it is just saying "you can run it" as in "i've finished settings things up, the idf.py build command can now work"
#circuitpython-dev
spare jacinth
and the way you will (indirectly) run it, is by make BOARD=<whatever>
pulsar sleet
I know. So make should work but it doesn't
spare jacinth
instead of "doesnt work" you should show what the error is, dont you think?
not paraphrase to cause confusion, but a straight copy-paste
pulsar sleet
I am getting the same exact error as this message
spare jacinth
hmmm, i thought the export.sh script checks that everything is installed, maybe it doesnt
there was something like idf.py install... maybe? did you run that?
pulsar sleet
no.
I did try to install xtensa-esp32s3-elf-gcc but downloading the archive and copying all the bin files so that may be a issue
spare jacinth
(im assuming you have already folllowed this before asking, maybe you havent: https://learn.adafruit.com/building-circuitpython/espressif-build)
spare jacinth
make BOARD=<...> clean just in case
if you ran install.sh and gotten esp-idf to run export.sh without complaining mid-way, i have no idea what could be missing, tbh
pulsar sleet
it works after removing xtensa-esp32s3-elf-* that I manually downloaded
spare jacinth
kewl
pulsar sleet
what type do I use for a arg wanting a python int
tulip sleet
what type do I use for a arg wanting a python int
type annotation? foo: int
pulsar sleet
In C I have a function that needs a python int
tulip sleet
In C I have a function that needs a python int
in CircuitPython C code? do you need to extract the int from an unknown Python obj?
is this in a constructor?
pulsar sleet
yes. I am trying to add the proto arg to the socket but I don't know what type to use
tulip sleet
you would specify the type as MP_ARG_INT. See for example busio_spi_configure() in shared-bindings/busio/SPI.c
pulsar sleet
Ok thank you
tulip sleet
that will automatically do type checking that it is an int, and you can give a default value too. .u_int = 0, etc.
pulsar sleet
Can I just do a cast to int with that or does it work just like a int
aka how can I use that as a int in C
tulip sleet
args[ARG_proto].u_int will return an mp_int_t
look at at that configure code. It shows how to set up the arg list, give names to the args using the enum, and do range checking: mp_arg_validate_int_range()
all kinds of examples in there.
Generally, steal the idiom from an existing call that is like what you want.
if it's a required keyword arg in CPython, ues MP_ARG_KW_ONLY, otherwise omit that
spare jacinth
aka how can I use that as a int in C
those values use a union so that you can pass the same type (compiler's POV) and have polymorphism(ish) inside it
there are a bunch of bits on that args[ARG_proto] can that be interpreted in different ways, in this case, by using .u_int you say "hey, these bits are an int" (for other args you would "interpret" them as a float or whatever)
tulip sleet
in this case, u_ means union (not unsigned)
slender iron
the ble workflow will use some bleio apis when the vm isn't active
pulsar sleet
do the args have to be a list or can I just add a arg like
mp_arg_int proto = 0
tulip sleet
STATIC mp_obj_t socketpool_socketpool_socket(size_t n_args, const mp_obj_t *pos_args, mp_map_t *kw_args) {
enum { ARG_family, ARG_type };
static const mp_arg_t allowed_args[] = {
{ MP_QSTR_family, MP_ARG_INT, {.u_int = SOCKETPOOL_AF_INET} },
{ MP_QSTR_type, MP_ARG_INT, {.u_int = SOCKETPOOL_SOCK_STREAM} },
};
socketpool_socketpool_obj_t *self = MP_OBJ_TO_PTR(pos_args[0]);
you are going to add something to the enum, and to the allowed_args list., and then validate the args[ARG_proto].u_obj value that you fetch.
also update the inline documentation that's above that
pulsar sleet
```c
STATIC mp_obj_t socketpool_socketpool_socket(size_t n_args, const mp_obj_t ...
where is this code
tulip sleet
shared-bindings/socketpool/SocketPool.c
spare jacinth
moving here from #help-with-circuitpython message, i wont be looking at this for the following days either, so not a moving thing right now (maybe not in a few days either), just dropping a little idea
it feels like a bunch of the time is spent resolving the circuipython.local name to an actual IP (as the browser shows nothing at all), but i have no way of confirming this, any ideas how to measure such thing?
- if that's the case, i dont think we can do much on that end
- if resolution is fast, but the actual loading of front page is slow, we could maybe send a very simple HTML so that the user sees some movement rather than a dead loading screen... as of now, from a user perspective myself, it just looks like the device is dead or just barely alive 😛
mortal kernel
the ble workflow will use some bleio apis when the vm isn't active
I see that now. Is it your intention for each port's _bleio to have a ble_drv.[c,h]?
(Or similar. The BT workflow implementation makes me think so.)
pulsar sleet
whenever I try to build circuitpython for the pico w I get a permission denied error
make: cmake: Permission Denied
crimson ferry
@pulsar sleet I can test Pico W once artifacts build, do you have starter code you can share?
mortal kernel
the ble workflow will use some bleio apis when the vm isn't active
The call to ble_drv_remove_heap_handlers makes complete sense to me now. Thank you.
tulip sleet
@lone axle#2825 does this ring a bell? https://forums.adafruit.com/viewtopic.php?t=206964 I thought this was working. Not sure if it's just this bundle or not. debouncer requires ticks. Guide in question is https://learn.adafruit.com/mini-pinball-table-with-gemma-m0/code-the-gemma-m0-with-circuitpython
not urgent
lone axle
<@382939733107408897>#2825 does this ring a bell? https://forums.adafruit.com/vi...
That is odd. I don't recall running across it before.
I'm not sure how the learn project bundler works, but it appears like it has trouble with adafruit_ticks.
tulip sleet
i was wondering if it was an issue with the debouncer library
i don't know whether the bundler looks at requirements.txt
lone axle
The only two things I can think of as possible causes would be: 1) debouncer not having ticks in it's requirements.txt, but it does and it seems to be correct https://github.com/adafruit/Adafruit_CircuitPython_Debouncer/blob/main/requirements.txt or 2) ticks not being in the bundle (if the learn project bundler relies on the built library bundles which I assume it does, but could be wrong.) But that isn't the case either, I double checked the bundle and ticks is definitely in there.
tulip sleet
yes, I would look for some similarly transitive dependency and see if the bundle for that works. have you seen adafruit_ticks missing before?
lone axle
I don't recall seeing it missing before
tbh I don't recall ever seeing a bundle that was downloaded from learn be incomplete or missing anything.
tulip sleet
maybe it is just this bundle. I will look for other examles in the learn guide and see how their bundles are
lone axle
This one also relies on debouncer and therefore ticks, but is missing ticks in it's download: https://learn.adafruit.com/ambient-machine/code-the-ambient-machine And in this case the project uses the automated screenshot generator which I noticed also failed to include adafruit_ticks
I worked on the screenshot generator so I am more familiar with it's operation. I can try looking into it later on in the screenshot generator context and if I discover the root cause it may be the same as for the project bundler.
tulip sleet
pip thinks the dependency is OK:
$ pip3 install adafruit-circuitpython-debouncer
Defaulting to user installation because normal site-packages is not writeable
Collecting adafruit-circuitpython-debouncer
Using cached adafruit_circuitpython_debouncer-2.0.8-py3-none-any.whl (5.7 kB)
Collecting adafruit-circuitpython-ticks
Using cached adafruit_circuitpython_ticks-1.0.13-py3-none-any.whl (5.4 kB)
[...]
it doesn't look inside adafruit_debouncer.py, does it?
The import there is a from..import: from adafruit_ticks import ticks_ms, ticks_diff
debouncer has an optional_requirements.txt, which I have not seen before.
lone axle
Maybe https://github.com/mgedmin/findimports can only traverse 1 level?
tulip sleet
(re my previous post) never mind, they all have optional requirements.
lone axle
or at least does in our case only do one level. It wouldn't have access to the library .py file easily that I know of so it couldn't really recurse to check for imports that came from libraries.
lone axle
The screenshot generator does. And I think maybe I heard that the project bundler does back when I was working on the screenshot one, but I am not certain.
tulip sleet
says it is olde than setuptools, so maybe it doesn't even look at requirements.txt.
I would have thought the bundler and the screenshot tool would both rely on pyproject or setuptools and would be using requirements.txt at least indirectly
but I guess I am wrong
lone axle
I'm trying to think of other instances of 2 levels like Project depends on Library1 and Library1 depdends on Library2 like this. I'm sure there are other cases but I am drawing a blank at the moment.
Learn guide project code doesn't always have a pyproject.toml or requirements.txt though I don't think.
tulip sleet
Learn guide project code doesn't always have a pyproject.toml or requirements.tx...
sure that makes sense. well, I will try to run findimports by hand and see what happens. Its README suggests using pydeps
i guess I would think findimports on the learn guide code and then requirements.txt subsequently
ok, well...
halbert@cod:~/repos/adafruit/Adafruit_Learning_System_Guides/GemmaM0_Pinball$ findimports
./code.py: could not find adafruit_hid.keyboard.Keyboard
./code.py: could not find adafruit_hid.keycode.Keycode
code:
adafruit_debouncer
adafruit_hid.keyboard.Keyboard
adafruit_hid.keycode.Keycode
board
digitalio
usb_hid
no mention of debouncer adafruit_ticks
stuck elbow
the first one?
tulip sleet
sorry, no mention of adafruit_ticks
lone axle
I did come up with a 2 level dependency chain Adafruit_IO depends on Adafruit_MiniMQTT. But the constructor for adafruit_io actually needs to have an instantiated mqtt object to pass in to it, so the project code ends up importing adafruit_mqtt in adition to adafruit_io therefore it doesn't end up suffering from the same problem.
Here is the section of code responsible for building the list in the screenshot generator: https://github.com/circuitpython/CircuitPython_Library_Screenshot_Maker/blob/main/get_imports.py#L152-L167 it doesn't seem to account for multiple levels. It's only finding imports from the project code files, but not recursing into the library code as far as I can tell.
tulip sleet
hmm so do pyportal bundles omit portalbase??
lone axle
It's not an ideal solution, but based on the way that adafruit_io and adafruit_mqtt interact and end up working with the screenshotter and bundler it would seem a temporary workaround could be to add import adafruit_ticks to the project code.
tulip sleet
for example
lone axle
Ah good thought those should make a dependency chain with multiple levels I'll check one of those
This is a pyportal project and the bundle and screenshots do not end up including adafruit_portalbase https://learn.adafruit.com/pyportal-air-quality-display/code-pyportal-with-circuitpython
In that case it works when users try it in most cases because they device they'd be running on has portalbase frozen in I think so they don't end up missing an import even though it wasn't included.
tulip sleet
so the way the whole thing works now is ... poorly
lone axle
Unless I'm misunderstanding it, it definitely does seem prone to missing any dependencies past the first level. And we've kind of just have been mostly lucky that in the cases it misses the libraries it missed were included due to being frozen in.
tulip sleet
circup does a fine job:
$ circup install adafruit_debouncer
Found device at /media/halbert/CIRCUITPY, running CircuitPython 8.2.9.
