@edgy edge maybe you're missing MP_REGISTER_MODULE?: ```MP_REGISTER_MODULE(MP_QSTR_cyw43, cyw43_module, CIRCUITPY_CYW43);
#circuitpython-dev
onyx hinge
tulip sleet
@edgy edge if you can point to a repo with the code, that would give the most context
edgy edge
@onyx hinge originally I was missing MP_REGISTER_MODULE but fixed it last night. I'll clean up my repository and create a single commit isolating my changes for this issue
analog bridge
tulip sleet
https://github.com/adafruit/esp-iot-solution/pull/1
https://github.com/adafruit/...
I should have been subscribed to these new repos automatically, sorry, I didn't see these. WIll take a look.
edgy edge
@onyx hinge could you take a look at this commit? Ignore the readme, it is a work in progress it will get updated once the process works.
tulip sleet
@analog bridge I had some questions and passed the esp32-camera fixes to @onyx hinge.
onyx hinge
<@426200547024961539> could you take a look at this commit? Ignore the readme, i...
which "this commit"?
tulip sleet
@edgy edge did you mean to submit a PR, or just point us to the repo?
edgy edge
I meant to point to the commit in my fork but generated an incorrect pull request by mistake
tulip sleet
https://github.com/aulitech/circuitpython/tree/neuton
I looked at the diffs. In some places you use CIRCUITPY_AULITECH_NEUTONML and in others you use AULITECH_NEUTONML.
orchid basin
onyx hinge
I don't see the problem. I built the firmware and opened it in the debugger. the new module is listed in mp_builtin_module_table: ```(gdb) p mp_builtin_module_table
$1 = {{key = 0x156, value = 0x94a0c <mp_module___main__>}, ...
{key = 0x4416, value = 0xa052c <neutonml_module>},
...
}
If you didn't try "make clean", then try it now. This could be a class of change that is not correctly tracked by the build system's dependency management.@edgy edge ^
edgy edge
but ```Adafruit CircuitPython 8.0.0-beta.0-53-g676aa321a3-dirty on 2022-09-20; Aulitech Cato with nRF52840
help("modules")
future board onewireio touchio
main builtins os traceback
_asyncio busio paralleldisplay ulab
_bleio collections pulseio ulab
adafruit_bus_device countio pwmio ulab.numpy
adafruit_bus_device.i2c_device digitalio rainbowio ulab.numpy.fft
adafruit_bus_device.spi_device displayio random ulab.numpy.linalg
adafruit_pixelbuf dotenv re ulab.scipy
aesio errno rgbmatrix ulab.scipy.linalg
alarm fontio rotaryio ulab.scipy.optimize
analogio framebufferio rtc ulab.scipy.signal
array gc sdcardio ulab.scipy.special
atexit getpass select ulab.utils
audiobusio io sharpdisplay usb_cdc
audiocore json storage usb_hid
audiomixer keypad struct usb_midi
audiomp3 math supervisor uselect
audiopwmio microcontroller synthio vectorio
binascii micropython sys watchdog
bitbangio msgpack terminalio zlib
bitmaptools neopixel_write time
Plus any modules on the filesystem```
i'll do another make clean ...
winner. Thank you @onyx hinge and @tulip sleet
onyx hinge
yay!
It's a weakness of make that a change (like adding -DFOO=1 to CFLAGS) doesn't automatically cause targets that depend on the value of CFLAGS to be rebuilt.
edgy edge
yes, but I've been around long enough and should've known that. Sorry to waste your time
onyx hinge
it's no problem!
proven garnet
Blinka now has a python wheel for download from PyPI ๐
tulip sleet
Blinka now has a python wheel for download from PyPI ๐
actually, I don't know what this means vs the previous PyPi release ๐
proven garnet
Before only source distributions we're uploaded, so there was also the local build process after downloading. Python wheels (and wheels in general) cut that later part out. So for the most part only the download is required.
I don't think it's massively important for us, but for a package like numpy it's a game changer because a wheel cuts out a lot of the local building required if you download the source and have to do a local install process.
Still, time saved is always good
tulip sleet
thanks!
tidal kiln
@idle owl should this line be indented inside the if conditional?
https://github.com/adafruit/Adafruit_CircuitPython_seesaw/blob/6234787515e2f0ece40b6408722ff0b42824038e/examples/seesaw_rotary_neopixel.py#L51
idle owl
<@330227457296957440> should this line be indented inside the `if` conditional?
...
Probably?
tidal kiln
Probably?
meh. it was related to something that ended up being soldering. but still looks odd. think it should be like the simpletest code:
https://github.com/adafruit/Adafruit_CircuitPython_seesaw/blob/6234787515e2f0ece40b6408722ff0b42824038e/examples/seesaw_rotary_simpletest.py#L34
jaunty juniper
yeah saving the "last value" is something I tend to do indifferently in or out of that kind of test
idle owl
meh. it was related to something that ended up being soldering. but still looks ...
If you have a way to test it, feel free to update it. I'm not able to right now.
Or file an issue, I guess.
Label it "Good First Issue" if you do.
Because that's a pretty good one.
gilded cradle
Does anybody know if CircuitPython's serial speed is always 115200 (https://github.com/adafruit/circuitpython/blob/main/supervisor/shared/serial.c#L133-L135) or does it vary by board?
onyx hinge
@gilded cradle looks like that line of code would be common to any board using uart as console, just based on the file path
gilded cradle
Looks like it varies after doing a search.
Thanks @onyx hinge but it looks like it's constructed in UART.c
onyx hinge
oh! hm. I love being wrong
gilded cradle
oh! hm. I love being wrong
I think you might not be wrong. I'm still poking around, but other frequencies appear to reference the busio connection to other peripherals.
austere acorn
@onyx hinge is the Union(str, typing.BinaryIO) with () or [] ? My doc check failed ๐
tulip sleet
<@426200547024961539> is the Union(str, typing.BinaryIO) with () or [] ? My...
it is with []. If you look in the readthedocs, there are a bunch of examples
austere acorn
it is with []. If you look in the readthedocs, there are a bunch of examples
Thank you!
I'm finding my way through first core contribution ๐
onyx hinge
<@426200547024961539> is the Union(str, typing.BinaryIO) with () or [] ? My...
OOps, I may have given the wrong advice, thanks for working on it though!
proven garnet
Looking for feedback - if I'm going to show people how to get started with Hacktoberfest, is it better to do it from within Windows rather than Linux? I would guess that's what most people are going to use, but I have no idea. It's also my bias because I definitely did start there.
I imagine that showing people how to use GitHub Desktop might be more useful for a larger group of first-time contributors than using VS Code git window or command line...
tulip sleet
well, they might have a Mac
For library work, it doesn't matter, for core work, I would rather recommend vanilla linux
maybe WSL
proven garnet
Good point. And in that case, GitHub Desktop is available for macOS as well
midnight ember
If your target demographic is beginners on windows Iโd recommend Mu and GitHub desktop. Canโt use pycharm in windows with serial. VScode GitHub setup is very intimidating.
proven garnet
If your target demographic is beginners on windows Iโd recommend Mu and GitHub d...
Ah right, Mu is good
brazen hatch
How can I tag a circuitpython build to have a specific name?
brazen hatch
I kinda need it to propely tag the picow builds as 8.0
because for some reason they are tagged as 6.0
even though they are 8.0
jaunty juniper
I kinda need it to propely tag the picow builds as 8.0
because for some reason t...
that's a problem with pulling the tags, how did you clone the repository/branch ?
austere acorn
Hello ๐
Could someone help me with my PR please? https://github.com/adafruit/circuitpython/pull/6931
After making the requested documentation changes, all the checks passed... except for one, that timed out after like 6 hours.
Thinking it was a fluke, I did a rebase to trigger the checks again ... this time there were 12 failed checks with the same error as in the pic.
I made another small change + commit ... now one failed again, in a different spot, with a different makes-no-sense-to-me error.
I'm lost!
jaunty juniper
Hello ๐
Could someone help me with my PR please? https://github.com/adafruit/c...
looks like network errors on github's side, we can ignore it and merge if needed
austere acorn
looks like network errors on github's side, we can ignore it and merge if needed
Oh, okay... phew, thank you!
I didn't know what I was doing wrong, after the first commits went so well ๐
jaunty juniper
I kinda need it to propely tag the picow builds as 8.0
because for some reason t...
ah I know the issue, try that:
git remote add adafruit https://github.com/adafruit/circuitpython
git fetch --tags adafruit
The failed job was a CI failure (not your fault). GitHub Actions has been somewhat flaky the past few days. I am rerunning the failed job.
brazen hatch
ah I know the issue, try that:
```sh
git remote add adafruit https://github.com/...
I just git tab -a 8.0.0-picow and it good now
Got it to boot ljinux btw.
But the wanna-be kernel module did not like the wifi driver lacking ssl.
I should add a failover for that.
Many thanks though!
onyx hinge
@austere acorn well done on that PR to the core! ๐
brazen hatch
Since you here Jepler, this is a nice moment to thank you for your work!
I will go stress test what works on the picow.
onyx hinge
oh, yeah, not much works and we're not ready for bug reports about it yet.
brazen hatch
oh alright
onyx hinge
but you're welcome
tulip sleet
basically this is pre-alpha ๐
brazen hatch
yea tbh
got mine connected, pinging both ways tho
its not insignificant
Wish I could actually help with the code..
My C is terrible though.
tulip sleet
your testing is helpful, don't worry, there are many ways to help, and we appreciate it!
austere acorn
@onyx hinge @tulip sleet Thank you very much! ๐
austere acorn
<@196904073255976961> well done on that PR to the core! ๐
If you have the time, could you give me a bit more details about what you asked for the MP3Decoder ?
I want to make sure I have the right idea.
