#circuitpython-dev
tulip sleet
yah, the next one: I said: This is the third alpha release of CircuitPython 6.0.0. It is relatively stable. The release is still alpha because some API changes and additions are not yet included. Upcoming in a future release but not yet included are ESP32-S2 Wifi support, and bleio HCI support for ESP32 co-processors.
slender iron
are there any actual changes or just additions?
tulip sleet
i don't think there are changes, but who knows π
slender iron
kk, lets do the next one as beta
tulip sleet
i agree, it will be close to feature-complete
slender iron
we can always add features with minor versions
idle owl
@slender iron Another yaml file would work.
idle owl
@slender iron Merged.
slender iron
Thanks! I'll add that yaml to the few repos that have tests too
midi and imageload I think
indigo wedge
I just saw the release notes for Alpha 3 mention BLE with esp32 co-cpu, is nina getting new functionality or is this with a separate fw? Very exciting π
tulip sleet
we are just turning on the BLE functionality and routing the UART to the right pins
it's just a minor rebuild of NINA. It's already in our fork
indigo wedge
ah so wifi over spi and ble over hci uart?
tulip sleet
yes, they can't be used simultaneously, as far as I know. But there's a complete new imple of _bleio that talks HCI, compatible with the old one
indigo wedge
nice! is it the same uart that's used for programming?
cause if yes then my imx feather will have ble working π
tulip sleet
ooo! Yes, the setup code is in https://github.com/adafruit/Adafruit_CircuitPython_BLE/pull/99. This is actually slightly out of date. I'll update the PR in a bit.
it's peripheral ONLY right now. it has been tested with Nordic UART, etc.
tulip sleet
we wanted to get something working for projects. Then I'll work on central.
indigo wedge
even just peripheral is amazing
tulip sleet
tnx! also no bonding. The code started as the ArduinoBLE library, but I reworked it a lot. I got some useful include files for struct definitions from zephyr
indigo wedge
right, their hci driver is quite mature
tulip sleet
i was distressed to find out that HCI leaves ATT implementation to the host; I expected it to be part of the command set (as in the nrf SD)
i am really tired of searching the 3000-page BLE spec. Also, there seems to be no community of HCI developers that I could find to ask questions: it's maybe too small a cohort
solar whale
I have a branch of the CP repo in my fork -- I pulled and merged the lates main - after 6.0.0-alpha3 release -- How do I get the branch build to reflect the new alpha.3 tag? it still shows alpha.2 -- If I just clone the main repo it builds as alpha.3.
tulip sleet
git pull --tags, maybe
silver tapir
@thorny jay I talked to AndrΓ©s Sabas, the Electronic Cats owner, and he says that maybe some alternative shipping option can be arranged. If still interested, please email store@electroniccats.com
thorny jay
Thanks @silver tapir I think I will wait for some Feather Wing from Adafruit or use the Saola: https://www.adafruit.com/product/4693
I have enough gadget to keep me busy.
solar whale
looks like git fetch upstream --tags did the trick
tulip sleet
@slender iron @ionic elk ugh all the ESP32-S2 builds failed in the alpha.3 release:
Run source $IDF_PATH/export.sh
Adding ESP-IDF tools to PATH...
Not using an unsupported version of tool cmake found in PATH: 3.17.0.
Using Python interpreter in /home/runner/work/circuitpython/circuitpython/.idf_tools/python_env/idf4.2_py3.8_env/bin/python
Checking if Python packages are up to date...
The following Python requirements are not satisfied:
gdbgui>=0.13.2.0
To install the missing packages, please run "/home/runner/work/circuitpython/circuitpython/ports/esp32s2/esp-idf/install.sh"
Diagnostic information:
IDF_PYTHON_ENV_PATH: /home/runner/work/circuitpython/circuitpython/.idf_tools/python_env/idf4.2_py3.8_env
Python interpreter used: /home/runner/work/circuitpython/circuitpython/.idf_tools/python_env/idf4.2_py3.8_env/bin/python
this is just a silly problem
onyx hinge
@slender iron just a note that I think that I narrowed down Phil B's problem to rotation= with rgbmatrix and that I probably failed to deal with it for Framebuffer displays in general -- I may need to pick your brain monday or tuesday about what handling that looks like, in case a little more code inspection doesn't help
onyx hinge
@slender iron groups don't have the rotate/transpose/scale options, tilegrids do. is there a big ugly reason why groups can't?
lone axle
Groups can scale I think
slender iron
@onyx hinge my brain couldn't handle it
onyx hinge
@slender iron I know the feeling tbh.
crimson ferry
On the esp32-s2, what is envisioned for using both ota_0 and ota_1 ...actually doing a future OTA implementation for updating circuitpython?
slender iron
@crimson ferry yes, I don't want to rule out OTA
thorny jay
Is it OK to download alpha3 directly from Amazon S3? Or there is a problem and it should be avoided?
tulip sleet
@thorny jay Yes, go ahead and download from S3. There were errors in the release process and circuitpython.org did not get updated
I'm working on doing the circuitpython.org update manually, but it's been tricky.
manic glacierBOT
manic glacierBOT
onyx hinge
visual benchmark of adafruit_sdcard vs sdioio vs sdcardio (from right to left)
I don't understand why the adafruit_sdcard one has the long pause between images, the three are running the same code. sdioio is plainly the fastest. sdioio is using stm32f405 feather, the other two are nrf52840 feathers.
manic glacierBOT
thorny jay
Because I don't have a Metro M4 Express AirLift... I wanted to replicate adafruit_matrixportal demo from John Park and @gilded cradle but on "Feather M4 Express" + "Adafruit RGB Matrix Featherwing Kit - For M0 and M4 Feathers" + "Feather Wing AirLift". After tweeking the code to use the RGB Matrix, then tweeking it to use the AirLift, I got the final error: "ValueError: D11 in use". It is not possible to do that... because some pin you need for the AirLift are also required for the RGB Matrix. I guess I need to wait for the ESP32S2 with build-in WiFi. π¦
crimson ferry
You could rig it I think, but not with featherwings stacked or doubled pin-for-pin
This may seem like a silly question, but should the following expression: led.value = not led.value toggle the output true-to-false or false-to-true, no matter what?
I've used this before and never thought much about it, it just works, as I think it should, unless I'm missing something.
But, here's something curious. This code:import board from digitalio import DigitalInOut, Direction led = DigitalInOut(board.L) led.direction = Direction.OUTPUT led.value = True led.value led.value = False led.value led.value = not led.value led.value led.value = not led.value led.valueworks fine on PyPortal. But on ESP32S2 (alpha.3), it does not work as expected.
On PyPortal, setting to True or False always changes the LED correctly, and prints the boolean value correctly
but on ESP32S2, negating the value doesn't behave the same
crimson ferry
>>> print(led.value)
False
>>> led.value
False
I realize reading an output is iffy, but to behave differently on different platforms is... unintuitive
onyx hinge
It doesn't seem very kind to the user to have "led.value" of an output pin not reflect the current state of the pin.
you might also say:: should led.value = led.value ever be able to change the state of the pin
likely it has to do with the esp port being young, and if you don't care to dive in and simply fix it, you should file an issue.
crimson ferry
I don't have the C chops to find and fix, so I'll file an issue. But I needed a sense check, this had me questioning the nature of my reality
onyx hinge
*
* @warning If the pad is not configured for input (or input and output) the returned value is always 0.
*
* @param hw Peripheral GPIO hardware instance address.