Searching for dependencies for: ['adafruit_debouncer']
Ready to install: ['adafruit_debouncer', 'adafruit_ticks']
Installed 'adafruit_debouncer'.
Installed 'adafruit_ticks'.
it's doing the right thing
lone axle
circup does reference either requirements.txt or pyproject.toml I belive.
I think it'll need to have a second "stage" added to the operation to work through the libraries it finds on the first pass and find thier dependencies based on one of those files as well.
Perhaps the code from circup can be re-used in the screenshot generator for that. I don't know how the backend of the bundler works or if it is python though.
tulip sleet
it looks in requirements.txt
the bundle itself has a requirements directory with adafruit_debouncer/requirements.txt, etc.
lone axle
One thing to consider is if we do want to take into consideration the list of frozen libraries as well. i.e. should the screenshot and bundle actually end up including portalbase since most of the time they'll be frozen in?
although consistency between circup and the other tools would be ideal. So whichever we land on, all should behave the same.
tulip sleet
the bundler is not specific to a board, so what's frozen can vary. circup does not check for frozen, it always installs all the requirements
short tendon
Not able to fully see what your issue is, but feels related to something I wrote. I have a script that reads through my lib folder, and then fetches all the related requirements.txt to see if there's anything missing. For example if I have adafruit_adafruitio it fetches https://raw.githubusercontent.com/adafruit/Adafruit_CircuitPython_AdafruitIO/main/requirements.txt (I do have a lookup for some)
pulsar sleet
Does circuitpython support warning. I am working on adding the proto argument to sockets and I want to show a warning if proto is set put the socket is not a SOCK_RAW sock
spare jacinth
there is a subset of it, warnings.warn should work (not all the same options that you get on computer)
or, if you need to call it from C, it would look something like:
warnings_warn(&mp_type_FutureWarning, translate("%q moved from %q to %q"), MP_QSTR_Display, MP_QSTR_displayio, MP_QSTR_busdisplay);
(extracted from https://github.com/adafruit/circuitpython/pull/8524)
translated to human that is:
warnings_warn(type_of_warning, message)
messageis a format string with variadic args.... which in this case is equivalent of"%s moved from %s to %s" % ("Display", "displayio", "busdisplay")
pulsar sleet
what is a future warning
spare jacinth
you'd likely want (if implemented) something a UserWarning i guess
pulsar sleet
How can I build with CIRCUITPY_WARNINGS enabled or is it enabled by default with just a make BOARD=
spare jacinth
Check the PR i linked
Since it added the thing, it also has to have added the flag that controls enabling it (at the Makefiles)
(which likely defaults to true)
tulip sleet
Does circuitpython support warning. I am working on adding the proto argument to...
what is the behavior in CPython? We would want to emulate that
warnings is on by default on all boards, I think, except maybe the absolute smallest
mortal kernel
To activate BT workflow on the nrf52 boards the reset button is pressed while the neopixel rapidly blinks blue. The RP Pico W has only a green LED. Any thoughts on an activation sequence for these boards?
The RP Pico W also lacks a reset button.
midnight ember
a getenv() special variable? if it works for enabling web workflow then could be used to enable bt workflow?
stuck elbow
a getenv() special variable? if it works for enabling web workflow then could b...
but how are you going to edit settings.toml without usb connection?
midnight ember
how do they do it currently with web workflow? I have no idea, never used one in that way.
spare jacinth
They still have serial, dont they?
stuck elbow
not on an ipad
the whole idea with the bluetooth workflow is that you can use it on a phone or a tablet
mortal kernel
From a CP development perspective it may be useful to have both a button press at reset and a settings.toml variable that can activate BT workflow.
midnight ember
If you want a button press with a Pico W you'd have to use a Pico W Bell, then maybe.
stuck elbow
maybe there could just be a separate firmware with the bluetooth workflow enabled
I just realized that you need usb anyways, to flash the firmware
mortal kernel
If you want a button press with a Pico W you'd have to use a Pico W Bell, then m...
That's probably the case. I'm looking into the bootsel button to see if there's maybe a way to use it.
stuck elbow
That's probably the case. I'm looking into the `bootsel` button to see if there'...
iirc that one shorts the flash CS pin to gnd
spare jacinth
It does, using it at runtime is possible but quite hacky (and probably flaky), wouldnt do it
At boot, if it is shorted to GND, RP2040 goes to bootloader, otherwise gets configured as the CS for flash
mortal kernel
iirc that one shorts the flash CS pin to gnd
Yes, that's what the schematic shows. Not useful for this purpose.
stuck elbow
I have it connected to an additional pin with some resistors, so I can sense it without messing with the cs pin on my board, but not sure how pico has it
mortal kernel
Pico has it through a 1K to ground.
The reason given in the docs for the nrf52 behavior (BT workflow off by default) is security. I wonder if security really is an issue?
stuck elbow
you don't want random people under your window to connect to your device
mortal kernel
you don't want random people under your window to connect to your device
I'm not able to get a connection further than about 6" away.
How about this: BT is off by default (current behavior), but can be activated by a setting in settings.toml?
stuck elbow
I can see all of my neighbour's bt devices perfectly fine
mortal kernel
I can see all of my neighbour's bt devices perfectly fine
When advertising, BT workflow turns the radio power down.
stuck elbow
imagine a classroom
mortal kernel
OK, I'm starting to understand why it is like it is. Back to my question: BT workflow off by default but activated in settings.toml?
tulip sleet
The BLE workflow was conceived of before either MSC or serial worked at all with tablets, either Android or iOS. We can reconsider how it works now. It was designed to be simple to use, and originally boards like CPB were the target. Whether this still makes sense or not I don't know. I don't think we have much if any data about how it's being used now.
I would say just pick something for now, and we can change it later if we want.
manic glacierBOT
It might be tough to state an exact maximum achievable voltage for a particular variant
Well, CP uses the calibs. We do not have to bother with the chip variance (theoretically).
The technical reference manuals for esp32, esp32s2 and s3 state that:
But vref is not defined anywhere in this document.
Assuming it's 3v3, it would mean it's 3300÷4095×3300=~2660.
But that clear...
manic glacierBOT
slender iron
I see that now. Is it your intention for each port's _bleio to have a `ble_drv.[...
I don't think so. I probably just got lazy. Ideally it'd be functions defined in shared-bindings that all ports can implement.
slender iron
From a CP development perspective it may be useful to have both a button press a...
That would be ok. We didn't have settings.toml when I added it. Pico W could use the reset button if it had one. The green led can blink quickly like a neopixel would.
mortal kernel
I don't think so. I probably just got lazy. Ideally it'd be functions defined in...
Thank you! I'll stick with what I have now (it follows the nrf port's ble_drv pattern) and revisit it later to move the rp2 and nrf ports to shared-bindings for bonding flash and BLE events.
proven garnet
@lone axle do you want me to merge and release those PRs related to unpinning sphinx and adding sphinx-rtd-theme for RTD? I resurfaced now that the semester is over and am catching up on things.
onyx hinge
hi @proven garnet ! good to see you
hollow gazelle
my MEMENTO pycamera arrived
the installed firmware displays images from the camera on the tft display
double clicking reset brings up a
PyCamera
0.16.0-27-g3d79e6c
firmware.uf2 -> CAMERABOOT
circuitpython.org
screen, but the computer (windows 11) doesn't show a CAMERABOOT drive - so that seems out to be able to download new circuitpython firmware
I also tried to follow the instructions on
https://learn.adafruit.com/adafruit-memento-camera-board/install-circuitpython
and made further progress after switching from an external hub to not using the hub and directly connect to a port on the computer
( thanks for the note about avoiding the hub from the boot loader, I'll do some more research and get back here )
proven garnet
hi <@135415127879974912> ! good to see you
Hello! It's good to be back! Luckily I didn't see my shadow, so I'm out of my groundhog hole earlier than expected 
lone axle
<@382939733107408897> do you want me to merge and release those PRs related to u...
That would be great if you get a moment. I think there are only a few remaining open ones iirc. Nice to see you!
proven garnet
You too! I think I got the last remaining handful!
hollow gazelle
my MEMENTO pycamera arrived
the installed firmware displays images from the came...
well, I now have
Adafruit CircuitPython 9.0.0-alpha.6 on 2023-12-12; Adafruit Camera with ESP32S3
Board ID:adafruit_esp32s3_camera
installed, and am going through the https://learn.adafruit.com/adafruit-memento-camera-board
there seem to be a lot of references to board.LED, (probably generic instructions )
There is a warning that says
The KB2040, QT Py and the Trinkeys do not have a built-in little red LED! There is an addressable RGB NeoPixel LED. The above example will NOT work on the KB2040, QT Py or the Trinkeys!
should this also mention the MEMENTO ?
at least there is a link to the NeoPixel blink example!
If you're using a KB2040, QT Py or a Trinkey, please download the NeoPixel blink example. https://github.com/adafruit/Adafruit_Learning_System_Guides/blob/master/Welcome_to_CircuitPython/code.py
empty salmon
in the current code for raspberrypi boards in common_hal/digitalio/DigitalInOut.c the code appears to set output pins to 12mA using PADS_BANK0_GPIO0_DRIVE_VALUE_12MA.
In the Issues, there appeared to be some debate.
Is this intended to persist or might it change?
In my board_init() I need to set one pip as OUTPUT HIGH and it needs to be capable of the 12mA max (although I don't anticipate it will ever be required to deliver that much)..
stuck elbow
I think we set max driving strength on all boards by default
there are some bugs where we don't, that should be fixed
empty salmon
thanks
empty salmon
Thanks @tulip sleet. Those were the “issue” discussions I was referring to in my original question.
It appears the core will continue to set the output to the max allowed on the raspberrypi boards.
tulip sleet
we are talking about how to make setting drive strength more flexible. It may involve some fundamental rethinking of DigitalInOut and Pin.
in general we would like max drive strength to be the default, because it is the most useful for the novice user
empty salmon
Well, I’m taking advantage of the pin being max drive strength now 😉
short tendon
Asking if anyone is willing to put eyes on the ConnectionManager pull request here: https://github.com/justmobilize/Adafruit_CircuitPython_ConnectionManager/pull/1. Trying to see if this is a direction folks want to go.
tulip sleet
Asking if anyone is willing to put eyes on the `ConnectionManager` pull request ...