Was it to add another MP3Decoder class property, like the file pointer, bits_per_sample, channel_count and so on? One that would store the file name and path as a string?
onyx hinge
@austere acorn I'm not sure, that's why I suggested to postpone it. I know that existing code assumes the mp3file.file object is an open stream. so maybe a new method mp3file.open that takes a str...? My original mp3 project, jeplayer, uses the file property: ```py
def change_stream(filename):
"""Change the global MP3Decoder object to play a new file"""
old_stream = mp3stream.file
mp3stream.file = open(filename, "rb")
old_stream.close()
return mp3stream.file
Later it uses the tell method on that file to estimate the playback percentage: `playback_display.progress = mp3file.tell() / file_size`so what if change_stream becomes: mp3stream.open(filename); return mp3stream.file? That seems like a nice improvement.
austere acorn
so what if change_stream becomes: `mp3stream.open(filename); return mp3stream.fi...
Oh, I see, so you could change the file used by the Mp3Decoder, without having to worry about closing the old file and opening a new one
And I'm assuming that same behaviour makes sense for WaveFile too
onyx hinge
It's less important for WaveFile. It's important for MP3Decoder because MP3Decoder needs to allocate a few large chunks of memory. If you repeatedly construct your MP3Decoder within your program, you eventually get a MemoryError.
and because WaveFile is on even small devices like samd21 we don't want to unnecessarily add 'stuff' to it
so I'd do it only on MP3Decoder at this time
austere acorn
Okay, makes sense!
edgy edge
we need to maximize both memory and storage available for machine learning models. Would it be advisable to remove modules we're not using? Any red flags in this list? We are using the stock XIAO nrf52840 BLE SENSE. Would there be any significant benefit to including BLE and IMU in the circuit python build instead of as libraries? We do use the USB port for debug and file loading```
Remove
adafruit_pixelbuf audiobusio audiocore audiomixer audiomp3
audiopwmio bitbangio bitmaptools displayio fontio
framebufferio keypad msgpack neopixel_write onewireio
paralleldisplay rainbowio rgbmatrix rotaryiosd cardio
sharpdisplay synthio terminalio touchio usb_midi
vectorio
Keep
_bleio adafruit_bus_device aesio alarm
analogio atexit binascii boardbusio
countio digitalio dotenv errno
getpass json math microcontroller
nvm os pulseio pwmio
random re rtc storage
struct supervisor time traceback
ulab usb_cdc usb_hid watchdog
zlib```
tulip sleet
we need to maximize both memory and storage available for machine learning model...
The builtin modules are in C and cannot be loaded as libraries. So you'll just need to pick and choose what you'll really need. Can you store the models in the external flash
the display-related modules are probably not important to you and could be turned off
or at least some of them
edgy edge
would any of these be unsafe to remove? ```
adafruit_pixelbuf audiobusio audiocore audiomixer audiomp3
audiopwmio bitbangio bitmaptools displayio fontio
framebufferio keypad msgpack neopixel_write onewireio
paralleldisplay rainbowio rgbmatrix rotaryiosd cardio
sharpdisplay synthio terminalio touchio usb_midi
vectorio
do you have an attached display? If not, then they are fine to remove
keypad is useful if you have attached buttons
edgy edge
is there anything like a module dependency graph documented or derivable? It's hard to know where is something like terminalio is used
tulip sleet
there are some dependencies reflected in the builds, but they are not complete. See py/circuitpy_mpconfig.*
terminalio is used for displaying the REPL on a display
edgy edge
no displays no buttons, nothing attached. Just onboard mic, imu and ble. would terminalio be used for USB serial debugging/ development?
stuck elbow
no displays no buttons, nothing attached. Just onboard mic, imu and ble. would t...
no, only on the display
stuck elbow
it's pretty easy to see when you have unmet dependency, because you will get a compile-time error
edgy edge
thanks. I was afraid of unresolved externals at run time that might not be caught easily
edgy edge
makes sense and perfect answer
is setting CIRCUITPY_<module> 0 in /circuitpython/ports/nrf/boards/aulitech/mpconfigboard.mk the correct way to remove from build? Or will I expect some trial and error to find other build dependencies?
tulip sleet
is setting CIRCUITPY_<module> 0 in /circuitpython/ports/nrf/boards/aulitech/mpco...
That's exactly right. If you browse the other nrf boards you'll see various boards with things turned off (e.g. the satellite boards)
manic glacierBOT
analog bridge
Do we have a learn guide on dualbank module?
tulip sleet
Do we have a learn guide on `dualbank` module?
Nope!
jaunty juniper
huh funny, doing display.root_group.scale = 2 on a board with a builtin display, makes the scale keep growing on every reload
tulip sleet
whoopsie ๐
jaunty juniper
it's a hack to begin with, but it's cool that it (kind of) works, might make it easier to show text on an M0 without using adafruit_display_text
(no idea if it's actually corrupting memory by writing "outside of the screen" or anything)
stuck elbow
that's weird, what do you get when you print display.root_group.scale?
jaunty juniper
1
stuck elbow
I used that to get a bigger terminal on my robot, but I didn't notice it growing every time, just once as you'd expect
tulip sleet
maybe it's specific to a particular (kind) of display
stuck elbow
seems like it's getting propagated?
tulip sleet
something is doing scale = scale * new_scale ?
stuck elbow
right, on the contents of the group, probably
jaunty juniper
if I change it in boot.py it's fine since it changes only once, but I suspect I need to use reset_terminal
or something
the REPL text goes outside of the screen
(I'm testing on 7.3.3 right now too)
stuck elbow
the scale on root_group gets reset, but not on its contents, and when you set it to 2 again, it multiplies the contents again
I bet there is code that compares the old value and the new value, and updates the contents based on that
(just guessing, though)
proven garnet
Hacktoberfest 2022
proven garnet
What format do I need to upload something to YouTube? MP4?
proven garnet
If anyone that run/records the meetings remembers their settings ๐
onyx hinge
@tulip sleet do you want me to make the pico/picow share a USB PID/VID or do you want me to request a new PID from raspberrypi?```Running USB VID/PID and Creator/Creation ID Duplicate Checker...
Duplicate VID/PID usage found!
- VID/PID: 0x239A:0x80F4
Boards: raspberry_pi_pico, raspberry_pi_pico_w```
tulip sleet
<@329766224093249548> do you want me to make the pico/picow share a USB PID/VID ...
that is our VID (239a), so we can just assign one of own. It should definitely have its own PID
onyx hinge
oh it is?
OK, in that case -- can you do that for me? or is it time for me to learn how?
tulip sleet
it's easy - an internal repo
idle owl
If anyone that run/records the meetings remembers their settings ๐
mkv works equally well.
idle owl
If anyone that run/records the meetings remembers their settings ๐
Also if using OBS, it gracefully handles crashes by closing the file when using mkv. MP4 does not.
proven garnet
Also if using OBS, it gracefully handles crashes by closing the file when using ...
Thanks!!
onyx hinge
@tulip sleet for picow I need submodules that are current living inside pico-sdk. also those are needed ONLY for one board, not for the whole port.
What do you advise me to do in ci_fetch_deps.py for this?
jimmo noticed to me that micropython puts copies of those submodules under lib/ so they don't have any recursive submodules, should i do that? and just list it for the whole raspberrypi port?
tulip sleet
<@329766224093249548> for picow I need submodules that are current living inside...
separated submodules in ports/raspberrypi sounds ok to me, we just need to keep them sync'd up with the pico-sdk version
ideally in the long run I would like to see different options for different submodules: e.g. shallow for broadcom firmware, full for small repos, recursive as necessary ,etc. I think we could figure that out but not right now
jaunty juniper
we don't support raise ... from ... right ? Or more accurately it ignores the from part ?
jaunty juniper
it's used in the miniMQTT library and that kind of hides the original exception because of that
pine tartan
hi everyone, i'm using a QT PY ESP32 Pico and trying to subscribe to the adafruit MQTT broker. i'm following this Guide but when I try to run the code, I just get an exception and the exception doesnt Actually say anything.
jaunty juniper
it's used in a few libraries in fact, we should look into that maybe
ah there:
Exception chaining not implemented
https://docs.micropython.org/en/latest/genrst/builtin_types.html#exception-chaining-not-implemented
ah that doesn't mention "from" though, but same thing
inner orbit
Is there ANY way to get a Trinket M0 to act as an I2C "slave" with CircuitPython??? I got ten I need to talk to. Don't wanna use C. Teaching some students.
jaunty juniper
I don't think i2cperipheral (now i2ctarget in CP8) is implemented on SAMD21 at all
it's on SAMD51, RP2040 and Espressif
crimson ferry
uart token ring? ๐
jaunty juniper
makes me think that I would like to make an addition to the support matrix, or an alternate page, with support by port/variant
(so if it's not implemented on M0 you don't go looking through the list of boards to see if one has it enabled)
blissful pollen
Does somebody know offhand which source file sets the neopixel to indicate CP statuses?
blissful pollen
Like when it drops to the REPL the pixel changes colors. Just blanking on where that is
crimson ferry
jaunty juniper
oh the code that changes the color
proven garnet
ah that doesn't mention "from" though, but same thing
This just means not implemented though, right? CircuitPython basically ignores the from error portion right?
Not that it's a switch to be turned on, rather checking that writing that doesn't cause an issue.
jaunty juniper
This just means not implemented though, right? CircuitPython basically ignores ...
raise from is just an explicit way to do exception chaining, which is normally implicit when raising from an except, and exception chaining is not implemented
so yeah they allowed the syntax but ignore it
proven garnet
I don' think I ever realized it was implicit like that, thanks!
jaunty juniper
I didn't know either, the python docs mention it:
Exception chaining happens automatically when an exception is raised inside an
exceptorfinallysection. This can be disabled by usingfrom Noneidiom
>>> try:
... raise OSError("hello")
... except Exception as ex:
... raise ValueError("world")
...