* @param gpio_num GPIO number. If you want to get the logic level of e.g. pin GPIO16, gpio_num should be GPIO_NUM_16 (16);
*
* @return
* - 0 the GPIO input level is 0
* - 1 the GPIO input level is 1
``` the esp idf has defined the low level routine in an interesting way.this makes the otherwise relatively sensible implementation in CircuitPython not work: return gpio_get_level(self->pin->number) == 1;
(I say relatively sensible because I'd have written ... != 0 or just return gpio_get_level() because I've been burned by code which returned an integral type and (inadvertently) returned values like 2, 4, 8, etc for "true". However the esp-idf docs seem to state that they promise to only return 0 or 1)
crimson ferry
Thanks, Jeff
onyx hinge
or it might be possible to set the pin to mode GPIO_MODE_INPUT_OUTPUT / GPIO_MODE_INPUT_OUTPUT_OD and not GPIO_MODE_OUTPUT
crimson ferry
"configured for input and output" is a new concept I think for CircuitPython
onyx hinge
Sure, happy to be a check
gilded cradle
@thorny jay, there's not enough free pins to use the RGB Feather and airlift feather at the same time. I considered adding support, but it just wasn't possible.
crimson ferry
@gilded cradle You could do it, but not stacked or doubled, as long as you can sacrifice I2C? A0, A1, SCK, MOSI, MISO, SDA, SCL are available, and Airlift uses the (3) SPI pins + (3) other pins, CS BUSY and RST, for routine use.
gilded cradle
Ah, that's true. I meant in a tripler.
thorny jay
@thorny jay, there's not enough free pins to use the RGB Feather and airlift feather at the same time. I considered adding support, but it just wasn't possible.
@gilded cradle Yeah, I figured that out after making chirurgical change to your library for 30 minutes. I was confused as to why you made a library for only one combo of hardware where the "portal for matrix" concept would work for other... well, I figured the "hard" way that I was hitting a pin limit. However, I believe some modularity where you can give your rgbmatrix and your AirLift pin would be usefull to this library. As soon as an ESP32S2 is there, you could have wifi internally and drive an rgbmatrix (no support for that yet I guess, and likely another Feather Wing will be required as it is each time different.
@gilded cradle You could do it, but not stacked or doubled, as long as you can sacrifice I2C? A0, A1, SCK, MOSI, MISO, SDA, SCL are available, and Airlift uses the (3) SPI pins + (3) other pins, CS BUSY and RST, for routine use.
@crimson ferry Maybe with a Bring Your Own Matrix Feather Wing... it is already more or less my case since I drive a 64x64 matrix added the "E" address line. So this mean I need one more pin. And then maybe I will hit the contrain of having the pin in the same "register" so that it goes faster to toggle them simultaneously. As the end of the day, I will have a unique hardware setup and unique problem.
I added one more pin.
manic glacierBOT
manic glacierBOT
crimson ferry
We can find with CircuitPython code whether a pin is capable of PWM, I2C, UART, SPI, touchio. But is there some mechanism deeper than pins.c to designate whether pins can be analog? digital? input? output?
stuck elbow
@crimson ferry same as with sercoms/timers β just try and catch the error
crimson ferry
@stuck elbow My question is whether it's possible for code to detect? For example, a processor pin that is input-only... CP code lets me output to it to my heart's content, but nothing ever changes on the pin.
(don't get me started on MOSI-only SPI devices, and breakouts/libraries that don't return init status π
stuck elbow
It should throw an error when you try to switch it to output
crimson ferry
what's the mechanism?
stuck elbow
if it doesn't, I would say it's a bug
crimson ferry
it doesn't detect or throw exception in my current case
where in circuitpython core code is the encoding of pins being input-capable, output-capable, analog-capable, digital-capable?
stuck elbow
I suppose it would need custom check for that on platforms that have such pins
crimson ferry
it would have to be in port/board I would think
stuck elbow
same for pullups/pulldowns
for analog there must be a place where it looks up the analog channel for the given pin, and throws an error when there is none
crimson ferry
that sounds right for analog, but digital may just rely on info in a datasheet
it it's input-only, for example
stuck elbow
generally I would think this is something defined in HAL
crimson ferry
OK, I'll dig around there, thanks
orchid basinBOT
manic glacierBOT
ornate breach
The sam4s has circuit python support, right? Iβm thinking about making my own feather with the sam4s since it has native USB and whatnot, I just wanted to make sure it was CP supported.
tulip sleet
@crimson ferry there is a table of pin capabilities for each chip family: https://github.com/adafruit/samd-peripherals/blob/0f5f1522d09c8fa7d858edec484a994c21c59668/samd/samd21/pins.c
This is basically some of the info in the datasheet, transcribed for programmatic use
@ornate breach we have not supported the SAM4S. The SAMD51 is an M4 processor with more RAM
SAMD51 has more RAM
I believe the SAM4S is older
ornate breach
Probably so
crimson ferry
Thanks, Dan, looks like it covers: interrupt, touch, ADC, sercom, timers. The thing that prompted this line of thinking was that on ESP32-S2, there are pins that are input-only, but I'm not sure if there's a way to programmatically enforce that.
ornate breach
I was just cruising through Digi-Key for M4 MCUs
Cooking up new ideas for feathers
tulip sleet
SAM4S was introduced in 2011; SAMD51 introduced in 2017
@crimson ferry the format of that pin table varies between ports; it's somewhat different on STM, for example, so it would certainly be possible to add an "input only" flag
tulip sleet
@onyx hinge I think there's a leftover debugging print statement from the endpoint checking:
Endpoint 0 - vs max_ep 8
Endpoint 0 - vs max_ep 8
Endpoint 0 - vs max_ep 8
...
solar whale
@tulip sleet I was going to try the esp32spi with BLE -- I have the updated nina firmware installed on a metro_m4_airlift, but I am having trouble getting _bleio into my build. What do I have to do to enable it?
or should I try first on some other board?
oh -- nevermind -- I forgot to install PR 3310!
solar whale
@tulip sleet FYI -- I tried running the ble_simpletest but get ```ress any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 6.0.0-alpha.3-48-gf5a5fc4c8 on 2020-08-31; Adafruit Metro M4 Airlift Lite with samd51j19
import ble_simpletest
scanning
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "ble_simpletest.py", line 12, in <module>
File "adafruit_ble/init.py", line 267, in start_scan
NotImplementedError:
the first few steps go ok ```Adafruit CircuitPython 6.0.0-alpha.3-48-gf5a5fc4c8 on 2020-08-31; Adafruit Metro M4 Airlift Lite with samd51j19
from adafruit_ble import BLERadio
ble = BLERadio()```
tulip sleet
@solar whale this PR only implements peripheral mode, so scanning is not working. You can try ble_uart_echo_test.py: that's peripheral mode
the idea was to get Bluefruit Connect up and running first
solar whale
OK -- sorry -- will try that
tulip sleet
I will make that clearer in the PR
solar whale
also of interest, I tried building my own copy of the nina_FW and it did not work ( got a time out lookin for the HCI ) I used the one from your PR and it works OK
I'll look into that
manic glacierBOT
solar whale
@tulip sleet Echo test worked great!
tulip sleet
The NINA-FW change has not yet been PR'd. It's on this branch: https://github.com/adafruit/nina-fw/tree/BT_HCI
solar whale
hmm -- it looks like that has been merged, but maybe I misread something --
tulip sleet
hmm, you're right, that's odd. I'll bring it up with Limor.