I'm sure people will look at it. It's a quiet week and you might bring it up again next week, say in the Tuesday (not Monday due to New Year's Day) meeting.
short tendon
I'm sure people will look at it. It's a quiet week and you might bring it up aga...
Sounds good. I did add a comment to the meeting notes and plan to attend
empty salmon
What extra steps do I need to do when setting a GPIO pin in my board_init() such that the pin is still available to the end user to change in the code.py?
I copied the code from the Adafruit RP2040 USB Host board.c:
digitalio_digitalinout_obj_t _host_power;
void board_init(void) {
common_hal_digitalio_digitalinout_construct(&_host_power, &pin_GPIO18);
common_hal_digitalio_digitalinout_never_reset(&_host_power);
common_hal_digitalio_digitalinout_switch_to_output(&_host_power, true, DRIVE_MODE_PUSH_PULL);
...
}
Is my problem the never_reset() call or something else?
tulip sleet
What extra steps do I need to do when setting a GPIO pin in my `board_init()` su...
so you want board.HOST_POWER or something like that?
empty salmon
I want to have a pin which I set as output HIGH to start. the user may need to change it to LOW (and or HIGH) from their circuitpython code.
in my usecase its an "enable" pin
tulip sleet
set the pin to high without using common_hal_digitalio routines. Just use the pico-sdk GPIO routines
manic glacierBOT
empty salmon
thanks @danh. I will see if I can find an example as to what I need to include and what calls I will make.
tulip sleet
Take a look at some other boards that setup various pins in advance -- there are a numbe of such cases
empty salmon
OK. all of the examples I found were using the common_hal_digitalio_*() calls. I will dig deaper.
tulip sleet
OK. all of the examples I found were using the `common_hal_digitalio_*()` calls....
did they make the pin available?
or was it hidden?
calling the common_hal routines claims the pin
empty salmon
no. all of the examples I have found within the CP code is very simialr to the adafruit RP2040 USB board.
I'll dig
mortal kernel
no. all of the examples I have found within the CP code is very simialr to the a...
If you're using 9.0.0 alpha 6 you can manipulate pin drive strength by accessing RP2040 registers directly. Here is an example that does just that: https://docs.circuitpython.org/en/latest/shared-bindings/memorymap/index.html
empty salmon
thanks @mortal kernel - I want to set the pin from the firmware before the user sees it so it has a smart default state
tulip sleet
thanks <@849815081251766292> - I want to set the pin from the firmware *before* ...
ports/raspberrypi/boards/lilygo_t_display_rp2040/board.c is an example of initializing a backlight pin in advance using the pico-sdk GPIO HAL
main furnace
About to do some broadcom testing on Pi zero W and Pi zero 2W. Have heard reports that the CIRCUITPY drive isn't mounting for > 8.0.5. I'll be trying latest 9x builds first.
empty salmon
thanks @tulip sleet. I was close. needed the code from DigitalInOut.c to set the output strength and needed to not do the never_reset()
const uint8_t pin = gp_pin->number; // the VCC enable pin
gpio_init(pin);
gpio_set_dir(pin, GPIO_OUT);
hw_write_masked(&padsbank0_hw->io[pin],
PADS_BANK0_GPIO0_DRIVE_VALUE_12MA << PADS_BANK0_GPIO0_DRIVE_LSB,
PADS_BANK0_GPIO0_DRIVE_BITS);
// Turn on "strong" pin driving (more current available).
gpio_put(pin, 1);
// common_hal_never_reset_pin(gp_pin);
empty salmon
@tulip sleet - thanks again for the pointers and replies on my GPIO questions. The results are working exactly as I had hoped.
This resolved the prior discussion of “REPL output to an integrated ePaper display”.
The new code means the default behavior will be “just like every other board” and the user has the option to enable/disable the ePaper display as often as they like with the default behavior being the display is active.
lone axle
How is the UID that gets put into boot_out.txt generated? I've never paid close attention, does it stay the same across formatting the storage? Is there anything a typical user might run into that would cause it to change on their device?
spare jacinth
How is the UID that gets put into boot_out.txt generated? I've never paid close ...
it havent looked at it either, but would expect it to be some kind some serial numbering on the actual hardware (at least on the ports that do have something like that), so a read-only value
spare jacinth
How is the UID that gets put into boot_out.txt generated? I've never paid close ...
im lazy enough to not create the equivalent link to github ```c
// Write version info
mp_printf(&mp_plat_print, "%s\nBoard ID:%s\n", MICROPY_FULL_VERSION_INFO, CIRCUITPY_BOARD_ID);
#if CIRCUITPY_MICROCONTROLLER && COMMON_HAL_MCU_PROCESSOR_UID_LENGTH > 0
uint8_t raw_id[COMMON_HAL_MCU_PROCESSOR_UID_LENGTH];
common_hal_mcu_processor_get_uid(raw_id);
mp_cprintf(&mp_plat_print, MP_ERROR_TEXT("UID:"));
for (size_t i = 0; i < COMMON_HAL_MCU_PROCESSOR_UID_LENGTH; i++) {
mp_cprintf(&mp_plat_print, MP_ERROR_TEXT("%02X"), raw_id[i]);
}
mp_printf(&mp_plat_print, "\n");
port_boot_info();
#endif
eg, for RP2040 it pretty much just calls (imma guess Pico-SDK function that does the internal stuff to read it) function: ```c
void common_hal_mcu_processor_get_uid(uint8_t raw_id[]) {
pico_unique_board_id_t retrieved_id;
pico_get_unique_board_id(&retrieved_id);
memcpy(raw_id, retrieved_id.id, COMMON_HAL_MCU_PROCESSOR_UID_LENGTH);
}
so it looks like this populates it common_hal_mcu_processor_get_uid(raw_id);
okay, cool. Thank you. I did just test storage.erase_filesystem() and it does not result in the UID changing. I'm going to validate a few other scenarios but it looks like it can serve the purpose I have in mind.
spare jacinth
from a quick look, it has to be defined (ie there's no defined(COMMON_HAL_MCU_PROCESSOR_UID_LENGTH)) and no port defines it to 0 (or negative) so every port does (currently) implement it
welp, STM port apparently only covers F4, so probably you get a no-op for other families
manic glacierBOT
manic glacierBOT
manic glacierBOT
I was using the bottom of the RGB Matrix Featherwing pinout in the learn guide as a reference, except then it becomes backwards and upside down.
Seems as though I got the rows flip-flopped. That shouldn't work at all but does actually work. After looking at the physical pinout on the matrix panel can see that there's something obviously incorrect with my wiring...
short tendon
Opened a pull request for a header bug in Adafruit_CircuitPython_Requests: https://github.com/adafruit/Adafruit_CircuitPython_Requests/pull/149
manic glacierBOT
manic glacierBOT
That's how it should look like and it is intended to be used for (thats why a higher refresh rate is needed).
pulsar sleet
I am adding the proto arg to socketpool.socket and I want to know what oonstants should I add for IPPROTO_.
manic glacierBOT
Almost there! Flawless on 2 boards already 😁 And I started to read the 300+ page of the ST7789V documentation ...
First a picture that shows the strange bottom line with pixel noise:
So I followed your suggestion and changed the MADCTL register to 0b00010000 (as 0x10) and changed the rotation in the board.c to zero. It didn't had the intended effect:
turbid radish
There will be a New Years Python on Microcontrollers Newsletter - be sure you're signed up at https://adafruitdaily.com/
digital shoreBOT
adafruit
manic glacierBOT
manic glacierBOT
manic glacierBOT
atomic summit
Advanced happy new year to everyone here (I live in the future). Thanks all for the amazing work on CP and excellent support to the community in 2023.
empty salmon
list of preferred pin names?
Is there a list of preferred or recommended names to use for pin on boards with I2C, SPI, I2S, UART, analog pins, digital pins, and buses?
tulip sleet
**list of preferred pin names?**
Is there a list of preferred or recommended nam...
It depends on the form factor, and the tradition with the particular chip family. Is this Espressif or RP2040, and is it an existing shape factor? (I think not for the latter.)
empty salmon
@tulip sleet - unique form factor. RP2040.
tulip sleet
then I would suggest using GPnn, and taking other cues from the Pi Pico board def
empty salmon
I thought the PICO use GPnn ?
tulip sleet
empty salmon
Ideally, I’d like the pin labels to align with most of the Adafruit tutorials.
tulip sleet
yes sorry
spare jacinth
Imo something like "SPI" (ie default config for the bus) doesnt make sense unless you have some special connector, because pretty much any set of pins can output SPI (same thing with I2C)
empty salmon
For the generic pin labels, I am using the GPnn format.
tulip sleet
it's up to you how you are doing the silkscreen on the board. We recommend providing pin names that match the silk. You can also add pin aliases that make sense
empty salmon
The RP2040 does have pins for I2C, UART, etc. I assume, when exposing them I could use both labels (if space allows)
OK. I’ll look thru a bunch of boards to get a consensus on label names.
tulip sleet
The RP2040 does have pins for I2C, UART, etc. I assume, when exposing them I cou...
well, it's quite flexible, you can use I2C0 on multiple pairs, I2C on other multiple pairs, etc. Too many to come up with pin names.
empty salmon
e.g. for SPI, using MISO, MOSI, SCK, etc and for I2C using SCL and SDA. For power enable, I currently use *_EN corresponding to whatever it is powering.
tulip sleet
For a Feather, a Qt Py, a Metro shape, etc. yes, there are standard locations for those pins, so we label SCL and SDA, etc. For your board, there is no standard, unless you want to make one. If you have a STEMMA or Grove connector, certainly yes
empty salmon
I just wasn’t sure if there was a recommendation list somewhere. Sounds like the answer is “no”.
tulip sleet
use similar boards as a guide, since we have reviewed those choices
empty salmon
It does have a STEMMA QT connector.
tulip sleet
so you could label those pins as SCL and SDA, unless you ALSO want other pins to be designated for non-STEMMA I2C
empty salmon
It also has the analog pins so I will alias those as well.
All I2C are on the same pins so I will use SCL and SDA
stuck elbow
Advanced happy new year to everyone here (I live in the future). Thanks all for ...
tell us, how is it? any improvements?
atomic summit
tell us, how is it? any improvements?
2024 is a little warmer than 2023 was 😉
stuck elbow
oh no
orchid basinBOT
manic glacierBOT
manic glacierBOT
onyx hinge
Howdy folks & happy new year. Remember that there's no meeting today -- instead, it'll be tomorrow (about 25 hours from now). I'm looking forward to hearing what everybody's been up to! <@&356864093652516868>
slender iron
Happy new year everyone! Here is the kickoff post for #CircuitPython2024. I set the deadline at the end of the month because I'll be out the next two weeks to visit family. I'll try to blog up folks' posts though while I'm there. https://blog.adafruit.com/2024/01/01/circuitpython-in-2024-circuitpython2024-python-circuitpython/
manic glacierBOT
short tendon
Random question on .mpy files and typing. I've noticed (although not dug in too deep) that when typing information is defined in .py files, that it takes up space in the .mpy file. Is it possible to make that not happen?
manic glacierBOT
blissful pollen
Random question on `.mpy` files and `typing`. I've noticed (although not dug in ...