Traceback (most recent call last):
File "<stdin>", line 2, in <module>
OSError: hello
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "<stdin>", line 4, in <module>
ValueError: world
>>>
What is the best mapping in a shared-bindings implementation for a C library function with float *inputs parameters? Is mp_obj_get_array() the way to go? ```i
// CircuitPython
import neutonml
nml = neuton()
my_floats.tile(my_floats.float32(0), 10)
nml.set_inputs(my_floats)
// Shared-bindings/neutonml/Neuton.c
STATIC mp_obj_t neutonml_neuton_obj_set_inputs(mp_obj_t self_in, mp_obj_t inputs)
{
int len, status;
float items
mp_obj_get_array(inputs, &len, (mp_obj_t) items);
status = shared_module_neutonml_neuton_set_inputs(self_in, *items));
return mp_obj_new_int((mp_int_t)status);
}
// shared-module/neutonml/Neuton.c
int8_t shared_module_neutonml_neuton_model_set_inputs(neutonml_neuton_obj_t *self,
float inputs)
{
// call the C lib here
// int8_t neuton_model_set_inputs(input_t inputs);
return neuton_model_set_inputs(inputs);
}```
thorny jay
Just saying: Displaying 333 boards. <= I know some in Adafruit are obsessed by numbers... so that is a number that talk to me. ๐
blissful pollen
Anyone have a thought offhand on why when my ESP32-S2 custom board goes into the REPL all the pins seem to be set to high (the built in LEDs all turn on).
If I set up one of the pins as a DigitalInOut the pin gets set low
blissful pollen
Yeah they are set in there. I vaguely recalled some change that they may be resting high now (didnt do that in 7.x when I last tried this).
oh in pins.c I did. I guess in board.c I could then default them low of course
Thanks @jaunty juniper. Defining the double tap pin also fixed the problem then of reset always going into boot mode.
Now just to figure out why it seems to exception into REPL after a reset IF I'm connected to a computer (vs straight power)
edgy edge
I'm trying to figure out the best way to wrap a 3rd party C library for circuit python. Almost done, but need to learn the best way pass arrays of floats. The library contains two functions that don't match any examples I can find: /// /// \brief Make a prediction /// \param index - pointer to predicted class variable (binary/multi classification). Can be NULL. /// \param outputs - float[] array of neuton_model_outputs_count() elements, contains predicted target variable /// (for regression task) or probabilities of each class (binary/multi classification). /// \return Zero on successful prediction. Result > 0 - model not ready for prediction. /// int8_t neuton_model_run_inference(uint16_t *index, float **outputs); /// /// \brief Set input values /// \param inputs - input_t[] array of neuton_model_inputs_count() elements /// \return Zero if model ready for prediction. Result < 0 indicates error, result > 0 - model not ready for prediction. /// int8_t neuton_model_set_inputs(input_t *inputs);
edgy edge
should i wrap the returned results in a python dict? like this```
STATIC const mp_rom_map_elem_t mcu_processor_locals_dict_table[] = {
{ MP_ROM_QSTR(MP_QSTR_frequency), MP_ROM_PTR(&mcu_processor_frequency_obj) },
{ MP_ROM_QSTR(MP_QSTR_reset_reason), MP_ROM_PTR(&mcu_processor_reset_reason_obj) },
{ MP_ROM_QSTR(MP_QSTR_temperature), MP_ROM_PTR(&mcu_processor_temperature_obj) },
{ MP_ROM_QSTR(MP_QSTR_uid), MP_ROM_PTR(&mcu_processor_uid_obj) },
{ MP_ROM_QSTR(MP_QSTR_voltage), MP_ROM_PTR(&mcu_processor_voltage_obj) },
};
STATIC MP_DEFINE_CONST_DICT(mcu_processor_locals_dict, mcu_processor_locals_dict_table);
const mp_obj_type_t mcu_processor_type = {
{ &mp_type_type },
.name = MP_QSTR_Processor,
.locals_dict = (mp_obj_dict_t *)&mcu_processor_locals_dict,
};```
crimson ferry
Adafruit CircuitPython 8.0.0-beta.0-65-g5781bd318 on 2022-09-22; Raspberry Pi Pico W with rp2040
with requests.get(TEXT_URL, headers={"User-Agent": "๐ฅง๐ฎ",}) as response:
This is a test of Adafruit WiFi! If you can read this, its working :) ๐
lone sandalBOT
blissful pollen
I'm trying to figure out the best way to wrap a 3rd party C library for circuit ...
Just so I make sure Iโm clear youโre trying to return an array of floats from your function?
Iโm on my phone so canโt look up code easily but I believe you would create a Python array object and then fill it my mp_float objects.
Creating a float is something like mp_new_float(float f)
The array I canโt recall off the top of my head
manic glacierBOT
jaunty juniper
ah that bothers meeee
โฏ ag 'USB_MANUFACTURER *= *"Adafruit"' | wc
20 60 1967
โฏ ag 'USB_MANUFACTURER *= *"Adafruit Industries LLC"' | wc
45 225 4803
that too
ports/espressif/boards/adafruit_feather_esp32s2_tft/mpconfigboard.mk
4:USB_PRODUCT = "Feather ESP32-S2 TFT"
ports/espressif/boards/adafruit_feather_esp32s2/mpconfigboard.mk
3:USB_PRODUCT = "Adafruit Feather ESP32S2"
โฏ ag 'USB_PRODUCT = "Adafruit' | wc
4 27 482
I'm also super bothered by the upper cases in Seeed_XIAO_nRF52840_Sense
but that's because I have a script that lets me use tab-completion to build a board
blissful pollen
Don't look in the ATMEL ports and see all the Adafruit boards with the "Adafruit_" leading them but the similar boards all have it
jaunty juniper
yeah Adafruit didn't put adafruit_ in their board IDs in the beginning, but at least it's lowercase ๐
edgy edge
Just so I make sure Iโm clear youโre trying to return an array of floats from yo...
I'm using a third-party C library for machine learning. This is how it would get called from C input_t inputs[] = { feature_0, feature_1, ... feature_N}; if (neuton_model_set_inputs(inputs) == 0) { uint16_t index; float* outputs; if (neuton_model_run_inference(&index, &outputs) == 0) { // code for handling prediction result returned in index and outputs } } but I will be calling it from python and I'm not sure how to handle pass by reference parameters in the shared-bindings code.
blissful pollen
I'm using a third-party C library for machine learning. This is how it would get...
A recent module analogbufio is similar in taking an array of floats in for processing in the object BufferedIn. You create an array and pass it to the call
edgy edge
A recent module `analogbufio` is similar in taking an array of floats in for pro...
passing in the array is the easy part. I'm confused about returning the pass by reference results in index and outputs
instead of passing by reference would it be preferable to return an object { result: x, index: y, output : [] }
blissful pollen
In Python you don't have to return the array I believe. You just fill it in with your values and that object will have them. If you look https://github.com/adafruit/circuitpython/blob/e045415f59526bd0ec0107e0308834031f453d20/ports/raspberrypi/common-hal/analogbufio/BufferedIn.c#L150 The DMA loads into self->buffer (if you look above that is the array). The function call does not return anything.
As you say the alternative is to not pass in an array and create one yourself and return that. That may be better if you want to size the array according to result yourself.
onyx hinge
Just noticed this on my Android 13(?) phone
austere acorn
is that going to format an Adafruit device, or your phone? ๐
edgy edge
In Python you don't have to return the array I believe. You just fill it in with...
thank you for the pointer very helpful and very fresh
onyx hinge
@austere acorn my device I'm sure
austere acorn
Hi @idle owl ๐ (or anyone else who deals with the libraries ๐ )
I noticed that in this documentation https://docs.circuitpython.org/projects/charlcd/en/latest/api.html all the examples use lcd = character_lcd.Character_LCD_I2C(i2c, 16, 2), even those for character_lcd.Character_LCD, character_lcd.Character_LCD_Mono, and character_lcd.Character_LCD_RGB that all have a different set of parameters defined.
Is this a mistake, or am I missing something ?
jaunty juniper
@proven garnet I'd like to point out that Blinka does indeed work on a pico:
https://learn.adafruit.com/circuitpython-libraries-on-micropython-using-the-raspberry-pi-pico
(not that it's in anyway clear what that person in the GPS library issue is actually doing)
proven garnet
Ah gotchya, thanks for pointing that out. So downloading MicroPython firmware + Blinka libraries is valid use?
jaunty juniper
yeah
the person is linking to a MP forum discussion that doesn't seem to use blinka though, so ๐คท
proven garnet
Gotchya, updated the comment. Thanks for correcting me!
(if the comment is still wrong let me know)
jaunty juniper
it's a MP question, let them deal with the MP forums I say ๐
setting the clock from a source is an interesting use, maybe we could implement an rtc module in blinka just for that
proven garnet
Haha fair, I thought it was a running code locally on the computer or something. But yeah, looking more at it, I see that now. It was clear they had an interpreter, so now I see it with all the puzzle pieces that it was the MicroPython one haha.
jaunty juniper
it's tech support, people rarely start by saying what board they are using, what software, and volunteering the full code from the get go, instead... screenshots ! (Do they know we can't use the scrollbar in the picture ?)
analog bridge
Does the FatFs library used in CP provide functionality to re-size partition without loosing data?
manic glacierBOT
idle owl
Hi <@330227457296957440> ๐ (or anyone else who deals with the libraries ๐
)...
I don't think it's a mistake... But I'll be honest with you, this library isn't in my wheelhouse. I don't know it well enough to definitively say one way or the other. I assume each type of display is different, and therefore has different parameters, but that's only a guess.
austere acorn
I don't think it's a mistake... But I'll be honest with you, this library isn't ...
I asked because the examples in the code folder https://github.com/adafruit/Adafruit_CircuitPython_CharLCD/tree/main/examples are different, using the correct parameters, but the documentation is not consistent.
jaunty juniper
yeah all the inline examples for the base class (which takes pins) are examples for the I2C subclass
austere acorn
The docs in character_lcd.py are not consistent
idle owl
It's entirely possible that the docs are not correct. But I would want verification on that from someone more familiar with it than me before saying that.
austere acorn
If it's helpful, I could open an issue, or straight up do a PR with the changes, and have someone verify and confirm?
digital shoreBOT
idle owl
If it's helpful, I could open an issue, or straight up do a PR with the changes...