solar whale
thanks -- I'll try it again -- then I can open an issue on it.
tulip sleet
that would be great; she built the version I posted in the Cpy PR long before the NINA pr
The adafruit_ble_adafruit library should work also, on the peripheral sdie, though it has been somewhat less tested.
solar whale
OK -- I'll try to do some testing later today. Off to a good start! Thanks!
tulip sleet
thanks for testing!
solar whale
issue created https://github.com/adafruit/nina-fw/issues/26
manic glacierBOT
manic glacierBOT
digital shoreBOT
adafruit
manic glacierBOT
onyx hinge
<@&356864093652516868> Discord meeting in about 1 hour in the text and audio channels. If you have a chance to add your updates and your "lurking" status to the notes document, please do -- it's super helpful. https://docs.google.com/document/d/1fzARe32tu1diEPjFla3ipsz6jJehwYHlF7LYdjLaMYw/edit?usp=sharing
Reminder that if you plan to speak in the audio channel, you have to be a member of circuitpythonistas. We're happy to add you, let us know.
modern wing
Good afternoon all you wonderful folks! Happily (if a bit groggily) lurking today π β
The notes doc has been updated to reflect as such.
manic glacierBOT
tidal kiln
lurking
uncut nexus
lurking today
lone axle
Lurking today (noted in the document already), I will have to cut away early for another meeting.
onyx hinge
thanks @lone axle I'll be happy to get back in my role of URL paster
modern wing
The Duke of URL.
onyx hinge
OK make me a vanity role, I'll take it
stuck elbow
The Earl of Url
thorny jay
Sorry, I am also going to make it text-only.
onyx hinge
got it, thanks
onyx hinge
https://github.com/adafruit/circuitpython-weekly-newsletter/blob/gh-pages/circuitpythonday2020.md circuitpythonday@adafruit.com
https://hackaday.com/2020/08/24/running-a-successful-hacker-camp-in-a-pandemic-bornhack-2020/ https://github.com/bornhack/badge2020 https://youtu.be/1t1gyQ_yyU8
https://www.woolseyworkshop.com/2020/08/28/getting-started-with-circuitpython-on-compatible-microcontroller-boards/ https://twitter.com/JohnWWoolsey/status/1299444645441699843
slender iron
I wonder if we should split the overview into the three sections
onyx hinge
with 52 boards, you can do the adafruit blinka board of the week .. all year
modern wing
No problem π
onyx hinge
@slender iron want me to drop that screenshot in the chat or is it full of secrets?
MAKE := #problematic
modern wing
Awww.
solar whale
as long as you scratch her head
onyx hinge
π± . o O ( hug report to the big ones with thumbs who serve me food. it should be more often though. )
@slender iron only the top side connections are actually needed in that pinout?
slender iron
no, you need both sides
onyx hinge
if you want it pogo-able, do those need to be duplicated on the top with a little pad or something?
slender iron
pinout explained here: https://debug-edge.io/
modern wing
question about CircuitPython day -- do we/will we have a schedule of the adafruit broadcasts for the day?
slender iron
draft is here: https://github.com/adafruit/circuitpython-weekly-newsletter/blob/gh-pages/circuitpythonday2020.md
modern wing
Thanks Scott
slender iron
np
dire spruce
I am learning circuit playground. I noticed that some example codes just do import board while others do board and call out circuit playground. Which is right?
onyx hinge
belated hug report to scott for taking the next step in improving his development style
gilded cradle
another belated hug report to @onyx hinge for taking a look at the RGB Matrix issues in circuitpython.
onyx hinge
@dire spruce both are fine -- if you want to go beyond what is in the circuitplayground module, then you can do it by directly using board, etc.
if you're just starting you'll probably find the circuitplayground way easier
thorny jay
I wanted to use the neopixel library on a code that was already using neopixel_write... is there an easy way?
dire spruce
How do I know which boards are included in board library
onyx hinge
@dire spruce we're happy to help you but can you please move your question to #help-with-circuitpython ?
dire spruce
Thatβs where I am right?
onyx hinge
@dire spruce no, go up in the channel list, under "help with"
dire spruce
Oh thereβs a circuit python and a help with?
slender iron
(this channel is usually for circuitpython developmen)
onyx hinge
+t
dire spruce
Oh ok
onyx hinge
Thanks!
slender iron
I answered your question in the other channel
thorny jay
I played a lot with 64x64 RGB matrix this week, and it is only after a few days that I figured that my matrix has Blue and Green inverted (or maybe I made a mistake). Suddenly the image were much nicer. π
crimson ferry
@solar whale I've been getting hangs with SPI lately (esp32s2), though it was working in the past (had ESP32SPI connected to an Airlift working fine in July) Need to re-test with alpha.3
tulip sleet
limor said she will check out the firmware issue
onyx hinge
tidal kiln
wow. that's already coming up again.
onyx hinge
You generally need permissions to add labels
Sometimes you can restructure the code so that the lock is held for a longer length of time. If you are polling, then make sure you do it outside of that loop.
if (hri_sercomspi_get_CTRLA_CPHA_bit(hw) == phase &&
hri_sercomspi_get_CTRLA_CPOL_bit(hw) == polarity &&
hri_sercomspi_read_CTRLB_CHSIZE_bf(hw) == ((uint32_t)bits - 8) &&
hri_sercomspi_read_BAUD_BAUD_bf(hw) == baud_reg_value) {
return true;
}
``` @slender iron is reading these registers back going to be "fast" or is it slower than RAM?other thing of note is that samd_peripherals_spi_baudrate_to_baud_reg_value is called to get baud_reg_value and it uses floating point math
If it's less than 16 bytes it shouldn't look at background tasks
tulip sleet
// Avoid using C stack when making Python function calls. C stack still
// may be used if there's no free heap.
#ifndef MICROPY_STACKLESS
#define MICROPY_STACKLESS (0)
#endif
// Never use C stack when making Python function calls. This may break
// testsuite as will subtly change which exception is thrown in case
// of too deep recursion and other similar cases.
#ifndef MICROPY_STACKLESS_STRICT
#define MICROPY_STACKLESS_STRICT (0)
#endif
Thanks for running the meeting, @idle owl !
modern wing
Thanks!
gilded cradle
Thanks
mental nexus
Thanks everyone!
solar whale
π
modern wing
@idle owl If this is with a macbook pro, could it be the usb-c adapters themselves? I've had a few USB-C adapters not play nicely with external devices....
If you're a host of a zoom meeting, or the host gives you permission, you can record zoom meetings --- super handy feature. π
idle owl
MacBook Pro, yes. One of the monitors here is USBC directly. The other one is DisplayPort to USBC. In my other location, both USB C directly.
gilded cradle
Which one crashes it?
idle owl
Both setups crash.
gilded cradle
I mean which monitor? Either?
idle owl
Oh, I can't unplug the USBC one here because my network and peripherals are plugged into a hub plugged into it.
gilded cradle
Ah
idle owl
But it doesn't crash if I remove the DP-USBC one
modern wing
Hmm. I was thinking it could be the USB-C cable itself, first party vs third party? Or it could be the hub not playing nicely
idle owl
This is the USBC cable that came with the monitor. Huge gauge. Powers the laptop as well as providing a hub through the monitor
modern wing
The reason I'm diving down this rabbit hole is because I have some Microsoft-branded USB-C to USB-A adapters, and they simply do not play nicely -- but worse, they're inconsistent in how they fail.
idle owl
This is USBC to USBC cable.
gilded cradle
Ok, I usually use a displayport-usbc cable and had played with OBS a bit, but hadn't crashed. I just switched back to a DisplayPort to Mini DisplayPort since I could use a single hub.