If you haven't looked before there is the mpy-cross tool that builds the mpy files. I quickly glanced and did not see a flag to not include typing information. But you could remove the typing information from the .py files and recompile the mpys.
onyx hinge
Random question on `.mpy` files and `typing`. I've noticed (although not dug in ...
The import and try/except will unfortunately take up space no matter what. The annotations themselves do not take up space. I have experimented with re-writing Python code automatically before mpy-compiling to recognize common idioms , but never came up with anything good enough to ship. https://gist.github.com/jepler/9ae2dbe8fda6ee5ff9e9360efe3bc157 #circuitpython-dev message
We've also discussed various ways to minimize the extra mpy space but I don't know that we've written up a best practices guide. Folks just kind of do whatever they've seen somewhere else.
I think the ideal solution IS some kind of ast-manipulation one PLUS a best-practices document so people can get a clear statement of how to ensure their type annotations have as little overhead in CircuitPython as possible.
onyx hinge
@lone axle @gilded cradle let me know if you can read your usual sections in the meeing today.
lone axle
<@382939733107408897> <@478616919721771019> let me know if you can read your usu...
I cannot. Unfortunately I've had a different meeting for work that conflicts come up. Likely won't be able to make today's meeting.
onyx hinge
@lone axle that's exactly why I checked. No worries.
If any other meeting regulars would like to take a turn reading the library update let me know! otherwise I've got it covered.
gilded cradle
<@382939733107408897> <@478616919721771019> let me know if you can read your usu...
Yep, I should be able to
onyx hinge
thanks Melissa
thorny jay
Hi, no CP activity since the last meeting, so I will not write anything in the notes.
This would have been my busy "Non CP" section:
- Turning a Pi4 into a Home Assistant server
- Connecting the smart doorbell (ESP HOME)
- Reading power consumption from the electricity smart meter (P1 interface) connected to an ESP HOME
- Reading temperature/humidity from 4 LYWSD03MMC (flashed) using Bluetooth relay from ESP HOME
- Setting up a Pi5 + Pineberry PCIe bottom + SSD + Plex Server
- Growing my UniFi network with manageable switches and a VPN to France with dedicated Vlan/Wifi
- Got a PowKiddy RGB30 for Xmas and I installed my Pico8 licence to have a physical "virtual" console
(don't hesitate to ping me, maybe DM, if you have questions about any of those)
Have a good 2024.
onyx hinge
<@&356864093652516868> meeting in about 8 minutes, sorry for the late notice.
blissful pollen
Lurking today, no notes
orchid basinBOT
silver tapir
This is inaturalist
https://www.inaturalist.org
This is the library
https://github.com/fede2cr/cpyinaturalist
I'll certainly write about cpy24, but wifi with bt on esp32 families that have it, which is already being worked on.
OTA updates with uf2 might be very very cool as well. (or at least an update via the cpy site installer that works via wifi)
Thank you for hosting jepler.
gilded cradle
Thanks, have a good week too
onyx hinge
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1Z2eoPQUUiMq5NK6sINJRZ-byxr5CSbZQm2u0cEehjkU/edit?usp=sharing
proven garnet
The `import` and `try/except` will unfortunately take up space no matter what. T...
What about converting the types into stub files that can be included in the PyPI distribution? We could use mypy still (if that is still the future intent) and use stubtest in the CI to make sure they don't deviate over time.
short tendon
The `import` and `try/except` will unfortunately take up space no matter what. T...
So this block in Adafruit_CircuitPython_Requests: https://github.com/adafruit/Adafruit_CircuitPython_Requests/blob/4754bf3e5fdb559be9ea085bfb65bd435039e887/adafruit_requests.py#L44 takes up about 1kb in the mpy file. Not a ton, but it's still some...
orchid basinBOT
manic glacierBOT
pseudo idol
I'm having a problem where 4 different circuitpython boards (ESP32C3 by Seeed) used to perform flawlessly, but have recently stopped responding to the web workflow, and even to pings to their ip addresses. I would think they were dead, but they continue to post sensor data to an MQTT server. I'm now trying to understand the network services that live under the REPL and handle stuff like MDNS replies, web workflow, etc. Can I get a pointer to the circuitpython repo area that does that bit of service code?
lone axle
I'm having a problem where 4 different circuitpython boards (ESP32C3 by Seeed) u...
MDNS I am not sure about, but for web workflow and the REPL websocket I believe the code lives here: https://github.com/adafruit/circuitpython/tree/main/supervisor/shared/web_workflow
slender iron
I'm having a problem where 4 different circuitpython boards (ESP32C3 by Seeed) u...
what version of CP are you running? there is a bug fixed in 9.x where a reload at the wrong time would break web workflow until a reset
pseudo idol
@lone axle Thanks, I'll check. @slender iron 8.2.8 and 8.2.9. But this issue survives a power-cycle and other resets.
slender iron
oh! surviving a power cycle is weird!
pseudo idol
Tell me about it. 🙂
I thought maybe my code was interfering, so I cleared flash and installed 8.2.9 from scratch. Only patched up settings.toml to get on wifi, and it still fails.
Does the bootloader get involved? I've checked my network. I can see the MDNS request go out from my browser, and other MDNS replies from the printer and Tivo come back, so the multicast appears to be working.
slender iron
Does the bootloader get involved? I've checked my network. I can see the MDNS ...
the bootloader shouldn't matter. can you access it if you put the IP into the browser?
is it just MDNS that is broken?
pseudo idol
That worked for a while, at least as far as asking for log-in info.
But no longer. No ping response either given a numeric IP.
I don't understand the web workflow well, but circuitpython.local seems to make the MDNS request, which is unfulfilled
slender iron
hmm, can you get serial output over usb?
correct .local is mdns
ip should work too if mdns doesn't
pseudo idol
that takes me to cpy-<mac addr>.local which was intermittant (when I could get there.) If I go direct, that also fails the MDNS request.
I can get to serial over usb cable. I'm unfortunately becoming adept at writing file read and write utilities over the REPL
slender iron
that's good. you could build your own CP and add print statements in then
do you see the ip in the serial terminal title bar?
pseudo idol
yes
slender iron
but it doesn't respond when you visit?
pseudo idol
nope. ping <numeric ip> doesn't respond either. (should it? I thought it did in the past....)
slender iron
I'm not sure ping will work
http should
you can add debug prints into the code and they'll go out the serial link
pseudo idol
I misspoke: ping does work if you give it long enough. 9 fails, then reliable
slender iron
hrm
pseudo idol
and now http://ip brought me to the welcome page. full code editor link takes me to http://ip/code/ and wi-fi dialog again takes me to cuitpython.local/code/ which doesn't respond
is "http://circuitpython.local/code/#state={"path"%3Anull%2C"contents"%3A""%2C"pos"%3A0}" meaningful to you?
pseudo idol
I'm wiresharking the network, and I do see MDNS traffic from other units on the network, but no response for circuitpython.local.
which implies to me that it's not a firewall issue on my computer
slender iron
hmm, all cp devices should respond to circuitpython.local
pseudo idol
agreed.
sick creek
yeah that would be base to respond for sure
pseudo idol
in web_workflow.c, I see CIRCUITPY_WEB_INSTANCE_NAME is used in MDNS
slender iron
I've gotta run. I think you are looking in the right portion of code
sick creek
later tannewt
slender iron
ports/espressif/common-hal/mdns will have mdns calls into the ESP-IDF too
a debug build make BOARD=foo DEBUG=1 will turn on IDF logging which could help too
may go out a different uart from the one on the internal converter
pseudo idol
If I set that to anything other than it's cpy-<mac> (which I have) could that be it?
slender iron
ya perhaps. I haven't tested it much with overriding from user code
pseudo idol
@slender iron Thanks for the pointers so far. I'm trying to avoid learning to build cpy from scratch, but if a cleaned up settings.toml doesn't do it, I can see a "fun" evening ahead.
slender iron
building is pretty well documented here: https://learn.adafruit.com/building-circuitpython
sick creek
it could interesting how this thing will evolve with this new wit-bindgen standardization will change things and evolve as document for that https://github.com/bytecodealliance/wit-bindgen
pulsar sleet
where does the matrixportal s3 put debug data when building with DEBUG=1
manic glacierBOT
stuck elbow
where does the matrixportal s3 put debug data when building with DEBUG=1
in the binary?
pulsar sleet
The printed out data. Does it show on the uart to usb converter or do I need to connected up a uart reader on different pins
stuck elbow
The printed out data. Does it show on the uart to usb converter or do I need to ...
what printed out data? DEBUG=1 enables debugging symbols in the binary, so that when you connect with gdb you can see the variable and function names
lone axle
The printed out data. Does it show on the uart to usb converter or do I need to ...
I think it comes over this pin. I've not used it on a matrix portal, but on Feathers you can use debug pin with DEBUG=1 builds to get encoded tracebacks when the core crashes. There is a script in the core repo that can decode them. I've seen it spit out some other information during boot up and things but not sure what all it was.
I read it with a USB serial adapter
pulsar sleet
I got a KB2040 I got for free because I spent a certant about on money and I use that as a uart to usb converter (just printing uart input)
where is the script to debug the DEBUG pin. I tried to find it online but no luck.
lone axle
where is the script to debug the `DEBUG` pin. I tried to find it online but no l...
I think <board> argument should be the same name used when making a build for that device. It's been a while since I used it though I could be misremembering. Then you paste the hex or whatever was output on the debug pin into the prompt from this script.
manic glacierBOT
pulsar sleet
I think `<board>` argument should be the same name used when making a build for ...
I copied the output of the debug pin (from putty) and it just gave me
0x00000000: ?? ??:0 do I need hex
I think I am doing something wrong becuse in putty I am getting
0�ßÕ· !1H�¤&9�¥�§�¶)!)§�í¬��!�!)���!�í�§��!�§))))�í�)!äÿ�1¤)9)=1!)©���
maybe wrong baud rate
lone axle
yeah, it should be legible characters I'm pretty sure.
pulsar sleet
looking online the bause rate should be 115200bps
running a debug build with code that hard crashes (looks like it) doesn't go into safe mode. The circuitpy drive isn't mounting and there isn't a port to connect to. Connecting to the uart with a external device shows just the crash. It looks like it just doesn't reset after crashing
actually it did connect but I can't get into safe mode by pressing the reset butting while the neopixel is yellow.
pulsar sleet
actually it did connect but I can't get into safe mode by pressing the reset but...
