the python file in tools can help
#circuitpython-dev
slender iron
ionic elk
@slender iron got a sec today to talk sockets?
slender iron
later yes. not now
ionic elk
๐
manic glacierBOT
mild fog
CircuitPython for prototyping has been off the charts fantastic. I really cant offer enough gratitude for how great it's been to work on adafruit feather hardware and to use circuitPython to bring it all together. I'm curious though, if I were able to take my prototype that consists of, say a feather and a few feather wings, but I wanted to reduce costs and make a specific PCB with just the components I need, could I still use circuitPython provided that the components chosen were the same as the components needed from the feather boards? Are there any difficult problems with making the jump to more custom boards while trying to stay in the CP world?
(other than the problem of designing and sourcing said board)
fathom shell
CircuitPython for prototyping has been off the charts fantastic. I really cant ...
If you want to use a SAMD51 and simplify the process a lot, check this out. When this comes back in stock, it's a pricey way of skipping a lot of steps https://www.adafruit.com/product/4802
mild fog
The project is based off of the nRF52840
fathom shell
Ah gotcha
slender iron
@mild fog you can definitely run CP on a custom board
@ionic elk I have a little time now
mild fog
<@!509938944373293056> you can definitely run CP on a custom board
Awesome! Have you ever seen any info or blogs on this process?
onyx hinge
@mild fog https://learn.adafruit.com/how-to-add-a-new-board-to-circuitpython/overview https://www.hackster.io/wallarug/circuitpython-creating-custom-boards-8e979e may help
orchid basin
analog bridge
@slender iron I have updated the ota PR.
manic glacierBOT
inland tusk
@slender iron aew you deep diving today
crimson ferry
Is anyone else still seeing issues with I2C on ESP32-S2? I get intermittent forever-waits (can ^C) on i2c = board.I2C() after soft reboot using 6.1.0-beta.2-89-gf2204d7d8 on 2020-12-16
vivid temple
yeah but in some of the videos lady ada was showing a prototype so i wonder if it will still be released or not?
crimson ferry
There have been some other very cool prototypes shown #TopSecret but "don't ask", and who knows if or when they'll see the light of day
manic glacierBOT
analog bridge
@tulip sleet Is there an easy way to disable usb check... (i.e. force deep sleep).
when debugging...
hearty tapir
@analog bridge I think it was mentioned that the deep sleep is a "fake deep sleep" while you have the REPL active. I believe you'll have to do something through the battery pins to power the board. Scott yesterday showed during the Deep Dive stream how to monitor the power consumption etc, you may find it handy.
analog bridge
hi @hearty tapir yes... I know about the fake deep sleep... I am looking for a way to force deep sleep even when on usb.
currently working on TouchAlarm implementation...
I am able to get it working if I explicitly call esp_deep_sleep_start() just after I set the alarm... but it doesn't work with existing sleep workflow...
tulip sleet
<@!329766224093249548> Is there an easy way to disable usb check... (i.e. force ...
I had added such a provision to the original API, but it was too easy to get in a unbreakable boot loop in which USB went away before a looping sleep could get interrupted. This happened the very first time I started debugging sleep, and I had to erase flash completely to get out of the loop.
Are you trying to power via the USB port but have no data? I tested some of the time with a power-only USB cable. I have a switchable one and also one where I have just cut the data leads. Otherwise I also used ESP_LOGI() a lot. Also you could have CircuitPython write to a UART instead of to stdout if you are trying to instrument the CPy program.
If you just want to hack it temporarily for debugging, I think it would be sufficient to change this to if(true), in main.c:
if (!supervisor_workflow_active()) {
// Enter true deep sleep. When we wake up we'll be back at the
// top of main(), not in this loop.
alarm_enter_deep_sleep();
// Does not return.
thanks @tulip sleet...
one more question, I notice neopixel and status leds get disabled after entering deep sleep and are only re-enabled on a power cycle... is this expected behavior?
tulip sleet
yes, because the GPIO state is reset. We'd want to turn these off anyway even if it wasn't automatic to save the current consumption of those LED's. The power-on LED (not on GPIO) on the MAGTAG has a high-value resistor to minimize its consumption as well, and it's pretty dim as a result.
I think in the case of the status NeoPixel that is turned off deliberately, since just disabling the pins that feed it would not turn it off
onyx hinge
I think there's a problem with adafruit_bus_device-in-core and bitbangio.I2C. When I try to use this combo on esp32s2, I get a hang; if I hit ctrl-c it becomes a panic error. ```E (44776) i2c: i2c_master_cmd_begin(1113): i2c driver not installed
../../esp-idf/components/freertos/queue.c:719 (xQueueGenericSend)- assert failed!
I suspect that the in-core version of adafruit_bus_device assumes the device is a native busio.I2C device but it is not necessarily
tulip sleet
the four front-panel neopixels have a transistor for power control
analog bridge
is gpio reset called before entering deep sleep?
onyx hinge
.. unless anyone has quick words I'm going to file an issue about it ..
tulip sleet
i think so, but I'm not sure where.
analog bridge
I am asking this because touch pad needs to be initialized before entering deep sleep... if io is reset before entering deep sleep, it will de-initialize touch pad
tulip sleet
it may just be a side effect of esp_deep_sleep()
i don't see a call to our own reset routines.
onyx hinge
True
[11, 112]
>>> bus = bitbangio.I2C(board.IO10, board.IO11); bus.try_lock(); bus.scan(); bus.deinit()
True
[]
``` though .. bypassing `adafruit_bus_device` bitbangio.I2C sees only an empty bus ๐ฆthere is a call to board_deinit(), before both fake and real deep sleep, but that routine is empty on most boards. On the magtag it deinits the eink display
analog bridge
if I explicitly call esp_deep_sleep_start() just after setting the alarm... it works
tulip sleet
quoting from the esp doc:
In deep sleep mode, CPUs, most of the RAM, and all the digital peripherals which are clocked from APB_CLK are powered off. The only parts of the chip which can still be powered on are: RTC controller, RTC peripherals (including ULP coprocessor), and RTC memories (slow and fast).
One thing we may want to handle:
Some ESP32-S2 IOs have internal pullups or pulldowns, which are enabled by default. If an external circuit drives this pin in deep sleep mode, current consumption may increase due to current flowing through these pullups and pulldowns.To isolate a pin, preventing extra current draw, call rtc_gpio_isolate() function.
For example, on ESP32-WROVER module, GPIO12 is pulled up externally. GPIO12 also has an internal pulldown in the ESP32 chip. This means that in deep sleep, some current will flow through these external and internal resistors, increasing deep sleep current above the minimal possible value. Add the following code before esp_deep_sleep_start() to remove this extra current:
rtc_gpio_isolate(GPIO_NUM_12);
analog bridge
for touch_alarm to work esp_sleep_pd_config(ESP_PD_DOMAIN_RTC_PERIPH, ESP_PD_OPTION_ON); needs to be called
tulip sleet
so you are calling esp_sleep_enable_touchpad_wakeup(void) ?
analog bridge
my touch alarm branch https://github.com/adafruit/circuitpython/compare/main...microDev1:touch-s2
tulip sleet
you may want to take a look at Scott's PR for pin alarms, which is not merged yet. It mostly works but has some issues, but I think pin alarms and touch alarms are going to be very similar
the esp_sleep_pd_config() may happen automatically, similar to using rtc fast mem. It may record that you need that when you do esp_sleep_enable_touchpad_wakeup()
analog bridge
I tested this but not doing automatically in case of touch
analog bridge
I did some more tests now with forced deep-sleep enabled...
there's definitely something between setting up touch-pads and starting deep sleep which is causing touch-pad to de-initialize
tulip sleet
the sleep doc is not very clear about how to set up the touch pads. is there an example of deep sleep touch you are using as reference? i thought it would be quite different than regular touch, considering how the pin level checking is also different (only certain pins, etc.)
analog bridge
the basic touchpad initialization is same for normal as well as sleep... some extra parameters need to be dialed in for sleep
there is more info about that on the touchpad page instead of sleep_modes page
blissful pollen
@onyx hinge So looks like its an bitbangio.I2C issue not the bus device? I can look at the bus device if you think that's the problem, but may take me a couple days
onyx hinge
I think there are multiple problems going on. I would appreciate if you can take a look at bitbangio.I2C + built-in busdevice, that would be helpful
you can probably do it on whichever micro is convenient for you
blissful pollen
Sure never played with the bitbangio before, so totally possibly I didn't account for something
tulip sleet
@analog bridge ah, I see where the reset is is happening, I think, it's in cleanup_after_vm() in main.c, which calls many reset routines. That gets called when code.py finishes running.