Currently traveling, so using my laptop with an iPad as a sidecar.
modern wing
Sidecar is such a nice feature.
idle owl
I had forgotten about that. I also forget sometimes that I have an iPad.
gilded cradle
Yes, I just updated my ipad since my old one was too old for that.
It's a great travel monitor
modern wing
I run a 2nd gen ipad pro 12.9, and that's "new" enough for sidecar. It beats any USB monitor I've had hands-down -- and it beats Duet Display too.
gilded cradle
Mine is just a lower-end 10.2" iPad.
modern wing
Isn't it so nice though, older and lower-end hardware plays nicely
gilded cradle
Yes
I used a luna display with my old ipad and old macbook, but it was a little harder to set up.
modern wing
Anywho, I haven't tried OBS yet on my new laptop -- I'll give it a go with a few external monitors, see how it handles recording to disk with everything going at once.
slender iron
@lone axle you were working on a UI layout language right? https://forums.adafruit.com/viewtopic.php?f=60&t=168866
lone axle
yep. They are looking for almost exactly what I have created if I understand their last message correctly. I'll post a link to it in that forum post.
tulip sleet
yes
256kB flash (or 192kB with external flash, but we don't have any boards like that)
slender iron
(you really want more than 32k though)
tulip sleet
256kB on non-flash-chip boards is 192kB firmware + 64kB CIRCUITPY
slender iron
thanks @lone axle
ornate breach
So this would meet bare minimum? https://www.digikey.com/product-detail/en/microchip-technology/PIC32MX564F128H-I-PT/PIC32MX564F128H-I-PT-ND/2601579
if I used SPI Flash
tulip sleet
no, it's only 128kB internal flash. Also, we have never done PIC32
slender iron
the different architecture will impact the code size of circuitpython stored in flash
tulip sleet
we've spent a while looking for chips that meet the requirements
I'm not sure there are any "hidden" chips that look great
ornate breach
gotcha
slender iron
(the imx rts are underutilized imo)
slender iron
- M7
ornate breach
What about this? https://www.digikey.com/product-detail/en/microchip-technology/PIC32MX270F256D-I-PT/PIC32MX270F256D-I-PT-ND/4876287 256kB Flash and 64kB RAM
has USB OTG support too
Might be tough without the ARM core though...
hmm
slender iron
non-arm is doable
main thing is having gcc support
there is a little assembly needed for the nlr mechanics too
what is the appeal with pic32?
ornate breach
mostly just trying to do something that's not been done quite yet
I like challenges. though to be fair, I have a number already haha
slender iron
most challenges are software challenges
one challenge would be getting running on cortex-a
ornate breach
hmmm
I do see that there already is quite a big undertaking for RPi and other similar boards
slender iron
you mean blinka?
ornate breach
yeah
ornate breach
oh I see
slender iron
so you can plug in and edit away like any other board
ornate breach
that is a bit of a challenge huh
ornate breach
lots of low level software challenges when thinking about running right on the bare metal
ornate breach
I am working on some FPGA projects
like the ICE40 feather wing I talked about last week on show and tell π
it's a small step towards getting the biggest FPGA I can on a feather wing hehe π
slender iron
π what are you doing on the software side?
ornate breach
right now mostly just trying to load a simple blink verilog file to it
after that, reworking one of the binary to header libraries for C to work with CP
basically need to turn a binary into an array to be spit out over SPI
slender iron
have you looked at my systemonachip library?
ornate breach
I have not
slender iron
I've done a bit of that for the machxo2. I have loading over i2c from cp working
I'm trying to create python classes that are both the driver and hardware def
solar whale
@slender iron SPI works OK on esp32s2- saola-wroom -- with rfm9x featherwing.. Same code failed on wrover
sorry for double tag -- posted to wrong channel
ornate breach
very interesting, that might work for the ICE40 chips
slender iron
ya, my hope is that we can distribute bitstreams with the python lib so it's easy to load and then use
ornate breach
I'm not aware of i2c on the ICE5LP*K but I'm sure there are a few of the ICE40 Ultras that do have i2c programming
slender iron
the same approach should work over spi
you may want to try and do an 8 bit bus over the featherwing
ornate breach
yeah, just a few minor tweaks
I'll play around with it in the coming weeks. I just got the new regulators to replace the ones I got which didn't provide the right voltage so that will be happening over this week
slender iron
π
the next thing I'm hoping to do is create an i2c to wishbone harness for the xo2 to test peripherals on it
@solar whale mind answering https://forums.adafruit.com/viewtopic.php?f=60&t=168979 when you have time? thanks!
ornate breach
That would be interesting
slender iron
ya, I think so. basically start with individual peripherals and then work up to self-contained socs
onyx hinge
@slender iron where does displayio take care of the transposition and X/Y mirrors when actually sending pixels out to the device?
slender iron
@onyx hinge https://github.com/adafruit/circuitpython/blob/main/shared-module/displayio/TileGrid.c#L313
its done on a tilegrid basis. the buffer it fills is in native display orientation and the bitmap is read by row as well
onyx hinge
what about when the whole display is transposed, not a TileGrid?
onyx hinge
@slender iron if my display is native 32x16, and the rotation is 90, should its core.area = {x1 = 0, y1 = 0, x2 = 16, y2 = 32 } transposing x and y?
slender iron
yes, I thiiiiiink so
idle owl
Here is the notes document for Wednesday 9 September 2020's CircuitPython Weekly meeting. It is being held on a DIFFERENT day, at the typical time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if youβll be attending the meeting - itβs super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and weβll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1Zeeb_ZJltPcESMhRtofP4_QGoLWagNtAPfoAeA2q8Ms/edit?usp=sharing
manic glacierBOT
dire spruce
Anyone know why help with circuit python is gone?
manic glacierBOT
I reupload the default firmware (from https://github.com/wuxx/nanoESP32-S2), and I got this VID/PID used for tiny USB Mass storage
VID: 0xCAFE
PID: 0x4003
Should I use this one ? I don't know how to check if this is a valid VID/PID.
Here is USBDeview screenshot
manic glacierBOT
manic glacierBOT
ionic elk
Am I the only one who finds openocd super frustrating to set up/use? It seems like there's so little documentation on how to find the right configuration files/settings for your application
onyx hinge
I have not grown used to it either
but once I figure it out I put the settings in a script and forget about it
for my sam e54 which has to use openocd .. ```$ cat start-stub-same
#!/bin/sh
exec /home/jepler/.local/openocd/bin/openocd -f /home/jepler/.local/openocd/share/openocd/scripts/board/microchip_same54_xplained_pro.cfg -c "gdb_port 2331"
so if you figure out how to launch it for esp32s2 let me know
probably involves $HOME/.espressif/tools/openocd-esp32/v0.10.0-esp32-20200420/openocd-esp32/share/openocd/scripts/target/esp32s2.cfg
ionic elk
@onyx hinge yeah my biggest problem is simply that it loves to rely on these super specific files and whoever wrote their documentation just assumed that you would know what you need and where it is, magically?
onyx hinge
yup. If you had the time and energy you could try contributing to their documentation but you never know how that will be received when you're starting as an outsider
manic glacierBOT
tulip sleet
@ionic elk I used openocd once long ago to fix some fuses on Atmel boards, and found it quite frustrating. I was greatly relieved to be able to switch to the J-Link, which "just works".
ionic elk
@onyx hinge they aren't even on github, which gives me bad feelings about how receptive they would be to outside contribution
onyx hinge
o_O
ionic elk
Honestly the whole project is kind of impenetrable, does anyone know who even runs it? Like, is it just some guy, or is it sponsored, or what?
tulip sleet
@ionic elk Did you see there are mailing lists? http://openocd.org/discussion/mailing-lists/
ionic elk
I meant more along the lines of a wiki, or a readthedocs. Maybe it's the entitled millennial in me though lol. I've never really "gotten" mailing lists.