Pressing the boot button worked
crimson ferry
@pulsar sleet sometimes the trace you get on the debug console is corrupted, but if it isn't you can decode it using ports/espressif/tools/decode_backtrace.py
short tendon
@lone axle sorry for the direct ping, but trying to learn the flow. You reviewed and tested my requests header fix PR, is there a next step?
lone axle
<@382939733107408897> sorry for the direct ping, but trying to learn the flow. Y...
No worries. Often I will merge it when I approve a PR. In this case I didn't because I didn't look over the requested changes specifically. I intend to circle back to it this week if no one beats me to it.
short tendon
Cool. Thank you
vital dune
Hello folks. Newbie question. I am trying to install CircuitPython on a particular ESP32-S3 module. Is there an easy way to see if a build is premade for a specific variant?
gusty siren
This might not be the right place, but I was hoping someone on the adafruit team could take a look at https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT/pull/184 and https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT/issues/148 ?
manic glacierBOT
crimson ferry
@vital dune https://circuitpython.org/downloads
If your particular variant is not there, you can make a PR for it (in some cases, there may be a board that's close enough: ram size, flash chip type and size, etc)
manic glacierBOT
proven garnet
@lone axle the work you did on RTD, did that include updating the web hook? I just got an email saying that those need to be updated on some of the projects I created on RTD.
Apologies if I'm waaaaay behind.
lone axle
<@382939733107408897> the work you did on RTD, did that include updating the web...
I don't think so, but I'm not certain I know which part you mean. Is that something you do inside the settings page for the repo? Either in Github or on the RTD side? If so I don't think I did anything on that front (don't believe I have access)
spare jacinth
just saw #7305 and thought it would be interesting to dive into it
in order to achieve it, im looking for the piece of code that creates the new class at C level after parsing Py's source (still havent found), during my little trip across the code i've seen mp_obj_is_native_type but im not sure if that would be the best function to use (maybe some checking needs to be crafted into a new function instead)
also, my idea so far is to emit a Warning when code tries to inherit from a native class. ie: not actually crash the program but let the user know it may be a bad idea.
- reason not to actually
raisebeing that some native classes (or at least some particular inherits) might be safe (?) .... and i dont know (maybe just lack of knowledge about the codebase) it there's a good way in which classes could note "hey, im fine being inherited from" (in which case we could default toExceptionfor all types, but emit a Warning for the "safe" classes, probably) - downside being that crashes would not be prevented without a
raise🙃
proven garnet
I don't think so, but I'm not certain I know which part you mean. Is that someth...
I think it's technically an RTD thing that they want to be done in GitHub, if that makes sense. I think it's basically regenerating the webhook stored in GitHub repos, but I don't have permissions for that on the GitHub side. There is a "Resync webhooks" button on the RTD site you can use but again I don't have permissions to alter webhooks in repos.
manic glacierBOT
lone axle
I think it's technically an RTD thing that they want to be done in GitHub, if th...
I don't believe that I do either. Make a note to discuss it during a meeting if you can?
mortal iris
Hello folks. Newbie question. I am trying to install CircuitPython on a partic...
what is the board? Recommend asking in #help-with-circuitpython first, and sometimes the manufacturer of the board has compiled a circuit python bin specifically for the board, which isn't listed at https://circuitpython.org/downloads?mcufamilies=esp32s3
proven garnet
I don't believe that I do either. Make a note to discuss it during a meeting if ...
Will do!
knotty trail
If I wanted to add a board, that had a library specific for that board, would I put the library in the main circuitpython project? Or would I fork and compile it and create a frozen module?
It’s a C driver for an RP2040 that is specific to the board I made. It also has a python part that is higher level stuff, but relies on the C driver
Just not sure if it makes sense to put a board specific driver in the circuit python repository
knotty trail
Some more context: I made a RP2040 based board with a MCP2515 CAN controller on it. I know that Adafruit has (basically) the same combo, but my board has headers/features specific for my application.
I am not really satisfied with the Adafruit MCP2515 CAN driver because it uses blocking calls for CAN, and waits until a message is received or timeout to return the message. This is inconvenient because it prevents other timing-critical tasks like driving NeoPixels, displays, etc. What my driver does is dedicates the RP2040's second core to managing CAN, and stores every message in an array, and updates the array with the most recent message. It also has a queue for transmitting messages, creates a system for messages to repeat (helpful in automotive applications), and other small features helpful for me. The circuit python side (still in development) basically just communicates with the lower level CAN driver in a way that doesn't use blocking calls. I plan on using it for robotics and for automotive uses, so I will have features in the CircuitPython relevant to that as well.
Unfortunately, my CAN driver is specific to the RP2040 and MCP2515, and really my board's wiring, so it wouldn't necessarily work on the RP2040 Feather CAN (I haven't checked though), and it wouldn't be applicable for any other boards.
With that being said, I am wondering where the best place to put the library is. I am not sure if it's wise to put it in the circuitpython repo because it's so specific, but I also want to be able to sell this board in the future, and the requirement for my board to be able to use CircuitPython (https://github.com/adafruit/circuitpython#branding), it has to be supported by the primary CircuitPython branch. Plus, I want to be able to contribute to it if/when I do more development
onyx hinge
@knotty trail we don't currently have a solution for shipping native code routines that are not "in the core".
so you would hypothetically add your routines in shared-bindings and shared-modules (or ports/raspberry-pi/common-hal if it relies on rpi specific hardware features) + a new frozen library
that said, we are unlikely to want to offer multiple incompatible APIs for canbus within the core. is there any way to evolve the current API so that it is more suited to your use cases? mcp2515 is pure python code outside the core, while we have canio for some mcus with built-in can peripherals.
knotty trail
@onyx hinge I was looking at evolving the existing canio interface but my implementation isn’t as nice as the existing canio interface. Also, it is somewhat specific for the rp2040. Maybe I could create an rp2040 specific canio interface but I’m not sure if that will work for all projects since it uses the second core to constantly monitor the MCP2515 and that could affect other libraries.
I will look further into evolving the canio module later. I think for now, the best option may be to just fork circuit python and list my board as CircuitPython-compatible. That kind of sucks because I’d like for it to be listed on the circuit python site, but it is what it is
onyx hinge
can you say a little bit more about how it's specific to rp2040, when the mcp2515 is just a SPI connected device?
oh the second core, hmm
knotty trail
oh the second core, hmm
Yes. Basically all the second core does is poll the MCP2515 as fast as possible. I tried using interrupts but after trying to read/store over 1000 messages/sec, it locks up.
With continuous polling, I’m able to process over 8000 messages per second, which is basically a 1MB/s CAN bus near 100% utilization. I haven’t tried more than that but I’m sure it can handle it.
I also have a message transmit queue so you can queue up to 32 messages to be transmitted instead of only 3 (tx buffers). I also save the count and timestamp of each message. I also have a mechanism to send messages every x milliseconds, even if the data wasn’t updated. This is useful for vehicles or devices that require a constant stream of messages and would set a fault/dtc if it didn’t receive a message on time.
What’s also nice is that since it uses the other core, the first core doesn’t have to do any busy waiting. This use useful if you have any neopixel animations or other sensors you are polling
It’s really useful for gateway-ing messages, Robotics, and I even have plans to use it on a solar hybrid inverter
The limitations on the python MCP2515 implementation is that it doesn’t save every message, it will wait until a message is received before returning, which can mess up neopixel animations and other sensor reading routines, and it has no method to continuously send CAN frames every x milliseconds
short tendon
Pre-planning. Does anyone know if the right people will be in this coming up Monday meeting to talk about ConnectionManager?
short tendon
Also, if this gets accepted, if a board has frozen requests, this would also need to be frozen in, yes?
onyx hinge
yes about the freezing. I am not sure when Dan and Scott are back.
short tendon
Thanks. If it's given the go ahead, I'll open ALL the pull requests...
slender iron
yes about the freezing. I am not sure when Dan and Scott are back.
I’m back the 16th
tulip sleet
i just got back
slender iron
Yes. Basically all the second core does is poll the MCP2515 as fast as possible....
I’d suggest opening an issue so we can have a longer term discussion. One apiece I’ve thought about is a coprocessor api so you can run whatever code you want on the second core. Sharing peripherals and pins will be a bit tricky though. You can also have port specific bindings (instead of shared bindings) that may be more relevant too.
short tendon
Should I then wait until the 22nd (the first Monday both are back)?
bright robin
Is it possible to use 16-bit commands with DisplayIO's API? I'm trying to write a board.c file, but it seems like common_hal_displayio_epaperdisplay_construct is hardcoded to use one byte commands, then immediately send the data.
I'm thinking maybe I can setup everything in the display_start_sequence and then just set all the commands to NO_COMMAND ?
Actually, sequences work fairly well, it would be great if all the command parameters just used a sequence rather than a byte.
(I might trying making a pull request for that...)
manic glacierBOT
lone sandalBOT
manic glacierBOT
gaunt bolt
I would like to suggest adding a method to the module "wifi.Radio" for getting information about connected devices when the board configured as an access point (i haven't found a way to get this info). Should I write it on Github?
crimson ferry
@gaunt bolt yes, you can create an issue on the circuitpython repo
tame creek
wow that formatting came out weirdly
manic glacierBOT
empty salmon
@slender iron - I recall an open issue regarding the use of the second core on dual-core MCUs. The options quickly became untenable.
I like the idea of having a CircuitPython core level capability since the second core is a hardware capability, having the board’s CircuitPython firmware has appeal.
For my board, I struggle to see what makes sense to go on the second core because everything has “end-user” applications - e.g. neopixels, touch sensors, and SPI display, audio (often used with synthio).
To take advantage of any of these, a portion of the existing core code would need to be shunted over to the second core; then the user calls would need to be routed according.
Like before, things get complex quickly.
ornate cave
night canyon
#general-chat Itt megy a beszéd itt fejlesztés/ hibabejelentés megy
(sorry)
manic glacierBOT
knotty trail
I’d suggest opening an issue so we can have a longer term discussion. One apiece...
I dug deeper and the Pico DVI board/library is a similar situation. Super custom code running on core1
manic glacierBOT
slender iron
I dug deeper and the Pico DVI board/library is a similar situation. Super custom...
USB host as well
manic glacierBOT
lone sandalBOT
marble hornet
Hi all 😁👋 it's been a hot sec
I've been encountered a build problem (maybe🤷♂️🧐) on main* and, if someone has a few minutes, I wonder if it's worth a sanity check?
(typing...)
Oop! never-mind it looks like it's because there was a space in a directory that was part of the path to circuitpython 😅 .... 🤦♂️
I'll see if it's easy to add a check or fix to makeversionhdr.py (where the error occurred) hopefully before the meeting.
manic glacierBOT
empty salmon
Which arm-none-eabi?