If you don't mind the possibility of doing the work and then having it not be necessary, please do put in a PR with the fixes. If you'd rather avoid that, file an issue. Both allow for discussion and verification, but the PR route would be more forward effort on your part. (Unless Neradoc has an explanation as to why it's currently correct. They are typing, and I don't know whether it's a response to you or not.)
jaunty juniper
nah
idle owl
Wanted to make sure I wasn't stepping on anything!
@austere acorn In that case, what I said stands. Completely up to you, but one or the other would certainly be helpful.
jaunty juniper
the examples are correct, but they are for a subclass of the class they are written in, which is a little weird, but you can't put them in the subclass, since the methods don't exist in the subclass docs
idle owl
Huh.
jaunty juniper
I don't care much for examples in the methods documentation anyway
idle owl
Fair enough.
jaunty juniper
idle owl
Wow.
jaunty juniper
maybe it can detect if there's an error signaled with the path of a known module, and put a link to the docs
like, wrong number of arguments in neopixel.py ? here's a link to the docs !
idle owl
Oooh. That would be brilliant.
idle owl
@onyx hinge Since you're the only one around, I'll mention it to you. I'm going to put in a PR to reorder the pin names on the Metro M4 Grand Central. PhilB is doing the PrettyPins, and it came to my attention that the far less-used feature pin names are first in the list. So dir(board) is showing those, not the silk D names. Doing this before Phil did the PrettyPins would have been super helpful to him, but it is what it is. I still think it should be updated to match the now-updated PrettyPins, and to make the more likely-used pin name first in the list. Any objections?
manic glacierBOT
pine tartan
Hi everyone, I'm working with MQTT and had some implemention questions
When we run the MQTT loop, is there continuous polling happening in the background or is there an actual event driven things happening?
I'm trying to conserve battery and would like to implement something that isn't wasting battery by continuously polling for data
Since wifi is expensive battery wise
idle owl
@onyx hinge Nevermind. Looks like CircuitPython lists both the silk pins and the PCC pins in dir(board) which seems weird to me, but there it is. I thought it only listed the first instance of the pin name in pins.c? Which is why we swapped the order of LED and D13 on all the boards. Anyway, apparently no PR is needed.
jaunty juniper
<@426200547024961539> Nevermind. Looks like CircuitPython lists both the silk pi...
it lists them all, I believe in the order they are in pins.c
idle owl
it lists them all, I believe in the order they are in pins.c
That did not used to be the case, I thought.
jaunty juniper
which is why we put the preferred name first
jaunty juniper
oh actually no it doesn't change the order, it changes which is the default
by which I mean:
>>> board.D13
board.D13
>>> board.LED
board.D13
because:
{ MP_ROM_QSTR(MP_QSTR_D13), MP_ROM_PTR(&pin_GPIO13) },
{ MP_ROM_QSTR(MP_QSTR_LED), MP_ROM_PTR(&pin_GPIO13) },
OH..... right.....
yeah.
Ok.
I knew that.
What is that from? It should be LED first, lol.
Thanks for reminding me of what I apparently forgot.
jaunty juniper
it might change the order the underrated help(board)
D12 -- board.D12
D13 -- board.D13
LED -- board.D13
that's the Feather RP2040
crimson ferry
@pine tartan polling via .loop(), see: https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT/blob/main/examples/native_networking/minimqtt_adafruitio_native_networking.py
onyx hinge
<@426200547024961539> Since you're the only one around, I'll mention it to you. ...
Makes sense, feel free to tag me for reviewing, but I'll not look at it tonight
onyx hinge
<@426200547024961539> Nevermind. Looks like CircuitPython lists both the silk pi...
Okay I'll forget it then
manic glacierBOT
Build Artifacts can be found here.
Testing the artifact on waveshare_esp32s2_pico.
After a nice swift backup of the current ljinux installation, I ran the commands specified.
After CIRCUITPY came back up, I restored the ljinux installation and powercycled the board.
First 500kb wget:
Matches with my Pi400:
,1)
BASE_CFLAGS +=
-Wno-strict-prototypes
-Wno-missing-prototypes \ why was leaving to play with the last minute percent nice 96% OK we're leaving in commensurate
-Wno-float-equal
else
-Wstrict-prototypes
-Wmissing-prototypes
-Wfloat-equal
endif```
onyx hinge
@edgy edge headers for third parties should be in paths that are included by -isystem, not -I or -iquote. That stops any messages that would occur directly in the headers (so probably the strict-prototypes diagnostics would be fixed this way)
You can place your own correct prototypes for functions in header files you control, which will fix missing-prototypes.
If you determine that float-equal is a mistaken warning, you can disable it on a per file basis by putting a line in the file at the top: #pragma GCC diagnostic ignored "-Wfloat-equal"
You can also apply CFLAGS on a per-file basis, like so: py/py.mk:$(BUILD)/extmod/ulab/code/%.o: CFLAGS += -Wno-missing-declarations -Wno-missing-prototypes -Wno-unused-parameter -Wno-float-equal -Wno-sign-compare -Wno-cast-align -Wno-shadow -DCIRCUITPY
If you want to do it just for a region of code you can use #pragma GCC diagnostic push/pop.
and of course you should consider whether you can contribute improvements to your upstream's code generator to make it better
edgy edge
Thank you! Another great answer
onyx hinge
18% of adafruit_requests.mpy (1580 bytes) is due to static typing! https://gist.github.com/jepler/9ae2dbe8fda6ee5ff9e9360efe3bc157 is my newest idea of how to address it. The idea is, we'd astrogate the .py file to fix the static typing additions, then mpy-cross the result.
for instance in adafruit_requests.py, it replaces the line 'if sys.implementation.name == "circuitpython": with 'if 1:', which allows mpy-cross to eliminate the 'if' test itself as well as the whole 'else' branch. A couple of other idioms I identified are supported as well, and in particular you can write ```py
from future import annotations
try:
from typing import TYPE_CHECKING
except:
pass
if TYPE_CHECKING:
you can write a novel here, it's OK
If you need
```py
try:
from typing import TYPE_CHECKING, cast
except:
def cast(t, v): return v
the overhead seems to be 22 bytes, after it's transformed to def cast... inside an if 1:.
(you can also write except ImportError: to placate pylint and it's the same zero/low overhead in mpy-file-size)
it's called "astrogate" because it uses python's Abstract Syntax Tree, a parsed form of the Python code, to reason about what to do
jaunty juniper
what is that ??
>>> sphinx.__version__
'5.2.0.post0'
make html throws an error in the template because of that I think: too many values indeed
File "/Volumes/CaseSensible/circuitpython-dev/venv-test/lib/python3.9/site-packages/sphinx_rtd_theme/layout.html", line 13, in top-level template code
{%- set (_ver_major, _ver_minor, _ver_bugfix) = sphinx_version.split('.') | map('int') -%}
ValueError: too many values to unpack (expected 3)
(and that's after I sorted out the new :property: error)
jaunty juniper
so if I pip install sphinx==5.2.0 it's fine
austere acorn
Anyone know what this error means in core PR checks ?
jaunty juniper
Anyone know what this error means in core PR checks ?
it looks like it's complaining about the docs in imagecapture using / (declaring arguments before it as positional only)
but why did that never come up before ??
I can only reproduce it locally by running python 3.7
austere acorn
it looks like it's complaining about the docs in imagecapture using `/` (declari...
No idea... as far as I know, I didn't change anything there. I just made some required changes in my PR about MP3Decoder
jaunty juniper
also I see in your PR:
//| def open(self, str) -> None:
it should be something like:
//| def open(self, filepath: str) -> None:
but that's unrelated
onyx hinge
No idea... as far as I know, I didn't change anything there. I just made some re...
It looks like there are some funky things affecting the documentation build right now, we'll get it sorted out. I'll ping you on that PR if you need to take any action
austere acorn
It looks like there are some funky things affecting the documentation build righ...
Okay, thank you! ๐
My PRs do seem to find **all ** the weird CI errors... ๐
tulip sleet
@jaunty juniper I don't know if you are working on that PR fix, but the recommendation is sphinx!=5.2.0.post0, which seems good to me (rather than pin to 5.1.1). If you don't have time to do it now I will push a new commit to your PR.
jaunty juniper
<@199811382793863168> I don't know if you are working on that PR fix, but the re...
doing it now
onyx hinge
@tulip sleet we may need to also pull that fix back to 7.3.x
though I guess we are not otherwise planning a future 7.3.x release atm
tulip sleet
maybe it will get fixed before we do any more 7.3.x releases
if any is right, I have nothing in the pipeline for 7.3.4
onyx hinge
weird, this next error ... https://github.com/pandas-dev/pandas-stubs/issues/296 links to several bugs reported against mypy about it, and .. they fixed it by pinning python back to 3.10.6? https://github.com/pandas-dev/pandas-stubs/pull/298
tulip sleet
weird, this next error ... https://github.com/pandas-dev/pandas-stubs/issues/296...
mypy fix <https://github.com/python/mypy/pull/13500 > not yet released: https://github.com/python/mypy/issues/13385
onyx hinge
yes I should have linked that too. still don't grasp why a python upgrade would have triggered any difference though.
onyx hinge
oh this is why: https://github.com/python/cpython/issues/94996
analog bridge
Why do a hard reset in erase_filesystem()? Can't we just re-mount the new filesystem and soft reload?
void common_hal_storage_erase_filesystem(void) {
#if CIRCUITPY_USB
usb_disconnect();
#endif
mp_hal_delay_ms(1000);
(void)filesystem_init(false, true); // Force a re-format. Ignore failure.
common_hal_mcu_reset();
// We won't actually get here, since we're resetting.