We have one for A2, and it seems so hard to follow conversations - I always figured there was some kind of special client I was missing that made them practical to use
solar whale
Are there known issues with CLUE and the current CP main? I am not seeing the REPL displayed on the screen and the 'spirit level' demo does not work -- no errors, just no display.
works ok with CP 5.3.1
No REPL on display with 6.0.0-alpha3 and later -- REPL works OK in terminal
The display does show the blinka icon and the bootup messages
lone axle
Confirmed same symptoms on a CLUE on my end as well.
6.0.0-alpha.2 is working correctly for me also.
solar whale
Thanks for confirming ...
lone axle
Mine is one of the "early" CLUEs that got listed in those first few listings at small QTYs. I know there was a warning on the page that hardware might still change, not sure if it did end up actually changing or not, and if so whether that change could affect this issue or not, but mine would be one of the earliest versions that were actually shipped out.
manic glacierBOT
tulip sleet
@solar whale btw, does your clue present as CLUEBOOT, or F840BOOT when in the bootloader
tulip sleet
I am helping some folks with a batch of CLUEs that are acting badly on MacOS. OK, thanks, I seem to have one CLUE that has FTHR840BOOT, for unknown reasons.
solar whale
I have an older one as well
tulip sleet
maybe it was my fault
solar whale
IIRC the original batch were FTHR840BOOT
Mine is on MacOS now -- no issues with that.
I was testing the "bonsai buckaroo" and it was working fine, but then i tried one of the graphic demos (spirit level) and it would not display. That s when I noticed the display was not echoing the REPL at all. So everything except display code seems to be working OK.
tulip sleet
I seem to have "new old stock" with FTH840BOOT
solar whale
I found one that does FTH840bOOT as well
onyx hinge
I have had CLUEs with the FTHR840BOOT
solar whale
Have to walk dog now. Iβll try some other displays later and try to pin down when it broke
ionic elk
@slender iron did you ever get any further with ESP32 debugging than the thread issues you were running into when you did your stream?
I'm getting the same "The current thread has terminated" and "Cannot kill process" stuff
Google is surprisingly useless at turning up anything for it
As best I can tell we don't have "thread awareness" but it's really not very clear on how to fix that: https://www.esp32.com/viewtopic.php?t=2755#p12946
manic glacierBOT
ionic elk
I think I landed on the answer, adding -Wl,--undefined=uxTopUsedPriority to LDFLAGS seems to have done it... whatever that means π
manic glacierBOT
solar whale
Yay! I can safely say I have a CLUE again....
onyx hinge
This was "fixed" once before !? https://github.com/adafruit/circuitpython/pull/3249/files
oh that was closed without merging, because we thought https://github.com/adafruit/circuitpython/pull/3281 fixed it instead
slender iron
@ionic elk nice! I never figured it out
@tulip sleet can you reply to this: https://forums.adafruit.com/viewtopic.php?f=60&t=168968
tulip sleet
@slender iron yes, though not right away. will put it on my list
solar whale
@onyx hinge when you have a chance, would you please add those notes to the clue issue?
onyx hinge
@solar whale why doesn't discord do that automatically?
solar whale
add that to the bot
onyx hinge
Weird my clue works with 600a3
@solar whale was there more to reproducing this problem than trying to show text on the lcd?
solar whale
Does it continue to echo the REPL. Try loading a graphic program like the spirit level demo
It was clearly flaky. Not always consistent.
Running graphics hosed the REPL echo
I always saw the boot messages, just nothing beyond if I entered the REPL
dire spruce
Help with circuit python room is gone. Does anyone know why?
ornate breach
Hit the help category
If there are no new messages and itβs collapsed, it wonβt show the rooms.
solar whale
Itβs there for me, I see your post.
crimson ferry
Is the https://circuitpython.readthedocs.io/en/latest/shared-bindings/supervisor/index.html#supervisor.Runtime.serial_connected caveat still accurate for SAMD, and what is the cause of this limitation? What about other ports? I'm wanting to come up with a way to store exception info when not connected to serial, but if supervisor.runtime.serial_connected isn't a reliable indicator, then I want to avoid disable_concurrent_write_protection in storage.remount.
If there's no SD card, the only other thing I've come up with is using nvm with a simple wear-leveling algorithm.
manic glacierBOT
manic glacierBOT
onyx hinge
In case you were wondering, it seems you can change the default branch setting for yourself (in the case of repos you create, not repos you fork) at https://github.com/settings/repositories -- for an organization, it's at https://github.com/organizations/org-name/settings/repository-defaults
(so happy to switch it over for my github user!)
slender iron
goes and changes my orgs
onyx hinge
as tempting as it was to just merge and be done with it π
slender iron
π
onyx hinge
@slender iron thank you, good reviews really do help.
I don't always keep the big picture stuff in mind, and it's easy to miss things that will confound the next person working in or reviewing the code, like core.
slender iron
I don't always keep the big picture stuff in mind, and it's easy to miss things that will confound the next person working in or reviewing the code, like
core.
@onyx hinge I'm on the opposite end. I think too big picture sometime and miss the details. My sinus headache helped with thecoreconfusion. π
ionic elk
@slender iron @tulip sleet the H7 series STM32 also comes with a built in camera module
tulip sleet
oh my goodness, here's a Python idiom I didn't know about for substituting a value instead of a falsey value (usually None). E.g., Instead of:
if foo is None: # use default of 33 if foo is None
foo = 33
you can do:
foo = foo or 33
huh
tulip sleet
because the or operator returns the operand, not just True or False
you have to being willing to give up strict None checking but often on writes if not foo: or similar anyway
slender iron
you and can also do foo = foo if foo else 33
tulip sleet
right, I was doing that, but this is shorter and is the same thing
slender iron
of course if it's a kwarg then you should just set the default
tulip sleet
i can't atcually, in this case, because the default is board.ESP_BUSY, which might not exist... or would that work anyway, because it doesn't get eval'd if the arg is supplied?
I should check that
nope, can't:
>>> import board
>>> def f(x=board.FOO):
... print(x)
...
...
...
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'module' object has no attribute 'FOO'
it'd fail on import I think
tulip sleet
then it might fail on import anyway, oops, have to use getattr, maybe
no, that is OK:
>>> import board
>>> def f(x=None):
... x = x or board.FOO
...
>>>
it'll error when you call it
tulip sleet
not if x is not None
slender iron
correct
tulip sleet
so this works nicely:
self._reset_io = digitalio.DigitalInOut(reset_pin or board.ESP_RESET)
onyx hinge
I find that style perfectly good, lots of languages uses that kind of chaining. but it is a surprise the first 7 or 77 times you encounter it, depending how much you understand and/or (and whether you bring wrong comceptions of how and/or operate in other languages to the discussion)
(e.g., a || b in C is either false or true, it's not 'the value of a if it's truthy, otherwise the value of b')
onyx hinge
I'm going to give something a try soon, once some parts come. The problems I have on Linux with USB and crashing/rebooting/unplugging CircuitPython boards are not as severe as some people here, but I've decided to try having a dedicated computer to plug the CircuitPython board I'm developing for into. I'll be attempting to use sshfs to access the device's CIRCUITPY and XYZBOOT files from my desktop, and probably running the serial terminal also in an ssh.