This may be an obscure configuration but here it is: Using VMWare Fusion for Ubuntu guest (ARM64) on MacOS host (Apple Silicon). I am not sure which toolchain to download.
tulip sleet
**Which `arm-none-eabi`?**
This may be an obscure configuration but here it is: ...
https://learn.adafruit.com/building-circuitpython/linux#install-build-tools-on-ubuntu-2986713 points to https://developer.arm.com/downloads/-/arm-gnu-toolchain-downloads for the 13.2.Rel1 release. From there find the "macOS (Apple silicon) hosted cross toolchains" "AArch32 bare-metal target (arm-none-eabi)"
The host is macOS; the target is AArch32 bare metal (arm-none-eabi)
tulip sleet
this may also be in homebrew, but I don't know
mortal kernel
(That's for running the toolchain on your Ubuntu guest)
tulip sleet
whoops sorry, yes, bob is right, I was looking at the macOS host, not Ubuntu guest on VMWare
empty salmon
I tried arm64-arm-none-eabi.tar.xz and I cen an exec format error cannot execute binary
So perhaps the aarch64-arm then?
tulip sleet
mortal kernel
I tried arm64-arm-none-eabi.tar.xz and I cen an exec format error cannot execute...
The ARM site is a touch confusing. You need to scroll down to the AArch64 Linux hosted cross toolchains section.
empty salmon
Thanks. We are all landing on the same conclusion concurrently 👍🏼
empty salmon
Irony is the bold category labels are a slightly smaller font from the sub-category labels 🫣
we have a winner!
mortal kernel
@danh I'd like to take up reviving the CP fork of the CYW43-driver.
empty salmon
The aarch64-arm is what I needed
(appreciation to @tulip sleet and @mortal kernel)
tulip sleet
@danh I'd like to take up reviving the CP fork of the CYW43-driver.
that is fine
mortal kernel
How do we go about it? It needs to be re-synched, branched, and have a couple of local pulls merged. I'd be happy to do it with if you can give me repo privileges.
tulip sleet
How do we go about it? It needs to be re-synched, branched, and have a couple of...
i have sent you a collaboration invitation
candid sun
<@&356864093652516868> We'll have our weekly meeting in about 60 minutes from now in this text channel and in the circuitpython voice channel. Please take the time to add your notes in advance to the document: https://docs.google.com/document/d/1Z2eoPQUUiMq5NK6sINJRZ-byxr5CSbZQm2u0cEehjkU/edit -- I look forward to everyone's updates!
blissful pollen
No notes, just lurking today
onyx hinge
compared to last week this meeting is hopping 🦗
turbid radish
☺️
candid sun
@stuck elbow i think you need to mute
stuck elbow
<@351136009922019331> i think you need to mute
sorry
candid sun
all good!
errant grail
WaveViz in action
onyx hinge
[I keep hearing "wave biz". Hey, I make money in the wave biz.]
blissful pollen
That is really cool @errant grail haven't had a chance but wanted to try more with WaveTables and being able to see it is great
errant grail
One of the symptoms of the bad and misbehaving SD card.
onyx hinge
😦 😦 so many folks with covid right now.
tulip sleet
wastewater COVID results tell the tale. North Boston region:
gilded cradle
I'm open to using something besides threads if it works as well.
tulip sleet
https://biobot.io/data/ for more Covid wastewater data
lone axle
I think the threads work nicely, but if I come up with any other ideas I'll make an issue to propose them. PyGameDisplay had actually used Threads prior to the Blinka Displayio update and I think having both making threads was causing trouble. But the latest working revision I have actually removes the threading from PyGameDisplay and relies on just the threading in Blinka_DisplayIO which makes PyGameDisplay much less complex. Just gotta find and hammer out the last few quirks.
onyx hinge
I'm sure you'll get it!
@proven garnet Adafruit also got a message about a bunch of repos. I tried this with some of the adafruit repos and it did not seem to work; my assumption is that either something's not right at the organization level or my access level is not sufficient.
appreciate that Dan!
fwiw I do have the ability to log into readthedocs as adabot but it didn't work
yes we could do the manual process for them all but it'd be nice not to:)
candid sun
i'm also happy to help if repetitive actions are needed
onyx hinge
That sounds great, somebody just needs to implement it and make it fit on as many devices as possible.
lone axle
The specifics of handling that section of storage is over my head. But the functionality of having that traceback or perhaps other diagnostic info in an easy-to-access file sounds awesome to me.
mortal kernel
The big issue with a logging file/filesystem is host OS filesystem caching. Once a filesystem is mounted, the host OS will not see updates reliably or at all.
short tendon
Would be great to add logging in there too, so we have some breadcrumbs of what happened
mortal kernel
USB events are a possible solution, but I'm unclear whether USB events can be limited to just one filesystem. I believe the event in question forces a remount.
stuck elbow
Would be great to add logging in there too, so we have some breadcrumbs of what ...
that would require using flash storage, and you run into flash wear issues with it
onyx hinge
@tulip sleet fwiw I found that adabot can't do the necessary operation on readthedocs, but if I add myself as a mantainer first [logged into readthedocs as adabot] and then re-sync the webhook [logged into readthedocs as myself] it works. It doesn't matter if I sign in to readthedocs with github or not. maybe we can elevate adabot's permissions on github so that it can administer webhooks?
thank you Liz!
gilded cradle
Thanks everyone
mortal kernel
Thanks all!
short tendon
that would require using flash storage, and you run into flash wear issues with ...
It couldn't use the same virtual?
onyx hinge
Next week's meeting is on the 16th (tuesday)
stuck elbow
It couldn't use the same virtual?
ram doesn't survive powering off
lone axle
Hope everyone has a nice week. Thanks for hosting Liz!
errant grail
Thanks to Liz for being an excellent host and for reading the text-only sections.
wraith crow
Thanks 😄
candid sun
thanks folks!
short tendon
Thanks!
short tendon
ram doesn't survive powering off
How could you store it in boot_out.txt ?
stuck elbow
How could you store it in `boot_out.txt` ?
the boot_out.txt is only updated when it changes, to avoid flash wear
tulip sleet
<@329766224093249548> fwiw I found that *adabot* can't do the necessary operatio...
Thanks - I will see if I can (perhaps temporarily) make adabot be a Maintainer in the adafruit org, click the buttons in readthedocs while logged in, and then switch it back. How can I tell whether it worked or not? Is there a status message in each readthedocs project/subproject?
stuck elbow
and most of the time it doesn't change
short tendon
Ahhh, could the logs be stored in ram, and then only written if there is an exception?
onyx hinge
@tulip sleet it tells you "Webhook activation failed. Make sure you have the necessary permissions." on RTD if it didn't work.
stuck elbow
Ahhh, could the logs be stored in ram, and then only written if there is an exce...
you could do that with python right now
candid sun
Here is the notes document for next Tuesday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1BnXAE4EqNirRCWa8SfC291593gxhsrIrJBOcdCfqEeM/edit?usp=sharing
cold oar
Hey all - is there a document that lists Adafruit's pin name recommendations for common built-in buses, sensors, etc? I think I've tracked down a lot of them, but haven't found a list of recommendations anywhere.
crimson ferry
I don't know that there is a document. This may be a bit outdated, and is specific to Feather: https://learn.adafruit.com/adafruit-feather/feather-specification. This covers sensor units rather than pins: https://docs.circuitpython.org/en/latest/docs/design_guide.html#sensor-properties-and-units. I'd look to the most recent Adafruit boards added, that should get pretty good coverage of pin naming.
manic glacierBOT
proven garnet
I looked into getting rid of inline type annotations, but the only way to provide them per PEP 561 is if the distribution is a package - there is no option for modules. So for any of the modules we would just be stuck with the status quo. There is a draft PEP for this issue that was started a year ago, but no development on it since, it seems.
proven garnet
You can trick Pylance in VS Code into finding it with a hack to data-files in pyproject.toml, but not mypy as easily.
manic glacierBOT
manic glacierBOT
mystic flume
I'm currently looking at the issue of excessive current draw after de-initing the UART. I originally tried to fix this in https://github.com/adafruit/circuitpython/pull/8089, but that broke some fairly major USB functionality, so it was reverted
Looking at this further the nrfx_uarte_uninit function that is in the nrfx library linked to circuitpython was buggy (https://github.com/adafruit/nrfx/tree/3f55e49eb11e6db0da1da09e189bb094222702c9). More recent versions of nrfx have fixed this, and indeed these have been ported through to the most recent adafruit version of nrfx
However, circuitpython still links to commit 3f55e49 of nrfx
I've tested just updating nrfx_uarte.c - and that works fine. However trying to compile circuitpython against the most recent nrfx fails with multiple errors - and I'm concerned that it could break other peripherals.
I suppose the question is whether I should create a branch of nrfx with just the uarte updates - just a handful of lines are changed, and link CP to that, or some other solution?
mystic flume
Looking further at the circuitpython history, the commits after 3f55e49 have never been used in circuitpython
I have now successfully compiled circuitpython against a slightly altered nrfx from the latest commit on that branch - and it seems to work. Not sure what a good testing strategy would be however.
tulip sleet
I suppose the question is whether I should create a branch of nrfx with just the...
Our fork of nrfx already has some changes we needed in it - most notably (and maybe only) the "uninit()" operations are idempotent. So our fork could be brought up to date with upstream and then further changes could be made. Another change we haven't made but need to is to provide a timeout in I2C transactions to prevent hanging when the bus is stuck.
mystic flume
Is there a process for testing these changes?
I'm thinking something like unit testing but with hardware. Or is it a question of asking people to beta test and report failures
tulip sleet
I'm thinking something like unit testing but with hardware. Or is it a question ...
we have no automated testing. We have worked on that in the past but never got an automated test. For this, I would test BLE, busio.*, etc.
mystic flume
I might struggle to test the SPI stuff, but I think my device uses pretty much all the other peripherals
So if my thingy works, likely most of everything else will. I'm just wary having broken USB MSD with a UART fix!
I'll have a play and submit a couple of PRs - one to nrfx and one to CP.
onyx hinge
I've been seeing a lot of safe mode resets when writing to CIRCUITPY today on my pycamera / memento with latest-ish 9.0-pre code.
I have a pile of local changes but none that would affect msc or flash intentionally
tulip sleet
I've been seeing a lot of safe mode resets when writing to CIRCUITPY today on my...
if it's semi-reproducible a bisect might be in order. Also there have been a bunch of fixes since 5.1.2 - I was looking through them while debugging something else
candid sun
i'm currently working on a new library (https://github.com/adafruit/Adafruit_CircuitPython_MCP3421) and stuck on a CI issue. i keep hitting this error:
Run circuitpython-build-bundles --filename_prefix adafruit-circuitpython-mcp3421 --library_location .