}
sinful spruce
Yesterday I was testing "CircuitPython's Web API" and it had a very modern looking user interface. Latter in the day the user interface was changed. What is going on?
jaunty juniper
Yesterday I was testing "CircuitPython's Web API" and it had a very modern looki...
there is a simple web page hosted on the board, kept simple for size, but it can load an advanced interface from code.circuitpython.org if it has internet access, so that might be what you saw
you can access /fs/ or /serial/ for the basic access or /code/ for the enhanced version
blissful pollen
there is a simple web page hosted on the board, kept simple for size, but it can...
Not sure if it was related but I tried the web interface for the first time yesterday and received the simple interface, though the board had internet access. Maybe something was up.
But honestly I forgot to just try the code.circuitpython.org url.
onyx hinge
>>> import board
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: no module named 'board'
```well that's a curious error"are you sure you didn't mean beard?"
lone axle
@idle owl if you get a sec sometime can you create a repo under adafruit for https://github.com/FoamyGuy/Adafruit_CircuitPython_DisplayIO_FlipClock This was my latest project from Limor and I think it's pretty much ready to publish / add to the bundle.
sinful spruce
there is a simple web page hosted on the board, kept simple for size, but it can...
Thanks so much -- I tried /code/ and it worked.
lone axle
<@&356864093652516868> the weekly meeting is set to occur in about 2 hours from now at 2pm Eastern / 11am Pacific US time. If you plan to participate you can add your notes to the shared doc here: https://docs.google.com/document/d/1eqMwdLfb53MKA5T1Ck5D1lJQCDSs7naD7hZwnAid0-c/edit?usp=sharing We look forward to catching up with everyone participating.
proven garnet
One of the rare meetings I can talk!
tulip sleet
Why do a hard reset in `erase_filesystem()`? Can't we just re-mount the new file...
It was because it's possible/likely we need to re-enumerate USB, and that the host might be confused. It was designed to be used for repair.
manic glacierBOT
I have looked at the other boards in the stm port. All of them had the USB_VID=0x239A (Adafruit) and for USB_PID I have picked up one unused ( I taught :) )
Maybe it should be the VID from ST since it's a development board from them ?
onyx hinge
@tulip sleet can you advise on #6945? ^
idle owl
<@330227457296957440> if you get a sec sometime can you create a repo under adaf...
Unless you have a reason to keep the repository local to you, you can instead transfer ownership to the Adafruit org, and then it appears as-is on the Adafruit repo list. If you prefer to keep the original on your account, I can create a repo for you.
lone axle
Unless you have a reason to keep the repository local to you, you can instead tr...
I forgot about transfering repo's being an option. I gave that a try using this screen in the settings (pic)
But afterward it reports this error to me (error pic)
idle owl
@lone axle Transfer it to me first.
I'll transfer it to Adafruit.
Always worth trying, but if it fails, do the intervening transfer first. ๐
lone axle
<@382939733107408897> Transfer it to me first.
looks like that one did go through
tulip sleet
<@329766224093249548> can you advise on #6945? ^
I can answer re policy
onyx hinge
thank you Dan
idle owl
looks like that one did go through
lone axle
https://github.com/adafruit/Adafruit_CircuitPython_DisplayIO_FlipClock
Thank you! I'll get RTD set up for it this afternoon. iirc I am able to do that and transfer it over to the acct with the rest of the library docs on it.
and PR the bundle.
idle owl
Thank you! I'll get RTD set up for it this afternoon. iirc I am able to do that ...
Pretty sure you can do RTD, but there's one step you might not be able to do. Ping me while you're doing it do I can verify the steps you're taking, and if needed, complete the final step.
proven garnet
I think the hook is still good if you do it before transferring too, for future reference. I tested it on the iBeacon library and it still seemed to work after transferring over.
proven garnet
We'll fix it in prod!
idle owl
๐ uff. Terrible!
lone axle
I think the hook is still good if you do it before transferring too, for future ...
good to know, thank you. I had assumed it needed to be set up after transfer.
minor plume
We'll fix it in prod!
twitches
idle owl
@onyx hinge @tulip sleet I want your thoughts on the current pin situation on the Grand Central. Not sure it's worth putting in the weeds. Might make more sense to save it for later here.
tulip sleet
<@426200547024961539> <@329766224093249548> I want your thoughts on the current ...
sure, can ask later; I think it may have a straightforward answer
onyx hinge
pidcodes may not be appropriate unless the board's open source .. or am I confused?
tulip sleet
hmm, yes, though I'm not sure what else to do in this case
manic glacierBOT
@jepler @dhalbert I replaced the USB VID and PID with the ones generated by STM32CubeIDE, and it seems to work.
royal rover
@lone axle time codes
onyx hinge
https://towardsdatascience.com/python-3-14-will-be-faster-than-c-a97edd01d65d (I think it's tongue in cheek)
negative seconds ๐
modern wing
Yay to new reviewers! ๐ฆ
lone axle
<@382939733107408897> time codes
thank you.
idle owl
Tekktrik!
onyx hinge
I thought so but I wasn't sure in the moment ๐
proven garnet
Easy to do when you have a home field advantage ๐ Thanks!
modern wing
And thank you for being here part of it @idle owl ๐
onyx hinge
@idle owl this community we've built together has become the one I measure all other communities against.
thank you for being such a driving force
modern wing
<@330227457296957440> this community we've built together has become the one I m...
At the risk of referencing internet memes/lingo, "this so much". There's so much that this community does right.
idle owl
<@330227457296957440> this community we've built together has become the one I m...
Admittedly I do the same. ๐
austere acorn
I was away from the computer, so late thank you for the hug report, @tulip sleet !
And thank you everyone else as well for being so awesome! ๐
I never dared to contribute to open source before.
onyx hinge
@austere acorn keep it up! we can always use more help
August 18 so a bit past 1 month at this point
(djdevon3's pcb)
brazen hatch
What a "the silkscreen is free real estate" chad
proven garnet
@onyx hinge if you didn't come across it, check out LibCST. It works with visitors the same way I believe ast does, but as a concrete syntax tree it can keep track of things like whitespace and comments.
minor plume
Those NeXT keyboards were so great!
austere acorn
not out yet, don't ask?
thorny jay
Ok, and there is a RP2040 feature I did not understood.
They will find it from the guide I guess.
learn guide I mean.
gilded cradle
Sounds fine with me
lone axle
idle owl
@proven garnet So basically like this? py mqtt_client = MQTT.MQTT( broker=secrets["broker"], port=secrets["port"], username=secrets["aio_username"], password=secrets["aio_key"], socket_pool=pool, ssl_context=ssl.create_default_context(), )
proven garnet
<@135415127879974912> So basically like this? ```py
mqtt_client = MQTT.MQTT(
...
Yup!
thorny jay
Maybe in the "support matrix"?
USB Host on RP2040 is like that.
Wifi support in some board where there is a wifi chip but no support are like that too.
Could be in the release notes? Because it change from version to version... But no one is reading the release note before buying.
austere acorn
100% what Kattni said, I saw people complaining about it
thorny jay
Sometime I know about a limitation because I follow those meeting... but sometime I was surprise even with my involvement.
Warning the product page is only for product from Adafruit or resell by Adafruit.
errant grail
We need a "pretty pins" or badges approach to included features and CircuitPython modules.
jaunty juniper
in general the shop page should also say "Circuitpython support is in alpha there might be missing features and isntability" on S3 and C3 and such, like it used to say on the feather STM32
thorny jay
Don't forget the non Adafruit board...
austere acorn
Implementing what CGrover said would also help with the M0 limitations
idle owl
Don't forget the non Adafruit board...
That's harder for us to control.
austere acorn
I see a lot of people trying to use M0 boards, and getting frustrated when they get the memory errors
idle owl
We don't have those boards, so testing the features on all of them is not possible.
austere acorn
Since they don't know CP is limited
idle owl
I see a lot of people trying to use M0 boards, and getting frustrated when they ...
Yeah, we're doing our best to push folks towards other MCUs. But as long as we're selling M0 boards, folks will buy them.
modern wing
We need a "pretty pins" or badges approach to included features and CircuitPytho...
Badges like bluetooth, wifi, LCD screen onboard, stemma qt, etc..
gilded cradle
Thanks
austere acorn
Yeah, we're doing our best to push folks towards other MCUs. But as long as we'r...
... and expect to use all the CP libraries at once too ๐
modern wing
Thanks!
idle owl
... and expect to use all the CP libraries at once too ๐
Exactly!
errant grail
Thanks!
minor plume
Thanks, everyone!
idle owl
That one we've gotten through to most folks, but it's not 100% prevented. Intermittent new folks still try to use the whole bundle.
minor plume
... and expect to use all the CP libraries at once too ๐
You mean an M0 board doesnโt have a terabyte of flash? ๐
blissful pollen
Was text only, but had a question I think was talked about during in the weeds: the BLE support on espressif is not fully complete? Was briefly looking at it last night and wanted to confirm
jaunty juniper
it doesn't support starting a service, which is most uses
austere acorn
You mean an M0 board doesnโt have a terabyte of flash? ๐
I did see someone complaining that it's like 90's limitations in MCUs, why don't they just add more memory ๐
thorny jay
Badges like bluetooth, wifi, LCD screen onboard, stemma qt, etc..
So if you have badge, you need a badge that say "no support" another one that say "full support" and one that is "partial".
jaunty juniper
want an HID device ? start a service. Want to be able to connect from the Bluefruit app ? start a service !
minor plume
I did see someone complaining that it's like 90's limitations in MCUs, why don't...
โWhy donโt they just add more memory? But I still want the board to only cost $3โฆโ ๐
thorny jay
Was text only, but had a question I think was talked about during in the weeds: ...
So I think S3 has limitation.
jaunty juniper
Was text only, but had a question I think was talked about during in the weeds: ...
but it can send advertisements and connect to existing services
onyx hinge
dangit I used to study french and I still never made the Glaude / Chaude connection
jaunty juniper
for example the adafruit_ble_radio library works fine between an S3 and a C3 board
or you can connect an S3 to a NRF52 running a UART service
(I did that as a demo for a reverse USB-BLE-UART adapter)
blissful pollen
Thanks. I noticed I couldn't create an UART service and then saw it wasn't implemented. Wasn't sure how much was done. Probably don't have time to play with it now anyways
thorny jay
dangit I used to study french and I still never made the Glaude / Chaude connect...