For the most part I don't have crashes on unplug, but the thing that's a real thorn in my side is that when I have gdb+jlink running and it is at a breakpoint for very long, my keyboard will stop being able to type until I unplug the debugged circuitpython board. So as long as the pi isn't super crashy in these circumstances, I will mind less if I have to reboot it multiple times per day than what I have to do now to recover from the confused USB due to debugger halt.
onyx hinge
@slender iron I have those but I almost always want to see the serial output too so I'm trying to figure out how to split the baby on this one
@tulip sleet if you were going to tackle the "rgbmatrix as built in display", please let me know if you run into trouble. I think it should be doable, but I didn't do it.
slender iron
I turn data off the moment gdb breaks
and if I'm using gdb I don't usually rely on serial usb
onyx hinge
if only there were a way for debug prints to come not through usb. even better if it was a standard, like swd.
onyx hinge
yeah same here .. and I'm not sure if pins are connected on the boards .. ?
slender iron
they may not be
onyx hinge
stm32f405 feather, the first I looked at, swo isn't connected at the debug header
onyx hinge
JTDO is used as FLASH_SCK so if it's SWO that won't work
slender iron
I don't think those are necessarily the same. what chip are you looking at?
onyx hinge
SWO is connected on Metro M4
slender iron
π
onyx hinge
first I was talking about STM32F405 Feather, had just taken a quick look at the schematic from learn
so I guess it would be interesting if in a debug build of metro m4, stdout went on swo also?
slender iron
ya
onyx hinge
I need more stuff to do to avoid making progress on can, eh
kindred zenith
With a pushbutton input, is there an easy way to detect long press as opposed to short press?
slender iron
we don't have a library. you can track the last time it was unpressed and then compare the duration when it is pressed
kindred zenith
What I want to do is e.g. "if pressed for one second, do x and stop"
kindred zenith
But not disrupt short press behavior.
slender iron
how are you detecting the short press?
kindred zenith
but1 = digitalio.DigitalInOut(board.A0)
while True:
if not but1.value:
keyboard.sendcode(Keycode.SHIFT, Keycode.P)
sleep(0.5)
slender iron
so you can get a reference time with time.monotonic()
kindred zenith
Sleep is to prevent multiple keypresses with a click.
slender iron
right, on a long press would you want to send the keycode too?
kindred zenith
So my question with time.monotonic() is how can I make it time out? I don't want to have to guess how long to hold the button.
Long press: different keycode
I want it to be like, press-release: send A, press-hold, timeout after time x and send B
slender iron
you save the time the button was last not pressed and compare the current time to that when it is pressed
kindred zenith
But that just evaluates on release, right?
slender iron
it depends on how you implement it
kindred zenith
I just couldn't figure out how to have it time out in the way I wanted.
slender iron
I think the tricky thing is that you will need to detect a release
otherwise you can't tell a short press from waiting for a long press
kindred zenith
I want it to be like: on press, start timing. On release do A. If no release after 1s, do B then stop.
slender iron
hrm
something like: ```python
but1 = digitalio.DigitalInOut(board.A0)
last_pressed = but1.value
last_press = None
while True:
pressed = not but1.value
now = time.monotonic()
if not last_pressed and pressed: # press
last_press = now
elif last_pressed and not pressed: # release
if now - last_press > 1:
# long
else:
# short
keyboard.sendcode(Keycode.SHIFT, Keycode.P)
sleep(0.5)
last_pressed = pressed
@slender iron thanks. But this waits for unpress, right?
slender iron
it does but you should be able to modify it
kindred zenith
Okay, I have to study it. I am not very good at python.
slender iron
np, this is good practice
silver tapir
Great news, we closed reception for speakers and talks for DΓa CircuitPython, we have a full agenda for the day.
https://diacircuitpython.org/schedule
One of the talks will be about designing new board for CircuitPython, and another on how to get UF2 to a new board. So we will have to add eng subtitles, as I'm sure you folks will love them.
manic glacierBOT
tulip sleet
Congratulations to the MicroPython maintainers, who released v1.13 a couple of days ago: https://github.com/micropython/micropython/releases/tag/v1.13. MicroPython is looking towards a 2-month-or-so release cadence: https://github.com/micropython/micropython/issues/6397
This opens up the idea of us merging from upstream from a defined release.
onyx hinge
Yay, and congratulations are sure to be in order
slender iron
anyone have a neotrellis? https://forums.adafruit.com/posting.php?mode=reply&f=60&t=169064
slender iron
thanks
olive lava
@pastel panther Bryan, thank you for posting your code for Adafruit_CircuitPython_ADXL34x and
Adafruit_CircuitPython_MPU6050 very very helpful!
@pastel panther ADXL34x tap, motion and freefall are solid examples. wondering if this could possibly be considered for the MPU6050 also, to be added
low monolith
Any news on the development of interrupts support for the CircuitPython? 8-)
tulip sleet
there are some async/await small libraries to check out in the main discussion issue: https://github.com/adafruit/circuitpython/issues/1380. Are you looking for real interrupts, or async event processing?
low monolith
@tulip sleet Thanks. Iβll have a look. I am looking for anything that could work π
manic glacierBOT
manic glacierBOT
manic glacierBOT
onyx hinge
huzzah! I just got a CAN packet out of my SAM E54 !
baud rate is not right
at least I think that's what Logic is telling me -- the dot/Γ should stay fixed with respect to the signal edges
onyx hinge
@pastel panther I'm ready to chat when you are
pastel panther
@onyx hinge yup, lemme just do a quick audio check
@onyx hinge not sure what's up, I just did an audio check in discord itself and it was fine
Β―_(γ)_/Β―
onyx hinge
odd
pastel panther
@onyx hinge can we use whereby?
onyx hinge
sure, we'll just make sure to mention any "results" of the meeting in here to keep the community in the loop
pastel panther
@onyx hinge can you do a quick check and make sure I have voice permissions in Amelia
onyx hinge
we used amelia last time didn't we? I'll see what I can figure out..
pastel panther
true... my phone was fine π€·
@onyx hinge nm, I can look into it later; whereby in one sec..
onyx hinge
I am pretty sure you are supposed to be able to connect & speak, you may not have video permission though
but you had video last time so π€·
//| def __init__(self, id: int, data: Optional[bytes] = None, *, size: Optional[int] = None, rtr: bool = False):
//| """Construct a Message to send on a CAN bus
//|
//| :param int id: The numeric ID of the message
//| :param bytes data: The content of the message
//| :param int size: The amount of data requested, for an rtr
//| :param bool rtr: True if the message represents an rtr (Remote Transmission Request)
//|
//| In CAN, messages can have a size from 0 to 8 bytes.
//|
//| For a non-rtr message, specify `data`. For an rtr-message, specify either `data` (a dummy buffer of the requested size) or `size`.