Scott's pin alarm code may defer setting up the alarms until after the program has stopped running, to get around this, or else he prevents reset on certain pains, but I think the former is more likely.
so I'd take the lead from his code
thorny jay
The 32blit kickstarter (from Pimoroni) is reaching the backer. Any CircuitPythonista have one? That is an STM32 board possibly supported by CP already.
analog bridge
<@!747673663419187281> ah, I see where the reset is is happening, I think, it's ...
nice... I assume this calls reset_port()?
tulip sleet
Scott's code calls alarm_pin_pin_alarm_prepare_for_deep_sleep() just before deep-sleeping. See line 157 in https://github.com/adafruit/circuitpython/pull/3830/files#diff-358598ea16b4ad1f25f165d7280842af0eb90c70c668c6b34f343ef39b2b09c5. That just sets up some pullups. Still looking...
yes, it does call reset_port()
he calls never_reset_pin_number() on the pins used for alarming here https://github.com/adafruit/circuitpython/pull/3830/files#diff-139b97e7159e35d9fe550989a19ac82e596cf83dfb92f88e250ff129c21101ec line 285
so that would prevent messing up those pins before going into deep sleep (my quick surmise)
analog bridge
๐ @tulip sleet thanks again for bug-fixing
tulip sleet
what bug did i fix? ๐
analog bridge
ok... for clearing things up
tulip sleet
ah ๐
tropic rain
What is the proper way to report a bug? Here or github?
tulip sleet
github is best; if you're not sure you can ask here first, but github is a much better to-do list
thanks!
manic glacierBOT
Here's what an attempt to read cell voltage looks like on esp32s2 near the tip of main (f2204d7d88b505155cd25b38b71ae3c35b86d438):
tulip sleet
@tropic rain which board are you using that doesn't have json?
and what version of CPy/
tropic rain
Circuit Playground Express. I'm trying to show images on the TFT Gizmo. boot.out.txt is
Adafruit CircuitPython 6.0.0 on 2020-11-16; Adafruit CircuitPlayground Express with displayio with samd21g18
and I'm using libraries fro adafruit-circuitpython-bundle-6.x-mpy-20201216
tulip sleet
yes, that board doesn't have json; there's no room. the code should not assume json. Roll back to version 1.4.0 for now, and I'll revise your t bug: https://github.com/adafruit/Adafruit_CircuitPython_Slideshow/releases/tag/1.4.0
tropic rain
Thanks, I'll give it a go.
idle wharf
I worked on a fix for Circup so it would automatically update its version from the release tags. Can someone with Library experience have a look at my changes. I don't know how to test this locally. The CI runs clean and I can install my branch and it works fine, but can I test the semver part without merging and attempting a release?
tropic rain
yes, that board doesn't have json; there's no room. the code should not assume j...
Thanks , that worked and I gotta say I'm really impressed with your quick response. This is how it should be.
timber mango
Hey guys im trying to get a python device that runs a simple script and maybe has like a screen on there so I can display messages to show when something finished. Can anyone refer me to a not-sold-out device?
idle wharf
Hey guys im trying to get a python device that runs a simple script and maybe ha...
That last link was bad advice...
How about Feather S2 https://www.adafruit.com/product/4769 and one of several Feather Displays. https://www.adafruit.com/?q=feather+display&sort=BestMatch
"at" me in Help-with-projects if you need more ideas.
fathom shell
That last link was bad advice...
How about Feather S2 https://www.adafruit.com/...
Why is the CLUE bad advice? Curious
idle wharf
You know what... I was wrong about that. I thought it was Arduino ... NVM, Clue was a great idea. https://www.adafruit.com/product/4500 Thanks @fathom shell
timber mango
I already went with py portal
fathom shell
Also thoughts on a pyportal? I think it's overkill now but I suggested it because it has no soldering involved and has a big screen
ah right wifi was a req
timber mango
So with py portal I can just press a button and it works?
fathom shell
how do you mean?
timber mango
like is there a button on it that runs the script
fathom shell
So with circuitpython your code.py or main.py script runs automatically on bootup until it ends
fathom shell
*after boot not on boot
timber mango
wait so I just plug it in, edit the file, and boot it up?
fathom shell
When it powers on it boots up and then runs the code in code.py and uses any libs referenced therein. You can load your code and then restart and it will automatically run.
timber mango
oh cool so I can use it as long as it has a wifi connection?
fathom shell
You should be able to use it without wifi
fathom shell
You do for power
strange pike
oh cool so I can use it as long as it has a wifi connection?
re: usb, wifi, etc ... you should get specific about what you're looking for. most circuitpython devices can be powered by external battery or usb, but many do not include integrated wifi.
timber mango
So can I just get a portable charger for it @fathom shell
fathom shell
you're gonna want to be careful about input voltage, I'd see if you can find a 5V only charger to be safe
tulip sleet
@tough flax asks:
I know that CP's USB stack uses TinyUSB and that TinyUSB supports WebUSB.... but do know if CP can benefit from that code? Can WebUSB talk to the serial port for example?
It can, though I don't know much about it. Here's a sample serial terminal: https://googlechromelabs.github.io/serial-terminal/ (https://github.com/GoogleChromeLabs/serial-terminal). I get a lot of <CONNECT><DISCONNECT>, but I find that if I turn on "Flush on enter", that problem goes away for a while, at the expense of only echoing on "Enter". I'm assuming this can be fixed. We are already using WebUSB as an alternate upload tools for the ESP32-S2 boards: https://learn.adafruit.com/adafruit-metro-esp32-s2/web-serial-esptool. Repo is at https://learn.adafruit.com/adafruit-metro-esp32-s2/web-serial-esptool
@gilded cradle did the upload tool.
analog bridge
@tulip sleet alarm.wake_alarm is taking weird values... It takes the value of previously set object. Example in the following case it takes value of <DigitalInOut>
led = digitalio.DigitalInOut(LED)
led.direction = digitalio.Direction.OUTPUT
led.value = True
print("WakeAlarm :-", alarm.wake_alarm)
that is some kind of bug then. It's a static entry in the module dictionary for alarm
i will look at that; i added it recently, in a bit of a hurry to make a demo work
it was there but unimplemented before that
analog bridge
above is with code in PR #3850.
in my local branch... I implemented a way to get alarm.wake_alarm.pin for touch_alarm and now I get the same crash as the one scott is getting with pin_alarm
tulip sleet
are you using alarm_set_wake_alarm() and/or alarm_save_wakeup_alarm() in shared-bindings/alarm/__init__.c?
analog bridge
I didn't alter that... yes alarm.wake_alarm is using the existing implementation.
tulip sleet
I will take a look, something sounds definitely odd.
analog bridge
the weird thing is that it works when I do ctrl+c just before print(wake_alarm) in main code and instead use repl to get wake_alarm
tulip sleet
ok, I was going to test it in the repl first! ๐
we have to delay setting alarm.wake_alarm until the VM and heap are set up when the program restarts
since it creates an object
analog bridge
yes I noticed that while digging into main.c
kind river
I got a few feather M4 CANs and started working on a canio based node framework that allows assigning callback functions to topic IDs: https://gist.github.com/ryang14/8d09939b9aa15716d4fbeca4fe982905
gloomy shuttle
<@!329766224093249548> `alarm.wake_alarm` is taking weird values... It takes the...
This looks and sounds similar to the bug that @slender iron was working on in his deep dive on Thursday in PR #3830
lone axle
It seems the "creating a library" page in the Creating and Sharing a Circuitpython library guide is having a 500 error: https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/creating-a-library
The rest of the pages in the guide seem to be working fine: https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library
tulip sleet
@tough flax I was told we got WebSerial working, but not WebUSB. But I think the demo I posted is WebSerial.
thorny jay
Thanks to @tulip sleet sleep memory, I made a kind of booloader for the MagTag. You can choose from 4 usage, hold the corresponding button, turn on the MagTag... and it will start that program at every startup (even after sleep): https://twitter.com/DavidGlaude/status/1340365817138044933?s=20
tough flax
@tulip sleet I get that WebSerial is easier, but it's still experimental on Chrome (and unavailable elsewhere) so that's kind of a problem... but I will take a look
lone axle
@thorny jay That is awesome!
thorny jay
Great that you like it.
I wonder if I can do "import table[2]" to put the boot config into a file. Also have another meaning for 0 because that is uninitialised memory value, so a power on without pressing a button.
languid sage
HELP!!! is anyone, besides me, trying to install the weather forecast display on the MagTag??? I'm having nothing but grief. I get a note about adafruit_imagload (from line 4 of the code) but adding that library to the lib folder results in other errors. and... OBTW the doc does mention this (not with the list of other libraries, but on a different page), but when I install it from the latest bundle (20201219) all I get is more errors (from line 176). Does this stuff get a QC check???
thorny jay
Is there a CircuitPython way to import a file given it's name as a string? (like this question for CPython): https://stackoverflow.com/questions/301134/how-to-import-a-module-given-its-name-as-string)
onyx hinge
__import__ is available, the other solutions likely are not.
tulip sleet
HELP!!! is anyone, besides me, trying to install the weather forecast display o...
Could you start a thread on the forums? Include the exxact errors you are getting and the versions you are using, including the version of CircuitPython you are using. Thanks.
or try #help-with-circuitpython, with a few more details like the above
manic glacierBOT
thorny jay
I am trying to "rate" the MagTag guide based on criteria. The goal is to identify those that could get some love. This is all related to beeing able to put more project on a single MagTag and have a selection button to switch from one usage to another. Here is a read-only version of the document in sharing. Maybe to be discussed "in the weed" (I made it read/write by all to help reading or contribute): https://docs.google.com/spreadsheets/d/1ZBrLID3p-dKbLkJ6x4Uq77l1FHkGS5ZRIoqBoD2kw_w/edit?usp=sharing
ivory oak
on circuitpython.org what is the difference between CircuitPython 6.1.0-beta.2 and Absolute Newest? Where is the current Stable release?
feral spoke
I'm looking in to some custom built split keyboards and was wondering if anyone knew if I could run CircuitPy on this keyboard with the Elite-C controller?
https://splitkb.com/products/kyria-pcb-kit
jaunty juniper
I'm looking in to some custom built split keyboards and was wondering if anyone ...
it looks like it's running a 32u4 and that's not gonna run Circuitpython
feral spoke
it looks like it's running a 32u4 and that's not gonna run Circuitpython
I'm confused about the comment about the 32u4. Why is that an issue? This appears to be 32u4 and works with circuitpy. https://learn.adafruit.com/adafruit-feather?view=all
jaunty juniper
that's feathers. Not all feathers support CPY, it runs on a few supported 32 bits chips with enough ram and flash
supported boards are listed on circuitpython.org, and any board with a similar chip can probably be added to it (provided the native USB is available on the board)
feral spoke
Maybe I should ask a different way. Can I put a CircuitPy boards on that keyboard and manage it with python?
jaunty juniper
oh probably
solar whale
@jaunty juniper Are you asking if you can enter keystrokes to the REPL from a keyboard attached to a board running CircuitPyhton? I have not seen any examples of that. Has anyone?
jaunty juniper
you meant to tag @feral spoke ๐ it's a good point, I might have misunderstood the question
solar whale
yes, thanks.
feral spoke
@solar whale I just want to use python to customize the firmware of the keyboard vs C and the QMK....but I like the PCB and layout of that board.
solar whale
OK -- I guess I misunderstood. Not something I am familiar with. Sorry I can't be of any help.
thorny jay
`__import__` is available, the other solutions likely are not.