That is going to help me explain. Or "Claude Francois" if that singer is know in the English Speaking World? (also Claude is a first name, so frequently I have to explain it is with a G, that is why I write David GLAUDE so that I try to make it obvious where is the family name).
minor plume
but it can send advertisements and connect to existing services
I will say this is one frustration Iโve had when starting projects - itโs hard sometimes to figure out which boards will support all of the things I need for a given project. But short of building a web app that lets you pick the specific capabilities and bundle libraries you want to use and then tells you which boards support those features and have enough flash for the libraries, I donโt know how youโd solve it. And building/maintaining such a configurator app is a substantial undertaking.
thorny jay
Back the "caveats" (is that the name), the first thing would be to list the most common mistake/complain from people that buy a board to later discover what they want to do is not supported.
idle owl
Back the "caveats" (is that the name), the first thing would be to list the most...
Yes, caveats is correct.
thorny jay
The badge idea, I suggested that to the support for sensor, to have an Arduino logo and/or a CircuitPython logo. Because I have a very few number of I2C sensor not supported yet, or not supported at acquistion time. But if I had listen carefully (and not forgotten) to Limor video, I would have know.
For the "WIO terminal"(?), I made the edit to the board to put a warning on the absence of Wifi support. But despite that, I had a friend that got it and was "sad" it was not there.
jaunty juniper
I will say this is one frustration Iโve had when starting projects - itโs hard s...
I have been thinking for a while of adding an automatic list of supported features per port to the matrix (or a new matrix) by parsing what modules are on each board for this port and then showing "all" "not supported" or "some", this could be added to the modules list already computed on every release and shown on cp.org
And I was thinking of adding a manual list of caveats/additional info to that, so that each module can be indicated as full or partial support (and "more info" on things like the alarm module limitations on pull ups vs down etc.)
thorny jay
So you have caveats per MCU: "On that chip we cannot do X (totally/partially/at all)."
tulip sleet
dangit I used to study french and I still never made the Glaude / Chaude connect...
it only sunk in when I heard how David pronounced it; somehow i thought it had a Germanic pronunciatiation for the "au"
jaunty juniper
the problem being maintaining the manual part
thorny jay
Then you have caveat per "module" where a module is not 100% full on some platform. (BLE is there, but not X or Y).
minor plume
I have been thinking for a while of adding an automatic list of supported featur...
I think itโs a good idea. Just challenging to keep up to date on the caveats.
tulip sleet
we could parse for .. note or .. warning or some other stock thing in the rtd docs
minor plume
But maybe itโs a case where imperfect information is still better than no information.
thorny jay
Yeah, it must be automated and I think in the Download page of CircuitPython.org, a bit like the support mattrix. Also see: https://github.com/adafruit/circuitpython-org/issues/1052
blissful pollen
the problem being maintaining the manual part
Maybe oversimplification but even looking for the Not Implemented exception being in a shared-module folder would be a strong hint
jaunty juniper
we could parse for `.. note` or `.. warning` or some other stock thing in the rt...
that would make the relevant info closer to the problem, and easier to maintain
tulip sleet
the rtd text would have a standard way of of listing the flagged ports and or boards, using the standard names. I don't think we have to specify what's missing in the circuitpython.org lists -- that will be in the doc when you link to it. Just flag that you should take note
because the actual restrictions can be complicated
thorny jay
Maybe oversimplification but even looking for the Not Implemented exception bein...
Sometime the "Not Implemented" is missing like: https://github.com/adafruit/circuitpython/issues/6527 but if there is a Not Implemented in the module, I guess that is true.
lone axle
Here is the notes document for next Mondayโs 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 youโll be attending the meeting - itโs super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and weโll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1p3AthJvTLLbLhqsiL6_ZHBQkj7CUSQ_IOJdc9Dnu4Hw/edit?usp=sharing
turbid radish
@lone axle - are you working on the Newsletter PR?
lone axle
<@382939733107408897> - are you working on the Newsletter PR?
pushed the changes and then didn't make the PR. Sorry about that. It's created now: https://github.com/adafruit/circuitpython-weekly-newsletter/pull/384
turbid radish
thanks!
idle owl
@onyx hinge @tulip sleet Will you both be around tomorrow?
languid whale
Was text only, but had a question I think was talked about during in the weeds: ...
There are several of us waiting for ESP32S3 BLE support. It's been a very long time coming...
tulip sleet
There are several of us waiting for ESP32S3 BLE support. It's been a very long t...
Good support requires changes in the nimble library we use internally. So it's not as straightforward as one would like.
onyx hinge
@idle owl yes should be available most of the day
jaunty juniper
Scott alluded to having NimBLE people working on adding dynamic service start to it, I don't know if that's a thing that is happening ? ( @tulip sleet @onyx hinge do you know ?)
lone axle
Pretty sure you can do RTD, but there's one step you might not be able to do. Pi...
I didn't get into this until a bit later than I had imagined. RTD Project is created: https://readthedocs.org/projects/adafruit-circuitpython-displayio-flipclock/ and I added adabot as a maintainer. From this guide: https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/sharing-our-docs-on-readthedocs it looks like the last step is the webhook which I don't seem to have access to settings page in order to set up, I must have been remembering one of my own community bundle repos for the ones I'd done before.
I think it's ready for the webhook setup. If there is anything else I need to do still let me know once you have a chance to get into it.
idle owl
I didn't get into this until a bit later than I had imagined. RTD Project is cre...
I will take care of the webhook. I think I have a backlog of projects needing it, so I'll try to get through them this week.
lone axle
Neither here nor there, but if there are any other folks here who looked forward to the PyDev of the Week posts that get linked in the newsletter. There hasn't been a new one for a few weeks but author indicates on Twitter that the posts will be returning, perhaps next week. https://twitter.com/driscollis/status/1574498928019415040
Twitter
main furnace
Is there a pure python wheel for Blinka? Big news for pyscript.
upbeat olive
Is there any reason Circuit Python would not work with the SAME51J20A Curiosity Nano Evaluation Kit? I have not tried, but was thinking of purchasing the kit esp if it can do circuit python.
tulip sleet
Scott alluded to having NimBLE people working on adding dynamic service start to...
he was discussing that with them, but I don't know how far it's gotten
tulip sleet
Is there any reason Circuit Python would not work with the SAME51J20A Curiosity ...
It would work, but would require a custom build, though it would not be hard to make one.
upbeat olive
It would work, but would require a custom build, though it would not be hard to ...
Thanks, good to know. I will consider buying it, if nothing else there are the regular methods of programming it, just looking forward to circuit python on it for some of the apps already available. ๐
main furnace
Not there yet.
Couldn't find a pure Python 3 wheel for 'adafruit-platformdetect>=3.13.0'.
languid whale
Good support requires changes in the nimble library we use internally. So it's n...
Thanks, Dan. ๐
clear halo
Hey folks, the September Melbourne MicroPython Meetup is on tomorrow evening (in Australia, 6:30pm AEST), I think some of you may like some of the content!
On the meetup page you'll find an agenda and links to the zoom stream, MicroPython discord and a site that displays the meetup time in your local timezone.
We will record and publish the sessions in case you can't make it. Hope to see you there!
onyx hinge
that causes a bit of churn.. 228 files changed, 1411 insertions(+), 2027 deletions(-)
manic glacierBOT
I think this is a good idea, since it canonicalizes the documentation.
One thing I do not understand is the removal of blank lines. Are you running black on the entire extraction of //| lines in each .c file?
Here's a typical set of changes, where blank lines between the defs are removed. This is not what black would do if they were all in one file:
Running black manu...
idle owl
@tulip sleet @onyx hinge So, Grand Central pin names. Turns out most of the pins are the D-names first (which has me confused about a different thing, but that's... a different thing). There is a chunk that is not. I'd prefer them to match the silk as the primary pin. I guess I wanted your opinions on it. This is what is affected. ```>>> board.D14
board.TX3
board.TX3
board.TX3```
I think it's a total of 6 pins.
onyx hinge
@idle owl if it's more consistent, go for it
tulip sleet
agree - make it like the silk. The order was probably done long ago before we realized what the order should be.
idle owl
Krad. Thanks!
onyx hinge
@tulip sleet I run black on each individual snippet of //| code, then string the results back together, so black doesn't have its normal memory from block to block that lets it know to add a newline between two 'def's
But also .. black has a distinct style for .pyi files!! ```jepler@bert:~$ cat bloo.py
class K:
def enter(self):
...
def __enter__(self):
...
jepler@bert:~$ black --pyi - < bloo.py
class K:
def enter(self): ...
def enter(self): ...
reformatted -
All done! โจ ๐ฐ โจ
1 file reformatted.
jepler@bert:~$ black - < bloo.py
class K:
def enter(self):
...
def __enter__(self):
...
All done! โจ ๐ฐ โจ
1 file left unchanged.
it squishes things differently
tulip sleet
๐คฆ oy
so you can run black on a def inside a class definition and it indents properly??