//| """
We didn't really have any big revisions to the draft API. Action items: @pastel panther will start sketching out adafruit_can, which will contain the non-Core implementations of Message, Match, and other ancillary objects; and work on the receive implementation on the MCP canbus chip. I'm going to add the missing "extended address" flag to my implementation of Message, and continue working on the sending end on the SAM E microcontroller. We'll probably have our next conversation after 9/9 CircuitPython day. And hopefully iron out the tech issues that stopped us from meeting on the Amelia voice channel.
assuming it's technically possible it would also be nice if adafruit_can had a specification of the CAN low level API (_canio.CAN) as an interface or whatever they're called in Python typing terms... I think it's possible.
the goal would be for that to be the general reference for everything CAN except the specifics of the constructor of a particular HW interface
pastel panther
Thanks for summarizing @onyx hinge
modest dagger
Hello, I am trying to build a custom c module and having a little trouble figuring out where to reference the module. I am following the guide linked below. I pulled down the 4.0 src to try and match up with the guide better but it still does not match. Specifically, the mpconfigport.h file. Wanted to see if anyone has any pointers on resources. Thanks
EDIT: Looks like I need to go further back for the guide to be relevant. Any info on doing this for newer versions is much appreciated.
thorny jay
I have a mysterious behaviour that I can reproduce, but not really know how to troubleshoot for the moment. My code works fine with the USB connected to a computer, but it quickly fail when connected to a powerbank or charger. I removed all the print but it does not change anything. My code seems to freeze in less than a minute when not connected to a PC. And that with or without opening Mu. (This ressemble to one of the current issue where CLUE does not behave the same.) But here my hardware is a Feather nRF52 Sense. It does: (1) Connect in UART over BT (2) measure pressure with I2C (3) change LED color (4) control the BT hardware. Similar code without the BT did work well in both case. Right now it seems to be related to BT and to negotiation of the CDC/mass storage with the PC. I would be interested in any known issue on alpha3 or any possible side effect of beeing connected to a PC.
I can try with nightly build to see if it is better now, or if it was better before.
I can try to simplify the code and make it reproducable without additional hardware.
But right now, I don't really know where to start to troubleshoot this issue... because it occurs when I don't have access to the console to troubleshoot it. And of course it does not occurs when I have the console.
How would you debug such a thing?
solar whale
@thorny jay is this on a CLUE? -- I ran into s similar problem and it is still in work to be fixed ... https://github.com/adafruit/circuitpython/issues/3367 If you can revert to CP 5.3.1 it should work.
In my case, the code actually runs, but there is just no display.
Sorry == now I see it is a Bluefruit sense. Not sure if it is related. Still, may be worth trying the earlier CP version just to check.
I found that it was OK at 6.0.0-alpha.2 so you might try that.
I read (or tried to read) your post before my coffee π -- It does look like you were aware of the previous issue and this is different but perhaps not unrelated....
AS far as debugging can you blink the D13 LED as an indicator of where in the code you are.?
manic glacierBOT
thorny jay
I read (or tried to read) your post before my coffee π -- It does look like you were aware of the previous issue and this is different but perhaps not unrelated....
@solar whale I reverted to 5.3.1 and it seems to work for me. It is not clear yet what the problem is... alpha2 did not work for me. There is a little bit of math so some stuff could go wrong such as float to int conversion, or divided by zero, so I could make things more robust, without any guarantee that the problem is in my code.
thorny jay
I believe there are issues with BLE in 6.0.0 and right now for two problems I encountered, I have things working with 5.3.1. One problem has been explained here 12H ago (something that stop working after less than a minute when not connected to a computer). And the other one has been discussed with @crimson ferry on #help-with-circuitpython just right now (ble.start_scan not exiting). I am reusing old code that I guess was developped in 5.x time, so it seems normal to work on 5.3.1. Maybe it need some adaptation for 6.0.0 ...
Maybe I need to rewrite from scratch based on simple_test demo code, just to make sure I am using the right/current pattern.
crimson ferry
This is weird, and I can't explain it...```import board
import busio
import digitalio
import adafruit_fram
this doesn't work
i2c = board.I2C
this works
i2c = busio.I2C(board.SCL, board.SDA)
wp = digitalio.DigitalInOut(board.D6)
fram = adafruit_fram.FRAM_I2C(i2c, wp_pin=wp)```
Traceback (most recent call last):
File "code.py", line 14, in <module>
File "adafruit_fram.py", line 225, in __init__
File "adafruit_bus_device/i2c_device.py", line 68, in __init__
File "adafruit_bus_device/i2c_device.py", line 170, in __probe_for_device
AttributeError: 'function' object has no attribute 'try_lock'```Adafruit CircuitPython 6.0.0-alpha.3 on 2020-08-28; Adafruit Feather M4 Express with samd51j19
adafruit-circuitpython-bundle-6.x-mpy-20200904
onyx hinge
You have to write: board.I2C() -- it is a function which returns the common I2C bus object. It is not an I2C bus itself.
onyx hinge
this is somewhat of a technical limitation, the value of every item in a built in module has to be fixed at build time
crimson ferry
i've done it a million times, but not this time
onyx hinge
hehe ok I'll stop explaining it then
lone sandalBOT
onyx hinge
I wonder how atrophied the ability to build without softdevice support is. that would get 56 or 64kB extra heap available, right?
onyx hinge
ooh it's slim enough to fit now. I haven't tested much of the functionality yet, or if it'll fit with sharp memory display framebuffer and the rest of my calculator program. ```Adafruit CircuitPython 6.0.0-alpha.3-92-g121975f8e-dirty on 2020-09-06; Particle Xenon with nRF52840
import udecimal
udecimal.Decimal("3.14159") * 2
Decimal('6.28318')
Decimal('0.3333333333333333333333333333')
onyx hinge
>>> udecimal.Decimal("1") / 3
Decimal('0.3333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333')
Decimal('2.302585092994045684017991455')
>>> Decimal(10).exp()
Decimal('22026.46579480671651695790065')
wow most of this even works properly π
onyx hinge
in addition to udecimal, I've also written trig functions atop it, based on the routines used in the unix bc calculator. https://gist.github.com/jepler/d6fd1133d926347192a1ff22d3995f0f -- the results agree with bc, except for rounding (bc rounds to zero, udecimal rounds to nearest) ```$ python3 -c 'from utrig import *; f="11.373"; print(atan(f), cos(f), sin(f), sep="\n")'
1.483094330462666230579325971
0.3685285931944669592773918397
-0.9296164133652692134932163577
only 1.5kB of mpy file, too
<@&356864093652516868> Just a reminder: the meeting is NOT happening tomorrow, it's moved to Wednesday at the same hour. Come celebrate CircuitPython Day with us on Wednesday 9/9!
cobalt grail
Found this string from shared-bindings/_bleio/Adapter.c in weblate - Cannot create a new Adapter; use _bleio.adapter;.
Why the use of semi colon here?
Weblate complained when I added a translation without it.
tulip sleet
@cobalt grail the trailing semicolon is a typo. The middle semicolon was just typical English punctuation
i can change either or both. I'm surprised weblat requires it
cobalt grail
It complains if I don't match the trailing semicolon. π
stuck elbow
it preserves punctuation of the original, I think
lone sandalBOT
turbid radish
?serverinfo
digital shoreBOT
adafruit
tulip sleet
@cobalt grail I'll fix it in some future PR, and remove the semicolon you had to add too :). Not worth doing a whole PR for a one character change right now π
manic glacierBOT
teal bear
Where is cpx build docs?