Great! import('magtag_weekly') works for me.
ionic elk
@feral spoke there's a keyboard emulator page for HID devices here https://learn.adafruit.com/circuitpython-essentials/circuitpython-hid-keyboard-and-mouse
Doesn't somebody make circuitpython powered keyboardy stuff? I can't remember who though.
feral spoke
Yes there is a MK60 board out there, but I want to use a diff board, but run python for firmware
idle wharf
on circuitpython.org what is the difference between CircuitPython 6.1.0-beta.2 a...
Let me take a swing and someone official can correct me if Iโm off. 5.x is the Stable release. 6.x Betas are the attempts to stabilize 6 for existing boards (non esp32-s2). When 6.x Beta something shows to be stable it will become a 6.x stable release. Main is the development branch and where esp32-s2 implementation is ongoing.
I hope that helps.
(And that it is correct)
thorny jay
Let me take a swing and someone official can correct me if Iโm off. 5.x is the S...
6.0 is Stable (not 5.x) and 6.1 is currently beta but the only recommended for ESP32S2 such as MagTag.
And I am not official. ๐
thorny jay
on circuitpython.org what is the difference between CircuitPython 6.1.0-beta.2 a...
Ok, CircuitPython 6.1.0-beta.2 is one particular chosen Absolute Newest from a few weeks ago. And by it's name, it is a beta release. But it is also the recommended version for MagTag and other ESP32S2 board. Absolute Newest is build every time one PR is accepted. It's name should contain the GitHub hash for that particular change in the code. In addition, when someone propose a PR, artefacts are created and compiled by the CI (and can be retrieved to test that particular build). Cc: @idle wharf
strange pike
Yes there is a MK60 board out there, but I want to use a diff board, but run pyt...
ooo that's nifty ... @feral spoke the kmk readme is recommending the M4 Express or nRF52840 Express adafruit feathers or similar devices due to RAM concerns. so not all CircuitPython boards will be appropriate. the big questions i'd have is how many pins you need to connect the board to your PCBs, and since you're doing a split kbd, whether you need one board for each half.
feral spoke
ooo that's nifty ... <@253259584867794945> the kmk readme is recommending the M4...
the 4 pads at the bottom are soldiered jumpers for the current build...but it appears to be 24 pin. ( and yes, there needs to be one for each board )
thorny jay
This is not Feather size, maybe ItsyBitsy?
There are people from Adafruit that build their own keyboard. This may make more sense to them, and there might be a channel for those discussion.
strange pike
teensy 4.0 has a circuitpython release, and that's probably closer to the intended form factor
thorny jay
Maybe that is the form factor: https://circuitpython.org/board/nice_nano/ (nice! keyboard: https://nicekeyboards.com/)
strange pike
the nice-nano should work, it's nRF52840-based
thorny jay
Good luck finding one...
feral spoke
hmm..looks interesting...and it's wireless too ๐ What is the chance that @meager fog chimes in on this topic? ๐
hollow kraken
Hi here! I'm cross posting a question here that was originally posted on help-with-projects. I'm working on trying to get a low cost switch interface for assistive tech users to plug their audio jack switches into in order to connect to a computer. I've been successful with getting a bluetooth version working using the Itsy Bitsy running circuitpython but when trying to connect the switch using the QT Py via USB/ HID, in the accessibilities setting (Mac 10.1 5) I get the following error: Unsupported switch input. Try a different switch or change the mode of your switch hardware. Is there something in the library settings related to USB connection I can change to have the computer "see" the switch. The keyboard HID works as expected and sends keystrokes without an issue.
feral spoke
welp... the nice-nano being 3.3v doesn't support OLED and LEDs on the keyboard. Maybe I just have to wait longer until the interwebs has a good split keyboard design that I can run a python based firmware ๐
strange pike
welp... the nice-nano being 3.3v doesn't support OLED and LEDs on the keyboard. ...
that might be work-around-able w/ level shifters ... most (all?) of the circuitpython boards are going to be based on 3.3v mcu's
hearty tapir
I'm unlucky with the MagTag ... when I have time and realise that there was a notification email, it is already gone from the shop again. ๐
I guess that this will not change over the holiday period now, so I'll continue to play with my featherS2 ๐
crimson ferry
I've got a makeshift MagTagAlike with a FeatherS2 and EPD FeatherWing queued up to work on ๐
Kind of how I used Feather M4 + TFT FeatherWing (and still do) before PyPortals were available
manic glacierBOT
hearty tapir
@crimson ferry did you see any Beacon again? ๐
hearty tapir
It is so disappointing if you theoretrically know what you want to achieve but C has a different agenda ๐ฆ
I'm trying to come up with a workaround for https://github.com/espressif/esp-idf/issues/6267 by checking if the struct member is NULL for &self->ap_info.record.country, but instead it is actually "<null>". I put an "else" in there, to disregard this for now and use esp_wifi_get_country(&self->ap_info.record.country) instead, which appears to work fine - but when I later try to access "self->ap_info.record.country" I have a backtrace. I think that my approach to overwrite that portion with valid data is not that valid.
crimson ferry
@hearty tapir I have not seen a Beacon again
hearty tapir
I can at least confirm that none happen during a backtrace either ๐
I saw plenty this night
manic glacierBOT
crimson ferry
FeatherS2 with ESP32S2
6.1.0-beta.2-124-gfd21ff59d on 2020-12-18
import supervisor
time.sleep(5)
print(supervisor.runtime.run_reason)
supervisor.reload()```repeatedly gives```soft reboot
Auto-reload is on. Simply save files over USB to run them or enter REPL to disable.
code.py output:
supervisor.RunReason.AUTO_RELOAD
```but expected `supervisor.RunReason.SUPERVISOR_RELOAD`?Also get AUTO_RELOAD by typing CTRL-D in the REPL (expected REPL_RELOAD)
crimson ferry
Pressing the reset button on the Feather S2 gives microcontroller.ResetReason.POWER_ON supervisor.RunReason.AUTO_RELOAD expected STARTUP?
ivory oak
@thorny jay you may have nailed what you answered, but didn't fully answer my question. To the beta description and absolute newest are one in the same. If the magtag is running beta code, I would say its not a production board. Why would something like this get released to the masses on beta code?
jaunty juniper
ah you are saying that the MagTag page doesn't have a stable release
ivory oak
@jaunty juniper if what @thorny jay said, its running the beta which is not a stable release according to their description and absolute newest also does not sound like the stable release either. So my question is what is the stable release for the Magtag?
thorny jay
Sorry, I don't have to deal with this. I do not work for adafruit. Maybe you should wait for working hours in the US and/or contact the support. This is a community and I am an happy customer running the latest 6.1. on my MagTag. I have happily run many of the guide and made my own project with it. Regards.
minor glade
<@199811382793863168> if what <@681528999520108564> said, its running the beta ...
there is no stable release of CircuitPython for the MagTag yet. but that's fine actually. It's not that the board is in beta, just the CircuitPython distribution for that board
turbid radish
?serverinfo
digital shoreBOT
adafruit
turbid radish
Yes, that's what I'm monitoring. I must be sure we have the milestone graphics in time for the milestone
ornate breach
๐
analog bridge
@tulip sleet regarding wake_alarm... delaying saving it does seem to work... I have modified the code and now I save the wake_alarm manually from code.
import alarm
alarm.save()
print(alarm.wake_alarm)
Is there a callback to know whether or not the heap is configured?
tulip sleet
@analog bridge There is bool gc_alloc_possible(void). And the new obj allocator should return NULL if there's no heap. But you should not need to make this decision at run-time. we should be setting that object when the heap is ready. It seems like we're calling all this too soon. I am planning to take a further look today.
tulip sleet
Hi here! I'm cross posting a question here that was originally posted on help-wi...
I'm not sure what kind of HID device switch devices present themselves as. Could you point to some other device that works on the Mac, so I can look up the technical details? Does this work on Windows (assuming Windows also supports switch devices)?
onyx hinge
<@&356864093652516868> It's about 2.5 hours until the last Circuit Python Weekly of the year -- there's no meeting on the 28th of December. Please add your notes to the document if you can. If you plan to speak, at a minimum we ask that you add your name as a placeholder. If you haven't participated before, ping me to make sure you're a member of the circuitpythonistas role. If you're just listening in (lurking), no need to do anything. Link for the notes doc: https://docs.google.com/document/d/1s3bS3z-5WW_0D5NdlRlKpDB-n_aP6eMbeFvtknM59nA/edit?usp=sharing
turbid radish
I will not be at the meeting, I have an appointment
onyx hinge
Take care!
hollow kraken
I'm not sure what kind of HID device switch devices present themselves as. Could...