I did see that it folded the ... onto the same line in a few cases. I didn't review all the files, but I reviewed about 50
onyx hinge
if the first line of the /|| region starts with whitespace, I add an if 1: to black's input, then take it off again
tulip sleet
nice hack
stuck elbow
by the way, is there a reason why we don't use pyi files?
seems like it would be a much better way of avoiding the annotations getting in the way than the import shenanigans
onyx hinge
seems like it would be a much better way of avoiding the annotations getting in ...
oh, in libraries? yeah, that might have been a better way to go...
tulip sleet
@timber mango so do you mean removing the annotations from the actual source and just doing them in a .pyi?
stuck elbow
<@456226577798135808> so do you mean removing the annotations from the actual so...
yes
or at least putting them in pyi files from now on for new code
tulip sleet
is that technique used by other projects to avoid adding annotations to source? I don't know of another use case besides ours
I know you are not fond of annotations
stuck elbow
mypy supports it
stuck elbow
I think that jedi -- and thus also jetbrains editors -- supports it too
idle owl
@tulip sleet @onyx hinge PR is in.
stuck elbow
not sure what visual studio uses
but considering how many python devs work at microsoft now, I would expect it to do the right thing too
stuck elbow
I know you are not fond of annotations
I am very fond of annotations, when they are used where needed, and when they are correct
tulip sleet
@onyx hinge I am still seeing removed blank lines in cases like this?
onyx hinge
@tulip sleet that removal is the "pyi style" mode of black. should I turn that off?
tulip sleet
I am very fond of annotations, when they are used where needed, and when they ar...
I apologize if I misunderstood you; I thought you didn't care for them
tulip sleet
<@329766224093249548> that removal is the "pyi style" mode of black. should I tu...
I would use the "standard". As long as it doesn't affect the readability of RTD, I don't mind.
onyx hinge
@tulip sleet so, remove the "--pyi" flag?
tulip sleet
no, I mean I would try to do what is done normally. If we are producing .pyi files, then --pyi makes sense.
stuck elbow
I apologize if I misunderstood you; I thought you didn't care for them
sorry, no offense taken
tulip sleet
<@329766224093249548> so, remove the "--pyi" flag?
I thought I guessed that ... would trigger pyi mode or something. Do you know why it's different?
tulip sleet
sorry, no offense taken
no reason to be sorry, it was probably an incorrect memory of mine from discord
onyx hinge
https://github.com/psf/black/issues/207 @tulip sleet I only know that there is a "--pyi" flag.
Black automatically formats *.pyi files in a package in addition to *.py files Stub files mostly follow the same convention with a few changes, as follows: If the body of a function or class is just ..., don't put it on a separate line, but just put it after the : Don't put blank lines between methods or functions, or between classes with an empty body Only put one blank line between other classes. Potentially allow longer lines. Typeshed's coding conventions currently allow infinite-length lines, but I'm not sure we should keep doing that.
stuck elbow
no reason to be sorry, it was probably an incorrect memory of mine from discord
I tend to argue about such things in emotional way, I'm working on that
tulip sleet
https://github.com/psf/black/issues/207 <@329766224093249548> I only know that t...
they don't give a ref for such differences, maybe just to be more compact? Not sure why anyone would care, since it's mostly programs that ingest .pyi's; I don't know that humans read them very often
onyx hinge
If we don't want the pyi-style that's fine, turning it off might be a bit less churn-y
tulip sleet
https://peps.python.org/pep-0484/#stub-files says:
Stub files are files containing type hints that are only for use by the type checker, not at runtime. There are several use cases for stub files:
we could run the generated .pyi files through black after they are generated, when we build the stubs
idle owl
@tulip sleet For the GC pins PR, you want each pin that has an alias to have space between it and other pins? Want to make sure I understand before heading down that path.
tulip sleet
<@329766224093249548> For the GC pins PR, you want each pin that has an alias to...
yes, when there are multiple names for the same pin, group them together, and put blank lines before and after
idle owl
Got it. Will do.
tulip sleet
If we don't want the pyi-style that's fine, turning it off might be a bit less c...
maybe try turning it off; I don't have a strong opinion on this; what do you think?
onyx hinge
I don't have a strong opinion either
tulip sleet
lol
I don't care about the churn; i am more interested about making it easy to write doc for a new module
idle owl
I don't care about the churn; i am more interested about making it easy to write...
Which one makes it easier?
Which direction I mean
tulip sleet
what do you mean by direction?
idle owl
Which choice. Maybe I misread, but it seemed like there were a couple of choices here to choose from.
onyx hinge
To black with or without --pyi? I do not have a strong opinion, except that the non-pyi style is closer to what we have today which means less re-learning
idle owl
Does running it with make it easier in the future though?
onyx hinge
but I have a feeling most people editing these docs will have changes that are caught by pre-commit. if they can run it locally then it's easy to apply the fixes..
it's like every other kind of 'black thing', it always just changes it for you .. if it's set up right
tulip sleet
https://github.com/microsoft/vscode-python/issues/13341 is an example of .pyi file editing using black --pyi (that's how they fixed the bug)
idle owl
it's like every other kind of 'black thing', it always just changes it for you ....
Right.
tulip sleet
ok, let's go with --pyi, since they are .pyi files, at least in some sense
idle owl
๐ (I'll pretend I helped.)
onyx hinge
OK thank you for deciding!
tulip sleet
Does running it with make it easier in the future though?
it will enforce whatever we choose, so ultimately doesn't matter ๐
idle owl
Hah ok
onyx hinge
the whole goal here is to NOT have to spend 2 minutes changing whitespace like Dan requested I do on a PR ๐ so now, 5 hours later, ....! ๐
idle owl
That's how to use time!
tulip sleet
@onyx hinge, ok now I am going to quibble about the trailing blank lines which you preserved ๐ If black --pyi would remove them, then let's remove them
onyx hinge
๐คฃ
tulip sleet
I think you can just roll back the last commit?
otherwise we are inventing our own style
idle owl
Anyone want to come solder this quad alphanumeric backpack for me?
tulip sleet
just make sure it is not upsdie down; I did that once
idle owl
I triple checked that ๐
There's writing on the front of the backpack, and they're supposed to be upright with that. Display decimals on the bottom.
There are also dots on the backpack, that you're supposed to line up with the decimals, but there's only two, and they're not actually lined up with the decimals. So... it kind of works?
So many pins though.
tulip sleet
i missed the decimal dots, yeah
idle owl
Also I have to clean off my work table to even get to my iron. ๐
It never remains clear for long.
Mostly because I haven't organised anything in here, and so it's all shoved off camera so it looks really nice in here, but really it's a monstrous mess.
tulip sleet
sometimes there is a lot of stuff on the floor behind my chair, you can't see that, or on the sides of the desk on the floor
idle owl
Right? The off-cam areas of this room are pretty much jam packed with stuff. Not much more can be stashed.
OK I should really go deal with this. I'm close to having everything together for this guide update, but one of those things is actually testing the code. Can't do that without an assembled backpack. So here we are.
manic glacierBOT
thorny jay
Something very strange with 7.3.3 on a QT Py M0: ```>>> import random
random.random()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "<stdin>", line 15, in print_repl_value
OverflowError: small int overflow
print(random.random())
0.411096```
Why does it OverflowError in the REPL??? What is that line number 15 in print_repl_value? So many questions... Does that deserve an issue?
stuck elbow
well, if you don't have bigint enabled...
idle owl
@proven garnet For your example that was added to the HT16K33 library, I see where you got the wonky licensing at the top of your example, but you should be using the SPDX format. Please update the existing SPDX to be you (not ladyada), leave it as MIT, and remove the "Author:" and "License:" lines. While you're at it, please change the example explanation comment to a docstring. Thank you!
Is there a way in CircuitPython to cycle through a series of characters and numbers without adding them all individually to something? Use case: 14-segment display backpack example, where it scrolls the alphabet, 0-9, and maybe some special characters. Arduino has a feature that does this, evidently. No idea if CP has anything.
@onyx hinge Is this ^^ a thing?
stuck elbow
Is there a way in CircuitPython to cycle through a series of characters and numb...
you mean like for c in "0123456789abc":?
idle owl
Um.... Maybe, but that's still typing them out individually.
This is what Arduino has: ``` for (uint8_t i='!'; i<='z'; i++) {
alpha4.writeDigitAscii(0, i);
alpha4.writeDigitAscii(1, i+1);
alpha4.writeDigitAscii(2, i+2);
alpha4.writeDigitAscii(3, i+3);
alpha4.writeDisplay();
delay(300);
}```
I was basically asking if this concept exists in CP.
stuck elbow
well, you can do that, but... it's ugly for c in (chr(i) for i in range(ord('!'), ord('z')+1)):
idle owl
Harder to read as a comprehension, I suppose.
stuck elbow
you have to explicitly convert between chars and ints, in C++ they are the same thing
idle owl
Ah fair enough.
@stuck elbow so... I tried to make it work, and it's displaying 1. only. I'm clearly missing something.
stuck elbow
>>> for c in (chr(i) for i in range(ord('!'), ord('z')+1)):
... print(c)
...
!
"
#
$
%
&
...
works for me
idle owl
huh.
Not in the display code, it works for me too.
This does not work. py for c in (chr(i) for i in range(ord('!'), ord('z')+1)): display.marquee(c)
idle owl
Hmm. Checking.
Tried adding sleep, no change. Tried adding a delay using the parameter to the marquee line, no change.
There's no show type thing being called in the existing example.
stuck elbow
maybe you want display.marquee(''.join(chr(i) for i in range(ord('!'), ord('z')+1)))?
idle owl
Hey!
idle owl
Even uglier, but it works!
idle owl
@stuck elbow The chr and ord bits are built into CircuitPython through Python? i.e. should I be looking in the Python documentation for an explanation of what they are doing...
idle owl
โค๏ธ Thank you!
stuck elbow
sorry, wrong lang
idle owl
Oh! Wrong that way. I thought you meant wrong programming language, and I was deeply confused for a moment.
Yeah, I would have struggled a bit to sort it out in French.
stuck elbow
no idea why the search engine gave me links in French
idle owl
Hmm. marquee seems to be a permanent loop. So you can't use it twice, apparently?
I guess I could use print and scroll together for the first one. The first one is a string.
That failed.
In a few ways. Sigh.
Code in play here: py display.print("Hello, world!") for count in range(7): display.scroll(count) time.sleep(0.2)
First, no matter what I provide to scroll, it results in this: code.py output: Traceback (most recent call last): File "code.py", line 27, in <module> File "adafruit_ht16k33/segments.py", line 220, in scroll File "adafruit_ht16k33/ht16k33.py", line 164, in _get_buffer IndexError: bytearray index out of range
Second, there appears to be no way to slow down the scroll, it jumps directly to rld!.