Traceback (most recent call last):
...
File "/opt/hostedtoolcache/Python/3.11.7/x64/lib/python3.11/site-packages/circuitpython_build_tools/scripts/build_bundles.py", line 267, in build_bundles
build_bundle(libs, bundle_version, zip_filename, package_folder_prefix,
File "/opt/hostedtoolcache/Python/3.11.7/x64/lib/python3.11/site-packages/circuitpython_build_tools/scripts/build_bundles.py", line 171, in build_bundle
build.library(library_path, build_lib_dir, package_folder_prefix,
File "/opt/hostedtoolcache/Python/3.11.7/x64/lib/python3.11/site-packages/circuitpython_build_tools/build.py", line 301, in library
_munge_to_temp(full_path, temp_file, library_version)
File "/opt/hostedtoolcache/Python/3.11.7/x64/lib/python3.11/site-packages/circuitpython_build_tools/build.py", line 174, in _munge_to_temp
with open(original_path, "r", encoding="utf-8") as original_file:
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
FileNotFoundError: [Errno 2] No such file or directory: '/home/runner/work/Adafruit_CircuitPython_MCP3421/Adafruit_CircuitPython_MCP3421/adafruit_mcp3421.py'
which makes sense because that file does not exist. the library structure is adafruit_mcp3421/mcp3421.py and adafruit_mcp3421/analog_in.py is there a setting that i'm missing somewhere? i'm looking at the CI script to see if there's something in there. the initial commit passed because it had the cookiecutter structure with that default setup. i deleted that workflow run in case maybe something had been cached but that didn't fix it
stuck elbow
i'm currently working on a new library (https://github.com/adafruit/Adafruit_Cir...
I think you need an empty __init__.py in there?
to make that directory a package
onyx hinge
i'm currently working on a new library (https://github.com/adafruit/Adafruit_Cir...
When there's more than one py file usually there is a (possibly empty) file that indicates it's a package: adafruit_mcp3421/__init__.py
candid sun
okay, thank you! i'll add that
spare jacinth
and some path may be wrong on part of the setup too?
no idea how this stuff works, but Adafruit_CircuitPython_MCP3421 appears twice, maybe it is pointing to the folder while it should point at its parent (and the library's name gets added)
candid sun
and some path may be wrong on part of the setup too?
no idea how this stuff work...
yeah that's what i was wondering
same error with the included init.py
onyx hinge
@candid sun researching, I'll have an answer for you in a moment
candid sun
<@347203168121126912> researching, I'll have an answer for you in a moment
thanks!
onyx hinge
@candid sun please try changing pyproject.toml: ```diff
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -41,7 +41,7 @@ dynamic = ["dependencies", "optional-dependencies"]
[tool.setuptools]
TODO: IF LIBRARY FILES ARE A PACKAGE FOLDER,
CHANGE py_modules = ['...'] TO packages = ['...']
-py-modules = ["adafruit_mcp3421"]
+packages = ["adafruit_mcp3421"]
[tool.setuptools.dynamic]
dependencies = {file = ["requirements.txt"]}
ah! i knew it was something like that 😅 thank you!
that did it, thank you @onyx hinge
onyx hinge
ah! i knew it was something like that 😅 thank you!
you're welcome
proven garnet
I played with ruff last night for formatting and linting my circuit python related stuff and it blew my mind how fast it was.
It would require changing the pylint disables to the equivalent style for ruff, but it's speed plus that it combined black, pylint, and for me pydocstyle made it a really good switch. Plus it has a pre-commit hook.
Not sure the libraries would gain much though given pylint and black don't take that much time for something that size, plus it only changes downloading two hooks into one. Plus the changing of linting disables is a pain.
onyx hinge
I have also been using ruff in some personal (non circuitpython) software & like it
ruff also auto-fixes more things
proven garnet
Oh that's awesome that it can automatically handle more stuff
onyx hinge
https://github.com/jepler/wwvbpy is a modest sized project but when I added ruff it knocked pre-commit time way down https://github.com/jepler/wwvbpy/pull/72 from 11s to <1s
~2400 lines or so
trim lion
Unable to upgrade ST-Link, need help
onyx hinge
@trim lion to get official help with hardware purchased from adafruit, use the forum: https://forums.adafruit.com/viewforum.php?f=19 -- since the product is from ST, you can also use their support methods, which I am not familiar with.
This particular channel in discord is for discussing development (coding) of circuitpython.
gilded cradle
Hello @crude blaze I hear you may be back from holiday and that you figured out a good fix for the rotation/offset with the ST7789 display on the pi.
onyx hinge
@trim lion you might also look over these matching threads to see if anything there helps: https://duckduckgo.com/?q=site%3Aforums.adafruit.com+st-link&t=ffab&ia=web
trim lion
thank you @onyx hinge
crude blaze
Hello <@375455651050160130> I hear you may be back from holiday and that you fig...
hi yea I'm back to work. Rotation of the display worked fine with the existing drivers, what was still an issue was that the TSC2007 touch driver did not support rotation. I submitted a PR but it looks like Pi might want us to go through upstream first, I need to follow up on their comment.
gilded cradle
hi yea I'm back to work. Rotation of the display worked fine with the existing d...
Ok, thanks. I'll see if I can get the BrainCraft working using the existing driver.
crude blaze
@gilded cradle cool let me upload the overlay somewhere, should mostly need pin changes
gilded cradle
Ok
crude blaze
@gilded cradle ah nvm forgot I had already put it into a PR:
https://github.com/adafruit/Raspberry-Pi-Installer-Scripts/pull/280/files
it works fine but the value for the touchscreen probably need some tweaking but for just a screen all is working, its only draft because of the TSC stuff
gilded cradle
<@478616919721771019> ah nvm forgot I had already put it into a PR:
https://gith...
BrainCraft uses ST7789 rather than ILI9341. Will I need a different overlay for that?
crude blaze
BrainCraft uses ST7789 rather than ILI9341. Will I need a different overlay for ...
ah didn't realize, yea that needs a different driver.
There is a DRM driver for it but it uses the same compatible string as the fbtft driver so not sure which one would get loaded as both seem to be enabled in the Kernel.
https://github.com/raspberrypi/linux/blob/rpi-6.1.y/drivers/gpu/drm/panel/panel-sitronix-st7789v.c
Otherwise they had recently merged that generic panel-mipi-dbi driver recently where someone had success using that under wayfire with the ST7789V I believe (and also their custom driver but not sure if that got merged into the kernel)
gilded cradle
Ok yeah. I've had success with the panel driver, but only if I used X11 instead of Wayland.
crude blaze
then it loaded the fbtft driver in that case
gilded cradle
Ah
crude blaze
panel-mipi-dbi seems to be the way to support SPI based drivers in the future, I think you had added a version of that to the pi scripts repo too but now its included "officially" so that makes it easy to use.
There are a couple good threads on the forum by 6by9 and aBUGSworstnightmare on using that with different panel ICs
gilded cradle
panel-mipi-dbi seems to be the way to support SPI based drivers in the future, I...
Ok. Well I'll take a look and try experimenting to see what works. Currently chasing down a separate issue with audio.
manic glacierBOT
short tendon
Random question... Looking at adafruit_portalbase/network.py there is this method get_local_time that does all this crazy breaking up of "%Y-%m-%d %H:%M:%S.%L %j %u %z %Z" to make a time.struct_time. Any reason it's not something simpler like:
TIME_SERVICE_FORMAT = "%Y,%m,%d,%H,%M,%S,%j,%u,0"
which would return
'2024,01,11,16,50,58,011,4,0'
which we could do:
now = time.struct_time([int(value) for value in response.text.split(",")])
and removing 9 lines of code?
lone sandalBOT
tulip sleet
Random question... Looking at `adafruit_portalbase/network.py` there is this met...
I think that would be fine. The code was probably adapted without rethinking the requirements.
short tendon
I'll test around with it a bit more and open a PR if it makes sense. Found it when I was looking to do something similar without reinventing the wheel
covert vine
Hi, I'm working on adding board support for a custom ESP32 board. I managed to successfully configure the files, build the firmware and flash it to the board, but I can't get Web Workflow working. The board is connected to the same wifi as my laptop and I can ping the board, but trying to access it via mdns or directly accessing the ip from a browser doesn't work (tried on Firefox and Chromium on PopOs and Firefox on Android phone). I also attached my settings.toml with replaced ssid and passwords, but it should be the same as the instructions... Any suggestions would be welcome, thanks!
tulip sleet
Hi, I'm working on adding board support
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
crimson ferry
someone please reassure me that raising an exception to pass an interrupt from C code to Python code is way too kludgy to even think about
slender iron
someone please reassure me that raising an exception to pass an interrupt from C...
It’d make a lot of code uncertain. If you wanted interrupts in cp, then do it like mp.
What are you trying to do?
crimson ferry
it was just a crazy thought
slender iron
🙂
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
cobalt grail
Not sure what is going on, but Weblate seems broken to me.
- It stopped mailing me notifications.
- The suggestions tab is totally broken.
Loads of the suggestions is just "%" or similar nonsense.
Is this a known issue? We are looking into using weblate in another open source project.
tulip sleet
Not sure what is going on, but Weblate seems broken to me.
1. It stopped mailin...
We didn't know about this but will look into it -- thanks!
manic glacierBOT
tulip sleet
Not sure what is going on, but Weblate seems broken to me.
1. It stopped mailin...
Did you get it to work? I see https://github.com/adafruit/circuitpython/pull/8801
onyx hinge
yikes I had not realized how close to full esp32s3 4MB parts are.
bitmapfilter is only ~7kB but it's enough to overflow qtpys & feathers -- camera fits but only because of stuff like audio being turned off.
<@&356864093652516868> just a reminder -- there is no meeting today due to the obervation of Martin Luther King Jr. Day. We will meet 24 hours later than usual this week.
tulip sleet
yikes I had not realized how close to full esp32s3 4MB parts are.
```1430656 byt...
I think this may be due to big growth in ulab from gcc 13. I think this is still unresearched. https://github.com/v923z/micropython-ulab/issues/648
tulip condor
Not sure but I think my question is too technical for #help-with-circuitpython : does AESIO takes advantage of the hardware on ESP32s and is it compatible with python3 aes decryption?? From: https://github.com/adafruit/circuitpython/blob/a501a21887e87cead1a37a9a3ced609a67b7801c/shared-bindings/aesio/aes.c it would seems not but maybe Im missing something
Im not seeing #defines for esp32 that is why I dont think it takes advantage of flash memory encryption aes and/or aes hardware acceleration features on the esp32s
onyx hinge
@tulip condor our implementation of aesio is in https://github.com/adafruit/circuitpython/blob/main/shared-module/aesio/aes.c and is just a software C implementation. It does not use microcontroller-specific facilities to work.