Thanks so much for getting back to me. This is an example of a commercial switch interface: https://www.rehabmart.com/product/atec-usb-computer-switch-interface-38517.html. I don't really know how it presents itself but I have been successful with connecting switches just using the standard Arduino library although I can't remember what board I was using. I'm going to try and load the CP 5 firmware when back from work and see if that might resolve things? Tried the Trinket (not MO) and it's HID/ keyboard library. Will report back and see if that works.
ionic elk
@prime flower I'm looking at your recv_into issue since I'm trying to more generally line up the error behavior of socket to Cpython. But I'm trying to figure out how Cpython actually returns the timeout error - from the Cpython Module/socketmodule.c file, it just returns Null https://github.com/python/cpython/blob/03c8ddd9e94c7bddf1f06cf785027b8d2bf00ff0/Modules/socketmodule.c#L3544
This isn't the first time I've seen CPython returning more specific errors in practice than the source code implies it should. Does anyone happen to have insight on why this happens?
prime flower
@ionic elk When I run MiniMQTT on CPython, it returns a socket.timeout which I can Except. It also does not close the socket if nothing was received within the timeout, the esp implementation seems to close the socket which is problematic for MiniMQTT because the client should poll for new data from the broker/server.
The specific error in the issue occurs here (https://github.com/brentru/Adafruit_CircuitPython_MiniMQTT/blob/cpython-s2/adafruit_minimqtt/adafruit_minimqtt.py#L872), where it attempts to read from a socket within a timeout and then throws an OSError and closes the socket (?).
I'm about to eat lunch but can talk more later.
ionic elk
@prime flower sure, as per your post the problem is almost certainly this block
if (received == 0) {
// socket closed
common_hal_socketpool_socket_close(self);
}
but I'm trying to figure out what to replace it with... I've been rooting around the Cpython module trying to figure out how they do things
prime flower
yep - points that's it
ionic elk
I guess that Cpython routes everything through sock_call_ex which I think is done to make things asynchronous. It handles timeouts and sorts null returns into errors, rather than the socket functions themselves doing it.
so I don't know how relevant it is. I could simply put in an OS error in that bloc instead of the socket close, but I don't know if that's the behavior we want?
This isn't the only case where I'd like to bring our socket module in line with Cpython so I'm trying to wrap my head around how they do things
prime flower
Yeah, understood.
Could we instead raise a timeout error if nothing has been received by the server during the time the client attempted to read?
Also - blocking/non-blocking functionality might need to implement something similar. I had issues getting non-blocking sockets working on the esp32-s2 implementation (setting settimeout to 0, https://circuitpython.readthedocs.io/en/latest/shared-bindings/socketpool/index.html#socketpool.Socket.settimeout)
ionic elk
I don't think we have any non-blocking capabilities at the moment
I don't know if we'd need to do that through FreeRTOS or what
LWIP does not have obvious looping operations that we could extract out and put our own background tasks calls into
prime flower
for now I can roll without non-blocking sockets, I'll just use a timeout and handle that. ESP32SPI impl. of MiniMQTT uses that and is OK.
tulip sleet
Thanks so much for getting back to me. This is an example of a commercial switch...
Could you open a thread on https://forums.adafruit.com and include the details, including the source code you're using? Then ping me with the forum URL and I'll take a look. It's probably a little bit too involved to do interactively here.
ionic elk
@prime flower for now, so you're unblocked, would you like me to just swap out the socket close for a quick OSERROR(0) so you can keep working?
we could always revise it later
I don't think the socket close should be there in any case, but there could be other reasons I don't know of for why that's in there.
hollow kraken
Could you open a thread on https://forums.adafruit.com and include the details, ...
Will do ๐ TY!
prime flower
@ionic elk Ok, sure. I'll try it out once you've added that.
ionic elk
hmmm. I'd like to fold this into my existing PR, since that needs a lot more error stuff added to it anyway, but I may need to check out other things that are messing up the build
@prime flower do you have a local build setup for Circuitpython? You could try deleting the socket close yourself as a branch.
if not I could put in a new PR for it
prime flower
@ionic elk I dont for ESP-IDF builds
idle owl
@idle wharf You around?
I'm only partially here, in a meeting, but looking into Circup.
idle wharf
I'm around
idle owl
You seemed hesitant with your automation PR. Reason?
I was about to merge it and then read closer.
idle wharf
I can't prove that it works locally
idle owl
Hmm.
idle wharf
But someone who does this all the time might look at it and go... "that's fine."
idle owl
We can always revert it. Maybe it's worth merging to run it.
Ah ok
I see what you meant. I'll look at it closer and then probably merge it and we'll see what happens.
I'm not certain why it wasn't done that way to begin with.
Which is my only hesitation.
idle wharf
It seems like circup isn't plugged into the library process since its not in the bundle so maybe it was more of a one off and got lost ? (Totally guessing and assuming stuff)
idle owl
Right. But automating version numbers seems like a logical desire.
Regardless of ties to lib process.
ยฏ_(ใ)_/ยฏ
idle wharf
Totally agree with you
idle owl
Anyway, my vote is merge and revert if necessary. You in?
idle wharf
I'm in !
idle owl
Excellent. I'll do that and release it and see what happens. And have you test it.
idle wharf
You can't fix things if you don't break them. ๐
idle wharf
I'm doing the day job, but can keep an eye on this too. Its easy enough to test it.
Thanks for following up @idle owl !
idle owl
You're entirely welcome! Thank you for keeping it on our radar.
@idle wharf Green! And PyPI shows it updated!
Test it at your convenience ๐
Make sure it actually runs - in the event that we made it so not everything gets packaged properly.
idle wharf
checking it right now
@idle owl circup list, freeze, show, update all worked. Pip list shows the version correctly. circup --version does not. I'll open a bug for that and will fix it tonight.
tidal kiln
@idle owl not fully. but what's up?
idle owl
There's a screenshot in your Tides Viewer guide that's not accurate anymore. There's no easy way for me to update it except to redo it completely. https://learn.adafruit.com/admin/guides/3194/editor/20984
@tidal kiln If you have time today to update it, or potentially reword the guide to not require the screenshot, that would be great. If not, I'll add it to a todo in Basecamp for when you return.
idle owl
@tidal kiln Oh I didn't even notice you hadn't mirrored pages. Yes, entirely.
tidal kiln
thank for pointing that out. i'll take care of it.
idle owl
I'm going to add a warning that you should leave on that page. Otherwise, it's all you. Thank you!
idle owl
Thanks for letting us know
tidal kiln
have a good one ๐
idle owl
You too!
modern wing
Good afternoon all you wonderful folks -- happily lurking, last notes doc of the year updated.
idle owl
Thank you for updating the doc!
modern wing
14" in Northeastern PA, 5" in Bergen County NJ
idle owl
๐ That's teamwork.
@inland tusk Are you participating today?
@strange pike Are you listening in today? Or would you like to participate?
strange pike
<@!309091952849321984> Are you listening in today? Or would you like to particip...
just listening thx
idle owl
Excellent, welcome!
tulip sleet
@onyx hinge there is a mild echo through your audio. It's turning green when kattni is talking
onyx hinge
weird, I turned down my headphones volume
tulip sleet
and i can hear you typing
idle owl
It sure is nice!
solar whale
I'm out of alphabetical order -- oops...
solar whale
no problem
idle owl
@gloomy shuttle Are you listening in today?
gloomy shuttle
<@!517804297401008138> Are you listening in today?
yes
idle owl
Excellent, welcome! We have a notes doc that we use for keeping track of folks, and I wanted to make sure I had you accurately in the list as listening in.
thorny jay
Jeff, if you can read that for me. Thanks.
onyx hinge
thorny jay
That is good question... what show are cancel and what show are running. I would love to try to make it for Show & Tell.
idle owl
Show and Tell is on this week.
ionic elk
@prime flower what board are you using?
prime flower
@ionic elk for what?
ionic elk
I thought I could get you a board bin mid CI but actually I guess I can't
ionic elk
anyway #3854 has the fix in and should pass CI pretty soon. Fixed the stub stuff
prime flower
@ionic elk whoo! I'm working on some other tasks before tackling that - how would I pull a .bin from the CI?
didnt know that was possible...
ionic elk
I can actually just build you one rn
prime flower
๐ that'd be great
ionic elk
but yeah once it passes CI it'll have a list of builds
onyx hinge
ionic elk
@prime flower here's where you find them
prime flower
sweet, thank you!
prime flower
@lone axle frame looks useful, can I put other widgets within a frame?
lone axle
It doesn't have any specif API for adding things inside but that is the intended use yep. For now its just up to user code to position things inside of it.
idle owl
@gilded cradle Was there some iteration of the MagTag lib that needed ESP32SPI at some point?
lone axle
It is extending Group though, it could perhaps override append() to handle the positioning and put stuff inside visually.
gilded cradle
No...
idle owl
@candid sun Is there a specific reason why the MagTag Daily Christmas Countdown needs the adafruit_esp32spi library? I'm not seeing it used in an obvious manner.
solar whale
latkes are good for all occasions
candid sun
@idle owl i think i was thinking that the magtag library needed it
idle owl
@candid sun Alrighty. I'll ask Noe to test it without.
candid sun
@idle owl apologies if i was mistaken. i can also test if that's helpful
idle owl
@candid sun Sure! That would also work.
