio 01:24:28] Press ctrl-t q to quit
[tio 01:24:29] Connected
ip: 192.168.1.29
hostname: UMFeatherS2
Code done running. Waiting for reload.
soft reboot
Auto-reload is on. Simply save files over USB to run them or enter REPL to disable.
code.py output:
Configuring Network and System Time in UTC
ip: None
hostname: None
Traceback (most recent call last):
File "code.py", line 42, in <module>
File "code.py", line 38, in wifi_join
File "code.py", line 22, in set_system_time
File "/lib/adafruit_requests.py", line 555, in get
File "/lib/adafruit_requests.py", line 535, in request
File "/lib/adafruit_requests.py", line 415, in _get_socket
OSError: 0
Press any key to enter the REPL. Use CTRL-D to reload.
And then I end up where CTRL-D or unplugging also doesn't work...
But this all worked a day ago...
Let me know what info I can help pull together to provide better info.
I'll try my Kaluga too
idle wharf
Just seems intermittent, some times it works sometimes not.
solar whale
@idle wharf I have been finding similar issues on my umfeathers2 and saola_wrover ... something has broken. I need to do more testing, but it also keeps causing my Linux box to reboot!! I need to move back to the Mac for testing. Will try to do more later this morning.. . I am also seeing intermittent behavior and once it fails, it does not seem to recover after a soft Reboot.
solar whale
For me -- it works properly after a Power cycle or hard RESET, but not after a soft reboot.
manic glacierBOT
onyx hinge
oh hoooray, pidcodes merged tannewt's big PR
solar whale
@onyx hinge I think I misunderstood -- how do you get the MacOS to compile?
onyx hinge
@solar whale I don't, I have no mac. cwalther's PR fixes the nproc problem, my PR fixes a different problem affecting only the github actions to build the macos mpy-cross program
solar whale
I thought you were suggesting your fix took care of both. Actually I had no issue with mpy-cross
onyx hinge
no, that's not what I intended to say.
solar whale
OK -- I will try #3577
manic glacierBOT
solar whale
@onyx hinge I just trued pr #3577 on my Mac and linux systems -- seems to work OK (for Pyportal build)
onyx hinge
Thanks @solar whale !
manic glacierBOT
slender iron
@onyx hinge I have write access now too.
@onyx hinge or @tulip sleet want to join the zephyr API call?
onyx hinge
No thanks
slender iron
I think they are talking about a pin muxing api today
tulip sleet
@slender iron is it now? I am free
tulip sleet
i'm on, thanks
slender iron
👍
looks like pinmuxing is just listed as "time permitting" in the agenda
which means it probably won't be discussed
@onyx hinge can we switch back to one lto partition? We can parallelize it when we switch to gcc 10
onyx hinge
@slender iron sure, if it's causing problems (will gcc10 do better at partitioning?)
slender iron
I suspect the space issue with the make_transparent PR is due to enabling the single partition
gcc 10 has better code size
onyx hinge
did it still not fit after the translation change I made? I didn't check back
slender iron
no, german isn't fitting
onyx hinge
guess not, okay
slender iron
96 bytes
onyx hinge
I only paid attention to zh_Latn_pinyin, oops.
you wanna push the reversion of the parallel lto to that branch and we'll see?
slender iron
ya, I can
onyx hinge
the mpy-cross-mac failure is addressed in #3578 ..
low sentinel
Whoa y'all are talking with Zephyr people? I just started looking at platformio Zephyr support for another project.
low sentinel
well shoot
I'm planning to learn how to write an application using Zephyr instead of arduino. I'll try to avoid asking you things :-p
slender iron
I've never actually used it though 🙂
low sentinel
Yeah the whole device tree looks a little like it will take some gymnastics to get circuitpython supported but I'm very far from grokking it
slender iron
ya, I don't think dts is useful to us
meager fog
@ionic elk hihi im doing more ESP32S2 testing, i have some analog-in questions
(PWM out works great btw)
slender iron
@tulip sleet did we just blow their mind? 😄
tulip sleet
in some way, yes, the idea of dynamic mapping is way off their radar. They think everything is a static predefined application, spec'd in advance
i'm not sure we are going to convince them this is a use case they want to support
slender iron
yup, agreed
low sentinel
Many static predefined programs do dynamic work, how is circuitpython any different 🚎
slender iron
I'm not sure they should actually do it
tulip sleet
@slender iron you mean it's out of scope for zephyer?
slender iron
ya, I think that's a reasonable conclusion
tulip sleet
i could imagine a restricted version where you can only use some predefined sets of pins (I2C always on board.SCL and .SDA, etc.)
slender iron
I wouldn't want to move us to that though
and if we don't move, then why should they do it
tulip sleet
I just mean that you would still do busio.I2C(board.SCL, board.SDA), but it wouldn't work on any other pins. It's not that interesting, it's only
for a case where you want to piggyback on an existing zephyr port
slender iron
right
maureen would have an idea of how well that'd work. she's done the recent mp zephyr work
pastel panther
@slender iron @tulip sleet was the zephyr API meeting recorded?
slender iron
@pastel panther I don't think so. we only talked the last five minutes about circuitpython basics
pastel panther
ah, ok
slender iron
there is a channel for pin control discussion on the zephyr slack
mental nexus
Hello World!