I have done it before, so somehow I figured it out, but I'd like to point others to something
lone axle
do you mean building Circuit Python for the CPX?
teal bear
@lone axle that would work as would expect all the build docs to be mostly consolidated. right now I am building for https://learn.adafruit.com/adafruit-stm32f405-feather-express
lone axle
This is the guide I always refer to no matter what I am building for. https://learn.adafruit.com/building-circuitpython
teal bear
yes that. thanks.
pip3 install huffman - isn't needed, it gets pulled in by https://github.com/adafruit/circuitpython/tree/main/tools huffman
manic glacierBOT
manic glacierBOT
pine plover
Does circutpython work with spi on the esp32 s2 now?
supple gale
Does circutpython work with spi on the esp32 s2 now?
@pine plover Not yet. Scott is planning to work on that during his live Deep Dive on Wednesday. At least not on th wrover module, maybe due to PSRAM support. https://discordapp.com/channels/327254708534116352/327298996332658690/752430434306228225
pine plover
Cool thanks :)
onyx hinge
huh. During a long debug-pause, information about the related USB device in /sys becomes unreadable. attempting to retrieve it hangs, but can be recovered with ctrl-c.
I wonder if it's related to the crashes I encounter during debugging on my desktop. root@raspberrypi:/sys/devices/platform/soc/20980000.usb/usb1/1-1/1-1.1# cat manufacturer
during that time, another USB device plugged into the same hub can't finish enumerating either
solar whale
Now that you are mentioning it, I do recall a warning from Scott to unplug the USB when debugging on my Metro M4. Easy to do there since it has an alternate power source. ...
onyx hinge
Yup, this is my workaround for unplugging the device under debugging
solar whale
I do have to ask -- why do you have a monitor (hdmi) plugged in to it?
stuck elbow
so it seems that the latest firmware update on my laptop fixed the crashing on disconnecting circuitpython devices
solar whale
Nice! There is hope....
tulip sleet
!!
stuck elbow
I mean, it was random, and it hasn't happened since, but you never know...
solar whale
I have been brave enough to try a few cycles in that past week, and survived ... what did you update - was it a new kernel or something specific to the laptop?
stuck elbow
system firmware and thunderbolt controller
onyx hinge
@solar whale just for the bring-up, I don't expect to continue needing it
solar whale
Thanks - just curious -- I did not see a keyboard connected, so i was trying to figure out how it came into play...
I happen to have a Zero-4u mounted on a pi zero (non w) with a WiFi dongle. I may see if I can get it working on that.
onyx hinge
I'll be interested to know how that goes -- on the 0W the onboard wifi is connected via an SDIO bus and survives USB getting confused, which is a plus
it let me actually inspect the state of the system at that time
solar whale
hmmm -- good point ... I have both w and non-W's available to compare ... the potential project list just keeps on growing...
What does tio add rather than using using screen?
oo-- I see -- it reconnects --- cool!
idle wharf
Anyone successful flashing their ESP32-S2 Kaluga from macOS? I see /dev/tty.usbserial-1413300 and /dev/tty.usbserial-1413301 are present but when I use the PORT on the make command along with flash it cannot connect to either device. I've also tried the holding boot and reset, release reset sequence. I feel like I'm missing a step...
example of my make command: make BOARD=espressif_kaluga_1 PORT=/dev/tty.usbserial-141330 flash
onyx hinge
interesting, cube roots are not exact in Decimal: ```>>> Decimal("1e3") ** (Decimal("1")/Decimal("3"))
Decimal('9.999999999999999999999999998')
but I guess 1/3 isn't exact either
Decimal("1e4") ** 1/Decimal("4") is exact
idle wharf
No issue with CP or my Kaluga. Bad cable. Cable now in trash...
manic glacierBOT
onyx hinge
7205759403792796.0
>>> round(_)
7205759403792794
``` well that's a bit odd(circuitpython unix port)
7205759403792794.0
>>> round(_)
7205759403792794
``` the error arises in the display of the FP value, not the rounded value (the second paste is from standard python3)
onyx hinge
Does anyone remember that some samd processors were hanging when trying to retrieve the microcontroller temperature? I've encountered that on the sam e54 xplained board this weekend.
onyx hinge
@slender iron if you're streaming the meeting tomorrow, does that mean you'll also take the timecodes?
tulip sleet
@onyx hinge It was hanging when reading microcontroller.cpu.temperature: I fixed in https://github.com/adafruit/circuitpython/pull/3052.
The temperature reading was difficult to set up and the latest errata say it doesn't work:
onyx hinge
ouch, okay
tulip sleet
But I did get it to work, albeit with less accuracy than I might have liked.
It might be that inserting a delay like i did with voltage would help
onyx hinge
on my sam e54 (sample of 1), it hangs in tip of main
tulip sleet
does .voltage hang?
onyx hinge
I should make double sure whether it was temperature or voltage. what actually hung was tab-completing microcontroller.cpu and I looked at it for a second in the debugger and then moved on
tulip sleet
well, the fix is already in main, so it -should- be ok. I insert a 1msec delay after setting the voltage reference for .voltage. You could try someting similar for temperature.
this was really tricky: I had two SAMD51 boards with seemingly identical chips (same revision) except for the date code: one hung up and one didn't
onyx hinge
>>> import microcontroller
>>> microcontroller.cpu.temperature
29.9796
>>> microcontroller.cpu.voltage
[hangs]```ugh; maybe try lengthening the delay
Line 298 in Processor.c
if you lengthen it, fine to have it be longer for both SAMD51 and SAME54, since they're almost certainly the same chip internally
onyx hinge
100ms delay didn't fix it
tulip sleet
it may be a newer rev of the chip that has just broken it completely
the muxing of the TEMP sensors is a mess internally
onyx hinge
2000ms + stopping in the debugger did
tulip sleet
you could try some delays before/after the SUPC bit settings, but that's just guessing. Or try adding a delay between the double calls to adc_sync_read_channel(), though it's probably hanging on the first call.
onyx hinge
well now I'm getting more puzzled. I put a delay of 1000 ms in the source, it reads the voltage back just fine, but .. there's nothing like a 1s pause
tulip sleet
but i wouldn't spend a lot of time on this. I think if you return NaN or something like that, shared-bindings will raise NotImplementedError
onyx hinge
i.e., it returns immediately. Shouldn't mp_hal_delay_ms(1000); be 1 second and very noticible by a human like me?
tulip sleet
uh, yes π . So maybe something broke recently in the tick stuff??
onyx hinge
OK, failure to load the new firmware
tulip sleet
so i bet the 1ms fix is fine, and there's something else wrong. if you call time.sleep() does it sleep? Does that use mp_hal_delay_ms ?
or maybe never mind π
onyx hinge
gdb with openocd is just a little bit different than gdb with jlink,and my fingers do the wrong things when I'm not watching them
also confusing things: if I decrease the delay it works ... until I read temperature (disturbing the mux?) and then read voltage again
3.32308
33.1896
[hangs]```right, I think the temp code changes the reference voltage setting. If you just repeat, the reference is already set, and doesn't need time to settle or change
the delay was just an empirical fix: there's nothing in the datasheet that says you have to delay. There is no READY bit to check, unfortunately
i bought one of these boards too, so I could try this too, and you could go back to CAN. I just have to find it; it's somewhere within 5 feet of me but I'm not sure where.
onyx hinge
I could also file an issue and move on
tulip sleet
i agree
tulip sleet
@onyx hinge I found my board. btw, how come you are using openocd instead of just the swd connector on board with a J-Link? is the EDBG connector openocd?