@candid sun No apology needed - I'm going through the MagTag guides to get a warning added about updating libs, so now's the time to find it. ๐
solar whale
lots of examples for guides here https://github.com/adafruit/Adafruit_Learning_System_Guides
candid sun
@idle owl cool cool ๐บ yeah just tested and it does not need that library
idle owl
@candid sun Excellent, I'll pass that on. Thank you for testing!
modern wing
Thanks everyone ๐ Happy holidays -- hope you and yours have a safe and enjoyable season.
thorny jay
Thanks.
strange pike
happy holidays all
candid sun
@idle owl no problem!
modern wing
One more year in the bag ๐
prime flower
@ionic elk Thanks for the quick fix, will test and update the issue later today. Fighting with a compiler and CI script rn ๐
...the discord angry emote is really angry
modern wing
Woah, very angry indeed. Got up off the wrong side of the bed levels of anger ๐
solar whale
๐
onyx hinge
@lone axle I disconnected you from the voice channel
lone axle
Thank you
hearty tapir
Hi @onyx hinge , I'm not sure it is related but maybe for #3843 this could also be caused by one "fix" by me. I saw it as a fix, but maybe that made your JTAG misbehave: https://github.com/adafruit/circuitpython/pull/3817/files
onyx hinge
Here is the notes document for Mondayโs January 4's CircuitPython Weekly meeting. It is on January 4 (we are skipping a week during the holidays) 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 youโll be attending the meeting - itโs super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and weโll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1pOGIqgcxtlSQ3GSRD1UtbvKXwCyFx_1Gx_n2GhrGD8A/edit?usp=sharing
hearty tapir
The good news is that I see a country code where before was no country code. I'll still need to cleanup the extra characters there and I don't believe what I do is "valid" (even a LinkedIn Learning C-course later).
ionic elk
my MagTag is complaining about display rotation with a fresh install:
File "adafruit_magtag/graphics.py", line 57, in __init__
AttributeError: 'EPaperDisplay' object cannot assign attribute 'rotation'
Have I goofed on something simple? This is a fresh install
ah I guess it's a feature so recent it's not even on my main branch yet
hearty tapir
This is as good as it gets ... feels a bit hacky, but the workaround works fine. ๐
manic glacierBOT
hearty tapir
@slender iron would you mind to have a look at my concerns (that I'd annotate on GitHub) and advise better ways if my solution is invalid or unsecure? (referring to the above screenshot)
slender iron
@hearty tapir I don't mind looking at things on github. just tag me
is way behind on email at the moment
slender iron
The 32blit kickstarter (from Pimoroni) is reaching the backer. Any CircuitPython...
I have a beta board but haven't poked it at all. The basics shouldn't be too bad. It's the RGB display that will need work.
jaunty juniper
oh I am having "AttributeError: 'SPIDevice' object has no attribute 'spi'" with the sdcard module but there is already an issue about adafruit_bus_device in core being out of whack, which seems to be the problem
tidal kiln
@idle owl tides guide updated - lib folder screen cap updated and mirrored in more pages from magtag main guide
solar basin
I followed the ESP-IDF Getting Started guide for MacOS and got an error when trying to compile the hello_word example. I'd like to use the IDF for my Feather S2 or MagTag.
FAILED: esp-idf/cbor/CMakeFiles/__idf_cbor.dir/tinycbor/src/open_memstream.c.obj
@solar basin I've never actually used the idf on it's own
what version of the idf are you using?
solar basin
I pulled the repo and switched to 4.2 because I saw you had targeted that in CircuitPython
slender iron
I pulled the repo and switched to 4.2 because I saw you had targeted that in Cir...
Yup, I'm moving them there now
@tulip sleet has the crash I hit on thursday been solved?
tulip sleet
@slender iron nope. I determined that it wasn't a compiler bug, as far as I can tell looking at the assembler code. microdev thinks the object is being created too early. If they defer creating the object, it works
i wrote up my research in the PR
tulip sleet
my first prio based on the mtg today is to fix the storage.remount() bug. I was also working on the FrequencyIn bug. I found a problem, but it doesn't seem to be the cause of the bug.
slender iron
kk, ya it's good to have you bug hunting
tulip sleet
@slender iron also Limor is quite interested in a second CDC channel, and not necessarily gating that on the presence of dynamic USB descriptors. Second CDC is a 6.2 kind of goal for the new year
onyx hinge
@jaunty juniper I think that's distinct from the other issues I've reported, so I filed a fresh issue for it. Thank you for your report!
slender iron
@tulip sleet I think we should do dynamic descriptor
with the move memory work from cwalther it should be pretty straightforward
ornate breach
Hey @slender iron or dan, question for you
tulip sleet
@slender iron I think there are two levels: one is just turning existing devices on and off, and the other is allowing user-specified HID descriptors. Also there's USB and there's HID. The CDC is not HID-related.
but we can strategize later...
ornate breach
I'm writing some driver code for a board I plan to develop support for with circuitpython, and I am starting with the C code side of it. Do I need to define a specific i2c read/write method for either the C or circuitpython driver code?
tulip sleet
if the chip you are using is already supported, then you don't need to write driver code
ornate breach
this is how the device expects i2s to be written to
as far as I can tell, it isn't supported
yet
tulip sleet
sorry, I misunderstood, this is an I2C device... I was thinking you were porting CPy to a new boqard
ornate breach
I think @meager fog may be developing a board around the Atmel QT touch chips but I'm not confident
tulip sleet
if the chip is similar to one we already have a driver for, then you can probably take some of that code. You don't need to write C code
not atmel, but just came out
ornate breach
yeah, MPR121 is an older cap touch chip and she did a Great Search about capacitive touch chips which is sort of what inspired the board I'm working on
I am writing both circuitpython and an arduino library for it. Kind of a big undertaking for me since I've not really done this before
tulip sleet
take a look at existing drivers that access registers on I2C peripherals. I don't think what you want is new, it can all be done in CircuitPython. We have the register library to help
ornate breach
I'll take a look at the register library, I think Kattni suggested it as well
slender iron
@tulip sleet think your ble fix pr should go in a 6.0.1?
tulip sleet
what do you think the timeline is for 6.1 stable? I think it's kind of unknown, could be 2-4 weeks.
yes, it probably should, and also the storage.remount fix
slender iron
ya, I agree. I'd like to do another beta
tulip sleet
which looks quite simple; i am fixing it now
tulip sleet
i have to take the adafruit magtag rename out of the ble PR
i can make a new pr and cherry-pick the ble fixes out of the main PR branch
slender iron
either way
tulip sleet
let's merge this PR and I'll make a new PR with git cherry-picks.
slender iron
no
tulip sleet
i am a bit baffled by the frequencyio bug, which also rightly should go into 6.0.1
slender iron
fixes should go in their release branch first
so we can merge the release branch back into main without two fix commits
tulip sleet
ok, no problem, i will revert the magtag name commit and repush and
slender iron
๐
tulip sleet
will re-request a review
idle wharf
@slender iron I wanted to help push the ntp library over the finish line.
Should I send you a PR or would you prefer I go direct to the NTP repo?
lone sandalBOT
tulip sleet
^^ @slender iron I think I have a fix for the pin alarm crash. I committed it locally but can't push to the PR branch:
halbert@salmonx:~/repos/tannewt/circuitpython$ git push
remote: Permission to tannewt/circuitpython.git denied to dhalbert.
fatal: unable to access 'https://github.com/tannewt/circuitpython.git/': The requested URL returned error: 403
I'm a little confused because I thought that wasn't the usual error when I didn't have maintainer permission to push to a PR. But maybe I'm wrong.
in any case, you can apply the patch file to try.
hearty tapir
Wow, that sounds amazing! ๐
(I mean that the issue is fixed, not that you have permission issues :))
tulip sleet
This kind of error happens more often than I would like. We create an object on the heap, but forget to make sure that whoever is holding that object is included in the gc mark phase, so that the original object is not garbage collected. The holding pointer is in static storage and we forget to have it scanned by the gc.
One way this is done is to include the pointer to the object in the ROOT_POINTERs. The other is to write a special gc routine that knows where these pointers are. It should be on a checklist that whenever we create a statically allocated thing, we need to make sure that if it holds any Python objects, they get included in gc.
hearty tapir
I'll push a PR this (my) evening for "wifi.radio.ap_info.country". I'm not sure if I'd need to keep this detail in mind as well.
tulip sleet
well, I'll take a look when reviewing ๐
hearty tapir
That would be great, thank you ๐
I've completed LinkedIn Learning courses on C Essentials, something on C Structures and Embedded Applications in C - but taking courses doesn't teach experience ๐
tulip sleet
The radio object is a singleton, and has its own gc routine: see common_hal_wifi_radio_gc_collect() in common-hal/wifi/Radio.c
hearty tapir
Yes, that is the one you added ๐
tulip sleet
most embedded programs don't have gc ๐
hearty tapir
(when the bug hunting for i2c remove/deinit went on)
tulip sleet
so if you have added a Python object to the wifi_radio_obj_t struct, it needs to get added to that routine
hearty tapir
I'm planning to add a debugger (jlink or similar) to my setup - but I need to be careful not to make it a shopping spree
tulip sleet
https://www.adafruit.com/product/3571 is fine and only $20
hearty tapir
I'm overwriting an existing structure partially to workaround the ESP-IDF bug that is currently present - therefore this may be a cornercase.
tulip sleet
but it only helps a little with ESP32-S2 debugging, providing the basic JTAG interface
hearty tapir
Yes, that is the item I have already bookmarked. This will certainly be in the shopping cart sometime soon. I prefer to combine some parts to make the shipment more economical.
slender iron
@tulip sleet I literally just woke up thinking that was likely the issue. The extra log causes the pointer to linger on the stack and get collected
@tulip sleet do you need an ssh remote instead of an https one?
tulip sleet
no, i have not needed an ssh remote when doing this in the past. In any case, I think you can just git am the patch file. I made the patch from the top of the repo
slender iron
Kk
onyx hinge
@slender iron can you remind me why do we not use the "stub" for uploading esp32s2 firmware with "make flash"?
analog bridge
nice find @tulip sleet ๐
one question... Is cleanup_after_vm() called after _setup_sleep_alarms()?
tulip sleet
@analog bridge it is after a deep sleep, because the VM is shut down, since the program exits
slender iron
@onyx hinge I think it was because the USB bootrom can't do it
onyx hinge
@gilded cradle weren't you successfully using the stub with the USB on magtag to program it?
slender iron
<@!252717193496756235> I wanted to help push the ntp library over the finish li...