Yay! I just successfully loaded CircuitPython onto the ESP32-S2 Saola. Here are a few more details that slightly differ compared to the instructions below. Note: I am using an iMac to build and load CircuitPython.
- When running the CircuitPython make command, you have to have the USB cable connected to the physically built-in USB port.
- This worked when I used the serial port labeled
/dev/cu.SLAB_USBtoUARTIt gave an serial error when I used thetty.XXXXport. - My make command was
make BOARD=espressif_saola_1_wrover flash PORT=/dev/cu.SLAB_USBtoUART - After it says that it has loaded, switch the USB cable to the externally wired cable (pin20 to D+, pin19 to D-). Look for the CIRCUITPY drive.
- Celebrate!
Getting started guide is here for the ESP32-S2 and CircuitPython: https://github.com/adafruit/circuitpython/tree/main/ports/esp32s2
idle wharf
I linked to the official page ^^^ from my gist so if anyone ends up there they will go to the new awesome readme for the esp32s2 port
manic glacierBOT
timber mango
Reading up on moving from "Arduino" to CircuitPython, how does the board know to run the Arduino code, or the Python code? And/or am I missing some fundamental concept?
ornate breach
When you want to program with Arduino, just double press the reset button to get the boot to appear. You just select the appropriate board in Arduino IDE, make sure the port is selected and upload
When you’re done in Arduino code, double tap the reset, drag the Circuitpython .uf2 to the boot and it should restart with Circuitpython boot loader running.
tulip sleet
@timber mango The CircuitPython firmware is "just another program" that runs on the board. It's a really big program that does a lot, but it's a program just like Blink is a program
so once you load the CPy UF2 (program), it's all set to store files, run them, etc.
timber mango
Gotcha, thanks! Will it wipe my Arduino installed code?
timber mango
Cool. WIll be sure to keep them both updated in my repos.
manic glacierBOT
manic glacierBOT
solar whale
@idle owl FYI -- this impacts the CPB guide -- the pinouts for SPI are incorrect https://github.com/adafruit/circuitpython/issues/3582
idle owl
@solar whale How does it impact the guide?
solar whale
the pinouts are wrong for the SPI pins SCK is on A1, MISO on A2, MOSI on A3
sorry -- the GIZMO guide is the one that is incorrect
CPB guide does not mention SPI...
idle owl
Do you have Learn access?
solar whale
yes
idle owl
Would you be willing to update the Gizmo guide if I add you to it?
solar whale
sure -- be happy to
idle owl
Excellent. Can you link me to it? Not certain which one it is
solar whale
this came up in a forum post https://forums.adafruit.com/viewtopic.php?f=58&t=170634
idle owl
Excellent, adding you now.
@solar whale You should have access to edit the guide now.
Thanks for updating it!
idle owl
@slender iron MacOS is the only OS that does the stupid file indexing, right? Nothing else does?
manic glacierBOT
slender iron
@idle owl mac is worst afaik
idle owl
Ok. Bleh. Thanks.
slender iron
yuuuup
manic glacierBOT
solar whale
@idle owl I was wrong about the GIZMO guide -- the GISMO does some funky things with the pins..... I will restore the guide and maybe ad a note.
idle owl
Alright, sounds good. Thanks for paying attention to the details!
solar whale
@idle owl I added a "warning" box -- Is it ok? https://learn.adafruit.com/adafruit-tft-gizmo/pinouts#spi-display-pins-3041879-5
idle owl
@solar whale That's great, thank you!
solar whale
No problem -- glad to help.
manic glacierBOT
@tannewt I have NO idea why all these commits are listed in this PR. Do you?
onyx hinge
@solar whale using "git rebase" is always somewhere I have trouble helping someone through the steps.
Honestly if it's an "easy change" then I would just make it freshly again.
solar whale
should I just close the PR and atart over.
onyx hinge
Yeah more or less.
Are you aware of how to start your branch from a different branch, for try #2 ?
solar whale
not really.
onyx hinge
Alternatively, you can simply run git checkout ownerid/main or git checkout ownerid/master (where ownerid is the name you assigned the original project's remote repo)
I'd go sort of based off of these instructions
so you'd git checkout ownerid/6.0.x and git checkout -b cpb-spi-fix-60x or whatever you want to call your new branch
later, when creating the PR, you'll have to take care to choose 6.0.x as the branch to merge TO
solar whale
OK -- sounds good --- I'll give it a shot --
onyx hinge
there's a 1 step way to do that, git checkout -b cpb-spi-fix-60x ownerid/6.0.x^{} but now we're getting into weird punctuation and that's not a good sign
solar whale
I think I'm on my way -- Thanks!
manic glacierBOT
solar whale
@onyx hinge Thanks -- that was easy.
onyx hinge
@solar whale you're welcome
low sentinel
oh hey someone's using vectorio in the wild. That's neat
lone axle
I've got displayio progress_bar and a few others partially working that I converted to use vectorio. Once we have a stable release with it in I'll start PRing more of those.