I don't think the module's API is compatible with CPython. However, it should comply with the AES standard. We test it against several test patterns from the AES specification in tests/circuitpython/aes.py.
aesio vs pycrpyotodome: ```py
CircuitPython:
import aesio
key = b"Sixteen byte key"
inp = b"CircuitPython!!!" # Note: 16-bytes long
outp = bytearray(len(inp))
cipher = aesio.AES(key, aesio.MODE_ECB)
cipher.encrypt_into(inp, outp)
outp
bytearray(b'\xab\xb1\xa1\xf9\x8fW@\x9eEZ\xc0nqS_\xfe')
Python3 with PyCryptoDome:
from Crypto.Cipher import AES
key = b'Sixteen byte key'
inp = b"CircuitPython!!!" # Note: 16-bytes long
outp = bytearray(len(inp))
cipher = AES.new(key, AES.MODE_ECB)
cipher.encrypt(inp)
b'\xab\xb1\xa1\xf9\x8fW@\x9eEZ\xc0nqS_\xfe'
thanks a lot !
onyx hinge
thanks a lot !
if you want to dive into low level microcontroller code and add support for AES instructions, that'd be awesome. it's just not anything we've prioritized yet.
For espressif it looks like "just" shifting to using the mbedtls implementation of AES would get you hardware AES support. Which might be a good idea to do in general, similar to how it was done recently with hashlib
tulip condor
hum Id have to figure out first if it is supposed to be used directly. It is in the mbedtls implementation of esp-idf but it's not used directly. Even the arduino esp32-arduino library dont use it
Also using flash encryption has a lot of side effects and is irreversible
cobalt grail
Did you get it to work? I see <https://github.com/adafruit/circuitpython/pull/88...
No, those were fully manual translations.
onyx hinge
@cobalt grail If the weblate website isn't working properly for you, it's appropriate to open issues on weblate's github: https://github.com/WeblateOrg/ -- you can mention that you are using CircuitPython which is a paid account, not sure if it makes a difference to how they triage issues. It seems they did change something about e-mail notifications recently: https://github.com/WeblateOrg/weblate/pull/10659 it was supposed to work the same as before but you might just go look at your profile and check the notification settings anyway..
fwiw I also see problems with automatic suggestions from libretranslate, suggesting just "%" as a translation. oof.
As for translations, I guess ...
There used to be a site-wide configuration at Hosted Weblate, but it is now gone for several reasons. So the project has to configure own if it is desired to use machine translation. https://github.com/WeblateOrg/weblate/issues/9689
I have deactivated LibreTranslate as we do not have an API key for it.
cobalt grail
So what translation services are supposed to be working?
onyx hinge
none.
cobalt grail
Oh, there were a bunch of them when I first started to use Weblate.
onyx hinge
Oh, there were a bunch of them when I first started to use Weblate.
It seems like they've found them too costly to provide, or something. Personally, I have no idea if it's helpful or unhelpful. Sounds like you found it helpful, though. When you used it, was there one translation service that was better than the others? I can check into price and find out if Adafruit wants to pay for a translation API key on an ongoing basis.
that is, "[hosted weblate] found them too costly"
cobalt grail
Not sure which ones was available, but there was two that was helpful. Mostly needed it when the error message used an unfamiliar term. If I recall correctly it was the Google and Microsoft services.
cobalt grail
that is, "[hosted weblate] found them too costly"
Yeah, used to get errors from time to time, I guess that was rate limiting in action.
tulip condor
seems that adafruit could self-host libratranslate
there is a flag to enable api keys or not that probably means that libretranslate is charging the bandwidth to do this which isnt a problem if self-hosted
tulip condor
besides pins availability / hardware capability what detrermines what modules are available for a board ? Is it the way the shared modules are included or not for each board as well? Does this mean one could recompile the circuitpython image for a board with some libraries the official download doesnt provide and without some libraries the official download provide ? I assume that if a board has X amount of space in KB that some choices have to be made for its circuitpython image ?
crimson ferry
Flash, RAM, port / board capabilities, (and whether a specific feature has been ported), determine what modules are included. Yes, decisions are made when flash (or RAM requirements) are constrained (which is often). Custom builds are possible to delete some modules and add others, provided the other constraints are met.
tulip condor
so if I dont see some defines for a board in the init.c/init.h or even an include of board.h or the board supervisor or the pins macros used are available according to my board config file (mpconfig and pins.h) I could assume the module doesnt have any board-specific limitations and it's just a question of including it on the image ?
crimson ferry
Someone else will have to answer that one. But my general sense is that when possible, if flash permits, all applicable ported modules are included. When flash doesn't permit, some priorities had to be decided to include some modules at the expense of other(s).
Separate question... I'm a little confused (or recall wrong) about some info on https://github.com/adafruit/circuitpython/tree/main/ports/espressif for ESP32-S2 and ESP32-S3:
It says pin 20 (D+) and pin 19 (D-) are used:
to access the CircuitPython drive
and that
A USB to UART converter can be used for connecting to ESP32-S2 to get access to the serial console and REPL and for flashing CircuitPython.
using pins 43 (RX) and 44 (TX).
But doesn't the drive and normal serial console / REPL both use the native USB, and the USB-to-UART is for the debug console (and flashing) and not the REPL (at least in standard builds)?
tulip condor
I see . By porting I guess someone would have to write it in esp-idf so that it works on esp32. So if it's not in the image it may simply mean that nobody has ported it yet for that platform not necessarily that the esp32 cant run it
or in whatever api/sdk the samd21 uses in its firmware
crimson ferry
if it's ported, there will be something in common-hal (or in shared-modules if it doesn't require port-specific code)
shared-bindings (the common API as documented in readthedocs) will typically make calls to common_hal_{something}, and those functions are (usually) either in common-hal (or shared-modules)
manic glacierBOT
manic glacierBOT
onyx hinge
Auto-reload is off.
Running in safe mode! Not running saved code.
You are in safe mode because:
CIRCUITPY drive could not be found or created.
Press reset to exit safe mode.
Adafruit CircuitPython 9.0.0-alpha.6-42-g029c6c2f0c-dirty on 2024-01-16; posix with amd64
>>> print(3+3)
6
>>> ^C
```It's not _actually_ working yet but it's close!It's becoming more and more of a hack to incorporate some circuitpython modules into the "unix" port for testing purposes, so this is an ongoing attempt to build something that runs on a POSIX system but is built "just like circuitpython".
lone axle
<@&356864093652516868> The weekly meeting is set to occur here on discord in about 30 minutes. We're on Tuesday this week due to the US holiday yesterday. We look forward to seeing folks for the meeting!
Notes doc is in the pinned message feel free to fill in status updates / hug reports before the meeting if you can.
midnight ember
errant grail
Extremely useful design documentation. Nice overall UI design, too.
slender iron
when you have many labels, are they in a subgroup?
midnight ember
some are main groups only, some are subgroups but overall yes. remove and append makes a performance difference.
slender iron
hrm, so hidden stuff is costly still
slender iron
Oh no! Have you fully recovered?
midnight ember
as of yesterday yup feeling much better.
midnight ember
What is a Pico B2? Haven't heard of that one.
slender iron
could be a version
candid sun
thanks @lone axle !
midnight ember
Thank you for hosting @lone axle
Welcome back @slender iron hope everything is going well
slender iron
thanks! excited to dig into CP things again
lone axle
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1hvExFQWMsV78bORs1uBxon_bncchrHPV96r4cAWEhcA/edit?usp=sharing
short tendon
Would next week be a good time to talk about the connection manager?
devout jolt
What is a Pico B2? Haven't heard of that one.
@slender iron RP2040B2 is a hardware revision to the RP2040. It fixes two issues with USB ("USB device fails to exit RESET state on busy USB bus." and "USB device endpoint abort is not cleared.") This is described in the "Errata" section of the RP2040 datasheet (See section "4.1.2.6.5. Errata") And these issue I believe have work-arounds in the Pico-SDK
midnight ember
Had no idea. Thanks!
spare jacinth
As an extra: B1 (second version) has a software fix on the SDK for an issue too, not sure if one of those or a third one
slender iron
Would next week be a good time to talk about the connection manager?
I should be around
onyx hinge
Yay, I just marked bitmapfilter out of draft mode.
lone axle
it seems that the page here: https://circuitpython.org/contributing/library-infrastructure-issues hasn't updated for a few days or a week-ish. Many of the librariers listed as having new commits since the latest release do actually have releases that are more recent than their newest commits. The release date shown on the page is isn't accurate for the actual latest release on them.
short tendon
I should be around
Great. I'll add it to the weeds...
main furnace
Thanks Tim!
gilded cradle
panel-mipi-dbi seems to be the way to support SPI based drivers in the future, I...
Hey, I have not been able to get the drivers working. I tried the 2.2" PiTFT with the pitft22 overlay first and that initializes the display and makes it black, but nothing is displayed. Next I tried compiling the overlay you may for the pitft 2.4 with a tsc2007 wired up using the gpios specified in the overlay, but it remains white. Any tips?
crude blaze
Hey, I have not been able to get the drivers working. I tried the 2.2" PiTFT wit...
if it remains white that usually means unitialized. Which Kernel version are you working with? Might need to run an rpi-update. Some of the latest display changes were only merged recently after the latest release
gilded cradle
if it remains white that usually means unitialized. Which Kernel version are you...
I'm using 6.1.0
I'll try updating
crude blaze
I'm using 6.1.0
ah okay, I've only tested on 6.1.65 so far. Maybe thats it
gilded cradle
ah okay, I've only tested on 6.1.65 so far. Maybe thats it
It just updated to 6.1.72
crude blaze
newer than 65 should be fine
gilded cradle
Still nothing on the 2.4". I'll try the 2.2".
crude blaze
you might need to do a cold boot with power removed. I had some issues during development with warm reboots where there was no power cycle of the display
gilded cradle
Ah, I'll try that next. I just hooked up me 2.2"
Same deal on 2.2". I'll try 2.4" with cold boot.
@crude blaze ok, 2.4" initialized from a cold boot, but no desktop on display. Basically same as 2.2".
crude blaze
do you have an HDMI display connected at the same time?
crude blaze
then make sure you have
dtoverlay=vc4-kms-v3d
max_framebuffers=2
but its default in newer distro releases
gilded cradle
Yeah, that's the settings I have in my config
I'm double checking my wiring. Weird that the 2.2" inited the 2.4" but the overlay with correct gpios is still white.
crude blaze
if it stays white then I would double check wiring and maybe double check that the overlay is unmodified. Otherwise unsure what the difference could be
gilded cradle
🙂