Either way works for me
gilded cradle
@onyx hinge yes, I use stub loading. It writes a small binary program directly to ram and tells it to run.
tulip sleet
@slender iron it may possible have to do with your repo not being a direct fork of adafruit/circuitpython, but of micropython instead. Not sure. I tried a couple of other things, not worth spending a lot of time on.
analog bridge
@slender iron should I wait for PinAlarm pr to get merged... than I can resolve conflicts with TouchAlarm
slender iron
yes please
tulip sleet
you may want to modify your code based on imitating the final version of pin alarm
tulip sleet
maybe, i never got a .wake_alarm that was not None, but the simpler example seems to be good, and prints the pin name
or maybe it was a display refresh issue, but I'm not sure.
i added the sleep_memory thing to make sure it was cycling
onyx hinge
@slender iron flashing magtag (with a debug build) takes ~17s with stub, ~37s without stub. --after-reset=hard_reset doesn't work over USB, but the stub DOES work and is faster overall. Maybe some older esptool did not have a working stub for usb? github shows that was added in May 2020..
slender iron
<@!252717193496756235> flashing magtag (with a debug build) takes ~17s with stub...
If it works now, go ahead and change it
onyx hinge
@slender iron thanks, done!
slender iron
ok, I've pushed the fixed version of pin alarm. Thanks @tulip sleet !
solar basin
So, I finally managed to get the IDF working, but I can't find any instructions for flashing CP back onto the FeatherS2
manic glacierBOT
tulip sleet
@solar basin hold down the DFU button, and press reset. Then it will be in bootloader mode and you can use esptool.py to upload the .bin, or follow simpler directions in the magtag or Metro ESP32--S2 guides. You can install a UF2 bootloader so you can load UF2's in the future.
last flame
@solar basin esptool.py erase_flash
then https://feathers2.io/install_uf2.html
then drop on the latest .uf2 firmware from https://circuitpython.org/board/unexpectedmaker_feathers2/ - thats how I do it ๐คทโโ๏ธ
solar basin
Thanks!!!
last flame
after entering bootloader mode as @tulip sleet mentioned
solar basin
You all are awesome as usual!
tulip sleet
@slender iron is there some debugging you want to do on light sleep? I started reviewing days ago but stopped due to the bug.
crimson ferry
fwiw, re: --no-stub, I found that I didn't need it when using the UART-to-USB, but did need it when loading over native USB (esp32-s2)
manic glacierBOT
analog bridge
I have run into an issue... cleanup_after_vm() is called after _setup_sleep_alarms() and resets the previously set touch configuration.
Is it advisable to use something like peripherals_touch_never_reset() to prevent this?
tulip sleet
@analog bridge, yes, exactly, you could use the same technique used for pin alarms, if that makes sense in this case. In any case you'll need to special-case those pins.
analog bridge
using never_reset(pin) won't work here as cleanup_after_vm() calls touchpad_deinit().
I am using prepare_for_deep_sleep() which is called from alarm_enter_deep_sleep() for touchpad configuration in case of actual deep sleep but the issue here is with pretend-to-sleep whoes configuration is set in _setup_sleep_alarms().
slender iron
<@252717193496756235> is there some debugging you want to do on light sleep? I s...
I think it's ready
I think we'll want to stop doing light sleep at all but that's an existing bug
esp light sleep shuts everything down
tulip sleet
@analog bridge can you just keep some info that touchpads will be used for sleep, and suppress the call to touch_pad_deinit()?
@slender iron I think "Circuitpython Light Sleep" != "ESP32-S2 Light Sleep". Other chips make it easier to keep some peripherals awake during a lighter sleep, I would think. We could make CPy Light Sleep for the ESP32-S2 not doing anything more than time.sleep() does.
alpine nimbus
I was just writing up installation instructions for my new MagTag projects and I realized how many libraries are necessary just for a plain MagTag installation. 10! That's at least 3 more than a few weeks ago when I started. What's the point of refactoring if the new libraries are required anyway? It just makes it more complicated for newbies. I can think of at least 3 libraries that I'm not using, fakerequests, neopixel and io. (i guess neopixel is used to flash when its booting up or starting the network or something)
although I used fakerequests during development and it rocks
analog bridge
@tulip sleet ok... I'll go with peripherals_touch_never_reset() to suppress touch_pad_deinit().
tulip sleet
or could you just pass params to touch_pad_deinit()
analog bridge
that would require altering peripherals_touch_reset() function which will break consistency with other module reset calls
tulip sleet
@alpine nimbus the refactoring makes it easier to keep all the "portal"-type code working the same way. ... We have been talking about a way of specifying a single URL that would download a custom zip package of the specified libraries
but that requires web infrastructure work; not hard, just another thing to do
alpine nimbus
@tulip sleet Thanks. That would be. nice
slender iron
<@252717193496756235> I think "Circuitpython Light Sleep" != "ESP32-S2 Light Sle...
right, I think cp light sleep should idle, not esp light sleep
tulip sleet
@alpine nimbus https://github.com/adafruit/circuitpython-org/issues/491
slender iron
@alpine nimbus you can also use the magtag without the library
splitting the big library apart should hopefully make it easier to use the pieces you need
alpine nimbus
The neat thing about the magtag (or at least my project that uses it) is that you don't need to do soldering or know anything about electronics and not much about software. I'm trying to make it as easy as possible for non-tech people to set up so that would help a lot.
alpine nimbus
One person already told me he bought a magtag because of it although he's a software dev so I doubt he'll have any trouble.
alpine nimbus
Oh, I just thought of something. Maybe in the future MagTags will ship with software already loaded so people won't need to install any libraries at all. The board in the new adabox is like that but I don't know about standalone boards.
idle wharf
Either way works for me
manic glacierBOT
ruby lake
tulip sleet
@ruby lake where did you get that link from?
ruby lake
one moment
The link is contained in the sentence "For example, the Circuit Playground Express guide has a great page on how to install the library bundle for both Express and non-Express boards."
tulip sleet
@ruby lake - fixed - that phrasing was from ages ago when the entire bundle could fit. Thanks!
ruby lake
I figured some things just sneak into the cracks, I just happened to by tinkering with a gps feather.
blissful pollen
@onyx hinge Just to confirm here would the 3rd method for bus device be the best then? It will probably take me a bit longer (maybe) but if it will work better no issue
onyx hinge
I believe it's the best all-around solution for compatibility. For code size and performance, well, it needs to be tested. but before it can be tested, it will need to be written.
I think right now we have a little bit of wiggle room on binary size due to the gcc10 transition, though I don't know the exact "free space" figures for the most constrained builds.
blissful pollen
It doesn't seem that hard just have to wrap my head around it. If its too slow could probably do a detect native I2C and if anything else do the python
onyx hinge
that's sure an alternative -- mix two approaches
Apologies for giving you background you didn't need, but perhaps it will be helpful to someone else interested in the task
blissful pollen
No worries it was all super helpful, I just want to make sure I do the best solution to it
I found when I did it the first time I used other modules as examples, having someone point out look here and there is a huge time saver, rather then having to hunt for them
night perch
When the necessary socket bind/accept/listen support is added for esp32s2, would that theoretically allow waking from light sleep when the device receives an incoming connection? Or would simply keeping the wifi connection alive rule out sleeping?
tulip sleet
@night perch we ended up not being able to use ESP32-S2 light sleep because it disables nearly all the peripherals. So we are just going to idle now during "light sleep". We will do some measuring to see if there is some power consumption reduction while idling.
night perch
<@!239513708257083392> we ended up not being able to use ESP32-S2 light sleep be...
Got it, thanks. Great tutorial you wrote on sleep btw.
tulip sleet
thanks! it will need revision soon: we'll add pin alarms and touch alarms when they're merged in
manic glacierBOT
crimson ferry
Is this line in start_station redundant? https://github.com/adafruit/circuitpython/blob/8b03951fee166b1eea06d114fc347982b7b186bf/ports/esp32s2/common-hal/wifi/Radio.c#L59 looks like it sets the wrong enum type/value, and it's overwritten correctly anyway in connect shortly after start_station is called: https://github.com/adafruit/circuitpython/blob/8b03951fee166b1eea06d114fc347982b7b186bf/ports/esp32s2/common-hal/wifi/Radio.c#L157 (scanning also calls start_station, but I don't think it requires any particular interface config/constant)
hearty tapir
Is this line in `start_station` redundant? <https://github.com/adafruit/circuitp...
Good catch! I'd even say that https://github.com/adafruit/circuitpython/blob/8b03951fee166b1eea06d114fc347982b7b186bf/ports/esp32s2/common-hal/wifi/Radio.c#L59 will not have any effect, as it has to be the wifi_interface_t interface https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-reference/network/esp_wifi.html#_CPPv419esp_wifi_set_config16wifi_interface_tP13wifi_config_t
It likely errors out, we could check with ESP_ERROR_CHECK(). I wonder now, if this is the corner case that you mentioned in which an APSTA is visible .... assuming you do import wifi and scan, but don't issue a connect.
hearty tapir
I've now tried a lot of different combinations and still I have no luck to catch a beacon.
crimson ferry
I haven't caught another either, and I'm starting to question my sanity at the time I did ๐
manic glacierBOT
This adds support to access the country code as announced by the Wi-Fi access point (if it announces it).