We need to figure out some kind of fallback for Blinka_Displayio as well I think before including them.
low sentinel
What have you got in mind for blinka_displayio?
lone axle
I think the lowest level kind of fallback is to catch import error and switch to bitmap or whatever other strategy is currently in use.
But I'd love to see vectorio supported more directly by blinka_displayio too
How much effort you think would involved in having a CPython implemenation for it. I have no real idea what would go into it.
low sentinel
Writing vectorio in python? Probably pretty easy. The algorithmic stuff is straightforward to copy and I imagine if displayio is already there it's not too much of a stretch.
If you weren't too worried about ram (as I was when making vectorio) you could just vector draw into displayio bitmaps.
Probably some minor gymnastics to make the apis look the same as circuitpython. I dunno, if you're good at python and can read C it's a weekend project to get it mostly done?
My estimations are usually +-factor of 3 though :-p
lone axle
Nice, I really appreciate your perspective. That sounds potentially within reach for me.
displayio groundwork is there already. If I understand correctly it builds the screen buffer into a PIL Image object and then feeds data from there to the screen.
low sentinel
Oh! Neat, yeah it should be pretty easy then. Vectorio internally has that Shape class that asks its associated.. shape implementation what the color of each pixel is one by one.
low sentinel
If you end up taking this on I'd be happy to chat. The layer._fill_area method is probably the most interesting part.
lone axle
Thank you 🙂
manic glacierBOT
blissful palm
@slender iron you have talked about re-using the M.2 connectors before, I see Sparkfun has now implement such a reuse what do you think of what they have done, will you guys support this 'Micromod' standard especially with many of their cores capable of supporting CP?
prime cove
So I've run into an issue with a PCF8563 RTC that appears to be linked to how the BCDDateTimeRegister handles data. Since the bits aren't getting masked off, I'm getting values returned that are larger than expected e.g. weekdays greater than 7. Would it make sense to add a masking list to the BCDDateTimeRegister class, or is it expected quirks like this should be handled in the chip specific library?
prime cove
def __init__(self, register_address, weekday_first=True, weekday_start=1):
self.buffer = bytearray(8)
self.buffer[0] = register_address
if weekday_first:
self.weekday_offset = 0
else:
self.weekday_offset = 1
self.weekday_start = weekday_start
self.mask_datetime = [0xFF, 0x7F, 0x7F, 0x3F, 0x3F, 0x07, 0x1F, 0xFF]
def __get__(self, obj, objtype=None):
# Read and return the date and time.
with obj.i2c_device as i2c:
i2c.write_then_readinto(self.buffer, self.buffer, out_end=1, in_start=1)
return time.struct_time(
(
_bcd2bin(self.buffer[7] & self.mask_datetime[7]) + 2000,
_bcd2bin(self.buffer[6] & self.mask_datetime[6]),
_bcd2bin(self.buffer[5 - self.weekday_offset] & self.mask_datetime[5 - self.weekday_offset]),
_bcd2bin(self.buffer[3] & self.mask_datetime[3]),
_bcd2bin(self.buffer[2] & self.mask_datetime[2]),
_bcd2bin(self.buffer[1] & self.mask_datetime[1]),
_bcd2bin(self.buffer[4 + self.weekday_offset] & self.mask_datetime[4 + self.weekday_offset] - self.weekday_start),
-1,
-1,
)
)
This is what I have for the BCDDateTimeRegister to correct my problem.
onyx hinge
@prime cove If this affects https://www.adafruit.com/product/3295 then the fix should go in the adafruit library. If it affects another RTC that is "similar to" PCF8523 then maybe not. Reading the PCF8523 datasheet, it is specifically the "seconds" byte that can have a 1 in its MSB. Other unused bits are documented to read as zeros.
I don't think I have a PCF8523 for testing
i.e., maybe only _bcd2bin(self.buffer[1] & 0x7f) is necessary for PCF8523. That MSB of the seconds is 1 when the oscillator has stopped or when the device has lost power, until the time is set by the microcontroller.
prime cove
Everything was good on the PCF8523, the PCF8563 is fairly similar though and had the issue. The datasheet has the bits as don't cares, I figured that would mean zero but in this case sniffing the bus that was not the case and they change frequently above the mask.
onyx hinge
PCF8523 datasheet states "Bit positions labeled as - are not implemented and will return a 0 when read". PCF8563 datasheet states "Bit positions labelled as x are not relevant" and does not guarantee the value which is read.
I think the mask should be 5, even if self.weekday_offset is some nonzero number _bcd2bin(self.buffer[5 - self.weekday_offset] & self.mask_datetime[5 - self.weekday_offset]),
i.e., you want the mask for weekdays of 0x7 applied to the weekday, no matter which register position holds the weekday
does that make sense?
prime cove
Ah you're right my mistake there
got a little too excited with the copy-pasting haha
So is this something that you guys would be open to for a PR? I figure this is something that could reasonably happen with other RTC chips
prime cove
Yes it is. I would just add the mask_datetime member.
onyx hinge
is a similar change needed to i2c_bcd_alarm?
thanks for taking the time to help me understand, I overlooked at first that you said 8563
manic glacierBOT