The workaround via esp_wifi_get_country() is necessary as esp_wifi_sta_get_ap_info() returns wifi_ap_record_t where wifi_country_t is empty. Reference to ESP-IDF bug report/issue: https://github.com/espressif/esp-idf/issues/6267
The discussion with @t...
hearty tapir
I wonder if "git" wakes up in the morning and thinks "What weird effect could I come up today with to confuse BennyE?!"
hearty tapir
I haven't caught another either, and I'm starting to question my sanity at the t...
I thought once that I got it, but it was the the neighbours robot vacuum cleaner (they'll show up as ECOVACS and they have an espressif inside)
crimson ferry
I get a bunch of Espressif "Sentry_" SSIDs with some hex digits, I assume they're some kind of security device in the neighborhood or in visiting cars
hearty tapir
(I do remember that you mentioned that it was your board mac + 1)
From a security POV many of the security-devices are questionable .. ๐
crimson ferry
(yes, the Beacon MAC was base+1, as expected for AP)
hearty tapir
@tulip sleet I don't really understand why the CI is unhappy with PR #3868 but at least all the xtensa builds were successful. I should probably apply the following to my PRs "Am I confident to make this a PR without a DRAFT PR first? If yes, make it a DRAFT" ๐
tulip sleet
<@!329766224093249548> I don't really understand why the CI is unhappy with PR #...
There is a new blank line at the end of common-hal/wifi-Network.c. That is the only error I see right now; it's checked by the pre-commit job
the trailing whitespace checker checks for that and also missing newlines at the end of files. Some editors have a setting or a plug-in to automatically clean up trailing whitespace on each line and at the end of a file, and ensure that they end with a newline.
hearty tapir
I'm using vim in my Win10 Ubuntu WSL. I haven't done any specific tweaks to make tabs spaces or similar. Do I need to do something about this? I'm planning to add some more wifi related enhancements.
tulip sleet
don't use tabs; always use spaces. But there are vim solutions for this: https://unix.stackexchange.com/questions/75430/how-to-automatically-strip-trailing-spaces-on-save-in-vi-and-vim https://vi.stackexchange.com/questions/22497/how-do-i-trim-whitespace-automatically etc.
hearty tapir
I've always manually hit the spacebar, there should be no tabs nowhere ๐ I'll look at the links, thanks
strange pike
I'm using vim in my Win10 Ubuntu WSL. I haven't done any specific tweaks to make...
i've found it good practice to check w/ git diff --ws-error-highlight=all before committing or submitting PRs ... it's not a big deal if i'm changing whitespace in my own code, but for contributing to a project i try to minimize changes as much as possible
hearty tapir
@crimson ferry , are you going to fix the Radio.c line 59 or should I?
hearty tapir
i've found it good practice to check w/ `git diff --ws-error-highlight=all` befo...
I'll make this part of my workflow. Thank you
tulip sleet
https://stackoverflow.com/questions/97694/auto-indent-spaces-with-c-in-vim might save you time. I use the tab key all the time (in emacs), but it's set up for spaces. Emacs has sophisticated autoindent modes. I was present at the creation of vi and used it for several years when I had to (1978-1984), but returned to emacs
crimson ferry
@hearty tapir Feel free, I'm still very early in the learning curve on C/C++, structure of the core, etc. I'm hoping I can focus on code and let GitHub do the builds for me (I know nothing about building software), but it's going to be a while before I'm comfortable PR-ing anything.
tulip sleet
of course many more modern editors like vscode also have nice autoindent
crimson ferry
(...just when I thought I was making some progress understanding parts of Radio.c (which make some sense being tied to a good API), I remembered about shared-bindings which lives between languages and is still pretty mind-boggling)
hearty tapir
(...just when I thought I was making some progress understanding parts of Radio....
I saw the shared-bindings during some of the deep-dive streams already and therefore took something that existed already (rssi) as example and walked through the files to understand the structure before adding the country code. I picked the country code, as I thought it would be the most simple element and didn't (possibly) imagine I'd run into the esp-idf bug that I now (hopefully) workarounded.
crimson ferry
That's my plan too, try to find something simple that has good close templates in other features.
tulip sleet
@slender iron I am going to start release notes for a 6.0.1 release that includes the merged and pending fixes: BLE, remount (EDIT:remount error started in 6.x, so not included), FrequencyIn. ... Do you think it's worth backporting the json speedup? that is well-loved and is the only other thing I can think of at the moment.
manic glacierBOT
I wired up a couple of LEDs and can confirm that open drain doesn't work, or at least not as I expect it.
In push/pull mode, one LED lights brightly depending whether the pin's value is True or False.
In open drain mode, neither LED lights no matter whether the value is True or False. However, the upper LED should light when value is False.
I did verify that in input mod...
crimson ferry
main() in main.c is run once and only once, on power-up, right? So I'm guessing that the case where wifi_inited would be true when port_init() calls wifi_reset() is RFU; ditto for the use of wifi_ever_inited (we never deinit wifi currently except at power-up)?
idle wharf
@crimson ferry thereโs an interesting post on ESP32 forum about making WiFi that reliably can be started and stopped. https://esp32.com/viewtopic.php?f=13&t=18633
Iโm not sure when WiFi is stopped in CP though.
Maybe when code.py vm is killed and you move to REPL vm?
hearty tapir
I played around with that, when we tried to fix the i2c issue. We no longer remove the event loop and also netif init only once
That is the function I also suggested to Scott, but after the i2c fix by Bruce it was no longer necessary
slender iron
<@!252717193496756235> I am going to start release notes for a 6.0.1 release tha...
Either way is fine with me. (I imagine you already decided.) Iโve got a dentist appointment now but will be around later
tulip sleet
@slender iron I didn't decide but did put the json fix in the release notes. I will do some PR's for you or jeff to review
considering how slow the old json is, i think it's almost a bug that got fixed
crimson ferry
@idle wharf afaict, wifi is only stopped (and [re-]started) using radio.enabled. Thanks for that link, I hadn't spotted that call in the API yet.
idle wharf
I scan that forum every couple of weeks... occasionally asking questions.
crimson ferry
I had been using enabled to try to reset wifi for a while to try to get around a bug, but it didn't do much discernible... maybe related to that post
hearty tapir
I had been using `enabled` to try to reset wifi for a while to try to get around...
You still hit an issue with Wi-Fi?
idle wharf
WiFi does have some things that need to be worked out.
If you do while True: connect, ping, sleep(60) you will fail on the 2nd or third loop.
crimson ferry
@hearty tapir it was a long shot to try enabled, probably are Requests issues
trim elm
@tulip sleet You wouldn't happen to have a .uf2 or .bin for the magtag for the very latest version of CircuitPython (with the Alarm library that had SleepMemory added) would you? I'm working on building it myself but it seems to be a pretty long process the first time
crimson ferry
@idle wharf common_hal_wifi_radio_connect doesn't currently check for existing connection/enabled, but I'm not sure the implications of that... there is a note on esp_wifi_connect: "2. If the ESP32 is connected to an AP, call esp_wifi_disconnect to disconnect." but there isn't an error response that says, "hey, you're already connected" so maybe it's no problem
none of the Espressif examples I've found have restart or even shutdown scenarios for connections or wi-fi running or wi-fi init, but they do say netif_init only once, and netif_deinit isn't implemented
hearty tapir
this is why there was โdestroy()โ used
idle wharf
<@!329766224093249548> You wouldn't happen to have a .uf2 or .bin for the magtag...
If the fix you're wanting is merged, you can get latest builds off the S3 link on the board specific page. https://circuitpython.org/board/adafruit_magtag_2.9_grayscale/ Click the "Absolute Newest" button
crimson ferry
@hearty tapir which led to my question about why we need to call wifi_deinit and netif_destroy in wifi_reset if it's only called on power-up (above) ...maybe RFU or a low-power thing?
trim elm
@idle wharf I had no idea that was a thing. Probably should have read a bit further down the downloads page lol. Thanks!
slender iron
@tulip sleet at my desk now
tulip sleet
i am trying out the 6.0.x with json speedup build. Want to take a look at the PR's?
slender iron
yup, I can
tulip sleet
something weird is going on with the json update; still testing
ok, I was building it with gcc10 by mistake
onyx hinge
should we suffix the build invocations so they call for arm-none-eabi-gcc-9.2.1 instead of arm-none-eabi-gcc ?
tulip sleet
it was just my problem lcoally, but we could do that. it's not a CI problem
for Scott, that wouldn't work if his toolchain advances versions due to arch keeping up with the latest
onyx hinge
it would make it possible to move seamlessly between 6.0 and 6.1 without modifying PATH which would be nice
tulip sleet
do you want to backport that change to 6.0.x, or just do it going forward on main?
onyx hinge
a reasonable question... I could arrange my symlinks so that the unsuffixed one was 9 as long as I needed it to be
hearty tapir
<@!586300798980587537> which led to my question about why we need to call wifi_d...
I found this more complete scenario really helpful from espressif: https://github.com/espressif/esp-idf/blob/master/examples/common_components/protocol_examples_common/connect.c look around line 278 on how they implement deinit
It is that example that let me try out esp_wifi_clear_default_wifi_driver_and_handlers()
slender iron
@tulip sleet I was thinking about that but it would help older portal projects
though they can always move to the beta
tulip sleet
i will test on a metro esp32-s2. The test json limor gave me is too big for samd51 anyway
slender iron
whoa. too big for samd51 is very big
tulip sleet
the file is like 250k
crimson ferry
@hearty tapir nice one, that's got a little of everything
tulip sleet
@slender iron we could discuss in the meeting tonight; it's getting relatively close to that time anyway, and maybe we don't want to release 6.0.1 on Christmas eve
slender iron
ya true. we could do next monday I think
tulip sleet
i could make a new beta earlier than that if we wanted with the BLE and FrequencyIn fixes, so anyone who's stuck could be pointed to the beta. But I"m not sure how stable main is right now
hearty tapir
<@!586300798980587537> nice one, that's got a little of everything
They use the same API / functions also for wired Ethernet
or at least very similar.
slender iron
@tulip sleet I think I'd rather just post the beta on monday as well
we can merge before that so S3 picks them up
tulip sleet
yes, so we can decide on the json->6.0.1 tonight, and merge that or close it, and merge the rest as soon as approved
slender iron
kk
thorny jay
There is that discussion about the power LED on the MagTag and @ladyada that want to have it.
But I discovered there is a way to have a LED only when 5V is present and not on when battery powered, this is what the FeatherS2 is doing:
The red power LED doesn't light up when running off battery
That is by design! The power and charge LEDs and corresponding 2K resistors drain the battery, so they have been isolated to the 5V power domain, so they will only light up and draw power when the board is either being power by the USB plug, or by a 5V power source connected to the 5V header pin.
(see https://feathers2.io/)
tulip sleet
@thorny jay @meager fog used a 10k resistor on the power LED to minimize the draw. I think it may have been conscious choice to have it powered in both domains, but I am not sure. Of course, one might use a USB power pack connected to USB as well, but usually those are hefty, so lighting an LED is not a big deal.
thorny jay
Discovering my FeatherS2, the replacement just arrived today. That is two travel from Australia to Belgium. ๐
slender iron
@gilded cradle https://forums.adafruit.com/viewtopic.php?f=60&t=173253
gilded cradle
Ok @slender iron, got it
slender iron
thank you!
slender iron
@tulip sleet need my review on anything else?
tulip sleet
@slender iron #3872, and #3870 if we decide to pull the JSON fix into 6.0.x
slender iron
donezo
tulip sleet
thansk!
hearty tapir
@slender iron I'll change the esp_wifi_get_country() to return mp_const_none if != ESP_OK. Does the way the cstr looks like ring any bell?
As per espressif docs the wifi_country_t.country.cc[3] so not sure why I see what was pasted above (https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-reference/network/esp_wifi.html#_CPPv4N14wifi_country_t2ccE)
tulip sleet
@slender iron ok, one more PR to get the BLE regressions into main. I had to do this by hand due to translation conflicts. The code part is identical. https://github.com/adafruit/circuitpython/pull/3874 Thanks!
manic glacierBOT
I understood what is happening here: This is the full country element, not just the .cc portion!
D (0x44) E (0x45) " " (0x20 - interpreted as "whitespace" although this stands for "Any Environment" in Wi-Fi context)
tulip sleet
@slender iron #3830 (pin alarming) looks like it needs make translate
slender iron
kk, will do in an hour. on the wrong computer atm
hearty tapir
lol, this country.cc is the second issue I'll have to open with espressif. Even when using the esp-idf directly, it has a length of six instead of two.
crimson ferry
what's the esp32s2 magic to get the ESP_EARLY_LOGW and similar to show up in the debug console?
tulip sleet
@crimson ferry you need to compile with DEBUG=1 on the command line. I have not tried the EARLY versions, just the regular ones.
crimson ferry
sorry for the dumb questions, I'm starting at zero... I don't find DEBUG in any of the config files at circuitpython/ports/esp32s2/
hearty tapir
@crimson ferry I've had the same confusion ๐ your make command needs to look like this: make BOARD=boardname DEBUG=1
crimson ferry
oh, I was hoping to use GitHub to do the builds for me (to defer that learning curve)
hearty tapir
@slender iron are you ok with mp_obj_new_str(cstr, strlen(2)); and a reference to the ESP-IDF issue ID ?
hearty tapir
no, actually not strlen(cstr) as the country.cc object of ESP-IDF function has a bug
slender iron
strlen(2) isn't valid
slender iron
ya, that's ok
hearty tapir
I'll get famous within espressif ๐
@crimson ferry if you're on Win10, I can recommend the WSL Ubuntu 20.04 LTS - this is what I use ๐
(Windows Subsystem for Linux = WSL)
crimson ferry
@slender iron I'm on macOS (11). Using Github to build may be a moot point... I assumed Actions would fork over too, like we grab artifacts from your repo to test. But no joy.
crimson ferry
OK, thanks, I'll start there
slender iron
np
manic glacierBOT
All suggested changes implemented. I also fixed the trailing line/whitespace that caused the CI to be unhappy.
tulip sleet
@slender iron I should have merged your pin alarm PR first. But I merged the ble fixes to main, and that caused another make-translate conflict. Sorry. I can revert the ble fixes temporarily, or you could do make translate again on the pin alarm PR.
tulip sleet
np
swift moss
Is there circuitpython support for the feather S2 built in wifi?
jaunty juniper
Is there circuitpython support for the feather S2 built in wifi?
yes
swift moss
Thanks, just found it.
half geyser
Is anyone here an expert at Github? I'm trying to create a PR (for the issue mentioned above), but it just keeps telling me "Unicorn!".
This is the page I'm trying to go to: https://github.com/adafruit/circuitpython/compare/master...xobs:fomu-fixes-6.0.0?expand=1
manic glacierBOT
Github appears to be broken, and can't create a pull request:
Please apply https://github.com/xobs/micropython/commit/f3e54414e5771a2463a8a93b4dcbb395584d2fe3 and https://github.com/xobs/micropython/commit/2f95cc95a494395658eefed6130b9a8732117482 to fix this issue
analog bridge
This is the page I'm trying to go to: https://github.com/adafruit/circuitpython/...
use main instead of master
half geyser
use `main` instead of `master`
Thanks for pointing that out. I have to manually modify the URL, since the Github UI shows me that unicorn whenever I try to create a PR.
strange pike
did you clone the repo to your account? might still have the clone set to master and need to manually modify
half geyser
It really wants to merge to micropython/master, to the point where I discovered that Github's UI lags, and I accidentally opened two PRs to micropython even though the UI said circuitpython.
My fork has fomu set as the default branch.
solar whale
@tulip sleet @slender iron I released an update to the Adafruit_CircuitPython_RFM9x Library. Will the next release of CP (6.0.x or 6.1.x) routinely be updated to incorporate the new release for the feather_m0_rfm9x build or do should I created an PR to update it?
hearty tapir
i've found it good practice to check w/ `git diff --ws-error-highlight=all` befo...
I made an alias for that in my .bashrc which helped me to correct the remaining whitespaces/trailing. Works like a charm, thanks again ๐
tulip sleet
@solar whale It is not automatic that they get updated; it tends to get done in batches. Do you feel it is a regression for 6.0.x from, say, 5.3.1, or that it is a kind of showstopper bug? Then happy to include it. You could submit a PR against 6.0.x (you have to change the branch dropdown when you submit the PR.
solar whale
@tulip sleet It is nothing major, just some enhancements to incorporate some errata from the radio datasheet and an update to the RSSI calculation. No real functional impact and no API changes.
If I submit a PR agains 6.0.x, then will that also get incorporated into main.
tulip sleet
not automatically, and given that it's minor, i'd rather just have it in main, and whoever wants it can get the beta. We are only fixing regressions or showstopper bugs in 6.0.x right now. But we should probably make a habit of updating the libraries shortly before every announced release. I have it semi-automated in terms of fetching the latest tagged commit.
solar whale
That would be good. I'll go ahead in put in a PR for main.
tulip sleet
Or just open an issue for 6.x.x, because we want to update all of the libraries
solar whale
OK -- will do!
manic glacierBOT
solar whale
I did verify that is still fits for en_US and zh_Latn_pinyin ... still several k to spare.
halcyon latch
Hey, wrt circuitpy implementation of neopixels, I can't find anything on the documentation about querying them to find out what color they are- can they do this, or do I just have to keep track of it elsewhere?
lone axle
@halcyon latch I think you can access them by index on the pixels object:
pixels = neopixel.NeoPixel(board.NEOPIXEL, 1)
pixels[0] = (10, 0, 0)
print(pixels[0])
Oh! That makes life a lot easier, thank you!
solar whale
@tulip sleet I just created a build from main for the feather_mo_rfm9x and it did not build in busdevice -- It built OK yesterday -- did something change?
tulip sleet
we turned it off for now because it does not work properly with bitbangio or other busio-"duck-typed" alternatives. https://github.com/adafruit/circuitpython/issues/3856
sorry wrong issue, edited
solar whale
hmm -- thats too bad...
tulip sleet
we know how to fix it but it is a regression for the moment
blissful pollen
I'm hoping to fix busdevice over the holidays. I think I know what to do and understand how the other types work now too
solar whale
FYI -- pip3 search is disabled ... https://status.python.org
hearty tapir
Ordered myself an Adafruit Feather STM32F405 Express, a J-Link Debugger and some Nordic stuff via Digikey as Christmas present ๐
tulip sleet
v nice!
hearty tapir
Rabbit holes everywhere .. I need to be careful ๐ haha
astral seal
Is it ever planned to use Websocket as a Circutpython lib?
jaunty juniper
I use uwebsockets on micropython and have been successfull in using it on Circuitpython with a few modifications (that I should probably post somewhere) https://github.com/danni/uwebsockets
astral seal
can you send me the modify version?
jaunty juniper
so I have this, it worked with my obs websockets code on the FeatherS2, but it's super hacky, based on an older version of uwebsockets, and it doesn't implement SSL (though that should be easily added back in)
jaunty juniper
I only use it through obswebsockets too