To clarify, line 27 is display.scroll(count) right now.
I tried it as py display.print("Hello, world!") display.scroll(various-values-here)
Same results both in the code failure and the display results.
It feels like scroll should have a delay parameter. Unless it's designed to jump to a different location.
Oh, hah. marquee has a loop param that takes a boolean, and if set to False, does not loop indefinitely. Which means I can do it twice.
Still want to know how scroll is supposed to work. ๐
proven garnet
<@135415127879974912> For your example that was added to the HT16K33 library, I ...
Will do!
gilded cradle
@idle owl for marquee, you can pass optional parameters to not have it do a permanent loop.
idle owl
<@330227457296957440> for marquee, you can pass optional parameters to not have ...
I did finally see that.
gilded cradle
Let me take a look at scroll. I can't remember what that did...
idle owl
Thank you!
gilded cradle
@idle owl maybe you wanted something more like this:
display.print("Hello, world!")
for count in range(7):
display.scroll()
time.sleep(0.2)
OK, that made it not crash.
But... it's still jumping right to the end.
Oh. Let me try that.
idle owl
Still w/o the 1, jumps to the end.
gilded cradle
Still w/o the 1, jumps to the end.
Oh, if you have auto write off, it will appear to do that unless you call display.show() at each step
idle owl
Oh so there IS a show()!
gilded cradle
yeah
idle owl
Does auto_write need to be off for marquee to work properly?
Oof, adding show into that for loop made it display weirdness.
gilded cradle
I'm not sure. It might handle that automatically.
idle owl
It's still not working how I picture it should. Instead it jumps to the end, then posts a bunch of ! before moving onto the marquee.
idle owl
Basically what marquee does. But maybe print and scroll aren't meant to work that way?
I want it to print Hello, world!, and then, scroll to the end of the phrase.
Thought it would be nice to show other features too.
gilded cradle
Ah, yeah. It never worked super well, which one of the reasons I wrote marquee
gilded cradle
๐
idle owl
Thanks for looking into this with me!
gilded cradle
No worries
idle owl
OK, this example is good.
gilded cradle
@idle owl a quick hack might be to add a bunch of spaces at the end of the string
idle owl
Eh, I don't want to point folks towards something that doesn't play nice.
gilded cradle
Ok, cool
idle owl
Showing how to create two marquees together, with the first one not looping is a good one to show.
gilded cradle
Yeah
idle owl
Let me ask you another thing.
gilded cradle
My main claims to fame with the segments part of that library are the marquee and adding multiple display chaining
sure
idle owl
This snippet writes 0 0 0 0 on each of the segments. py display.set_digit_raw(0, 0x2D3F) time.sleep(0.2) display.set_digit_raw(1, 0b0010110100111111) time.sleep(0.2) display.set_digit_raw(2, (0b00101101, 0b00111111)) time.sleep(0.2) display.set_digit_raw(3, [0x2D, 0x3F]) time.sleep(0.2)
Why are they written in such different ways?
gilded cradle
Oh and I added the set_digit_raw()
idle owl
How is that all doing the same thing.
idle owl
My main claims to fame with the segments part of that library are the marquee an...
Excellent claims!
gilded cradle
It's to show different ways to use the function, but you don't need to have that.
gilded cradle
Excellent claims!
Thanks
gilded cradle
Maybe it needs some comments added to make that clearer
idle owl
That property is confusing to me.
The set_digit_raw because I'm no good at binary or hex.
gilded cradle
That's why I have a diagram in the guide
idle owl
Oh, fair enough. That makes sense
I got more confused when I saw this code using it. Because each of those results in the same thing. ๐
gilded cradle
Hehe, I forgot about that
I think it was to show 4 different ways to do the same thing
gilded cradle
๐
idle owl
How would you comment it?
Best I can do is "This is four ways to display 0. It displays on each of the four digits" or some such.
gilded cradle
In order: 16-bit Hexadecimal number, 16-bit Binary number, 8-bit Binary Tuple, 8-bit Hexadecimal List
gilded cradle
Hold on, I'm editing...
idle owl
Ah ok
idle owl
Ah. I never would have noticed! ๐
@gilded cradle Thank you so much for your help with this. I'll add the comments to this new example, and I'll go back and include them in the simpletest example as well. Already making changes in the repo, easy to make more.
gilded cradle
Cool. You're totally welcome.
idle owl
Oh, two quick things. I'll tag you on the eventual PR. And Limor wants you to QA this guide update before she does. I'll ping you when I'm ready for that.
gilded cradle
Ok, sounds good.
idle owl
Thanks again!
gilded cradle
Thanks again!
you're welcome
idle owl
@gilded cradle Here's the PR. It's getting pretty late here. I still haven't eaten. So I'm heading out. I can fix any change requests tomorrow. https://github.com/adafruit/Adafruit_CircuitPython_HT16K33/pull/107
gilded cradle
<@478616919721771019> Here's the PR. It's getting pretty late here. I still have...
Ok, I'll take a look
onyx hinge
@stuck elbow Optional static typing has at least two purposes. One is showing the contract that you, as a library, propose to the library's user. The other is in type checking all or substantial parts of a whole program to "prove" (term used loosely) that type errors will not occur at runtime. I think separate ".pyi" stubs only do the first of those two things.
manic glacierBOT
stuck elbow
<@351136009922019331> Optional static typing has at least two purposes. One is s...
I think the contract is the responsibility of documentation, type annotations are not sufficient for this. And since the documentation tools read pyi files, the information from them is included, so they do contribute to the documentation of the contract.
lone sandalBOT
manic glacierBOT
ฮ do not have any boards that require serial-to-usb so I cannot test that, however I will test the pr artifacts on my beetle-c3 and pico and provide any findings (if any) on the pr thread.
(Re Re stdin:
n = runtime.serial_bytes_available
if n > 0 or self.stdin is not None:
i = None
if self.stdin is not None: # fake stdin
i = self.stdin # fake stdin
...@onyx hinge I am going out for a daily walk, but considering 8.0.0-beta.1, would you want to merge any part of the picow work in? Advantage is more testers, disadvantage is more testers ๐
onyx hinge
@tulip sleet mostly the state of it seems to be "it works fine for awhile, then it stops working". You can do useful stuff with it if you do the "reset when socket stops working right" dance, which is pretty much inevitable for socket code in CP anyway. I wish I would have time to rebase out all the commits that say "WIP" and might not even compile ๐ but it probably won't happen
tulip sleet
also I would want to get inhttps://github.com/adafruit/circuitpython/pull/6952. I am not trying to do this in the next hour or two, but I may start on the release notes (which take an hour or two to do and can always be added to easily).
onyx hinge
I should review that ๐ not enough hours in the day
tulip sleet
I think a simple squashing in rebase -i would be quick and safe
I know -- if you are on a roll, take your time. could wait for a few days
those are two things I think would be nice for beta.1 ; there's nothing else I see that is knocking at the door urgently
onyx hinge
OK. I just figured out one big "it stops working" cause so that's grand
now the ntp test goes for >500 iterations instead of stopping after 8
crimson ferry
is that related to the thing where internet stops working if you don't do it often enough?
I ping once a minute so that the once-an-hour http works
onyx hinge
@crimson ferry no, it would not help with that.
@crimson ferry can you do me a favor and see if that problem reproduces in micropython?
crimson ferry
yeah, I have one set up, haven't done much uP but will give it a shot
onyx hinge
that'd be quite helpful, thanks
onyx hinge
@tulip sleet should I merge that or do you want some actual testing done too? I don't know what to do to trigger the problems that were being reported in #6892 (I feel like there's a terminology problem or I'm mis-interpreting what I'm reading)
well, I'll let you decide now that it's reviewed
manic glacierBOT
Using artifact https://github.com/adafruit/circuitpython/actions/runs/3138352211 on raspberry_pi_pico
The bug is still there. It took 20 tries before I got it to show up.
It occured on a late connection.
In raw repl:
While it is still bugging, it doesn't go in stdin
idle owl
@tulip sleet Does it seem right to say that M0 boards don't have the adafruit_bus_device module included? I almost forgot to check this and would have not included it as a necessary lib in the guide. Or, looking at the dynamic screenshots, it's on there, so I probably would have remembered.
It would be nice to be able to filter the support matrix by boards that don't have a particular module.
No idea how to implement that. But it would be convenient.
tulip sleet
<@329766224093249548> Does it seem right to say that M0 boards don't have the `a...
Some do, some don't. The M0 express boards mostly do, so do the RFM non-express boards. but not Gemma, Trinket. Some Trinkey boards do also
idle owl
Some do, some don't. The M0 express boards mostly do, so do the RFM non-express ...
I saw there was a Trinkey with it. Ok, duly noted. So I need to note it in the guide regardless.
tulip sleet
No idea how to implement that. But it would be convenient.
I'll be Neradoc could do that. Could open an issue on that
idle owl
Ok, I'll open an issue, thanks for the suggestion.
@jaunty juniper Is this something you'd be interested in doing?
I'm filing an issue, and could assign it to you if you're interested. But I don't want to do that if you're not so much.
jaunty juniper
it's implemented, just add - in front
idle owl
Oh seriously?
jaunty juniper
I didn't document it because I don't know where to
jaunty juniper
I planned to come back to it but...
jaunty juniper
it also doesn't filter the regexp, since modules only contains letters and _, so you can put a RE
tulip sleet
I would say add another paragraph at the first arrow, or mention "-" in the gray text in the box
idle owl
Paragraph seems better to me.
Wow I am inordinately excited that I know this feature now.
jaunty juniper
that's my favorite kind of feature request
tulip sleet
thanks for including that -- I will use it!
idle owl
apt-get always requires sudo?
tulip sleet
`apt-get` always requires `sudo`?
Yes, you are installing things to the system directories
jaunty juniper
@idle owl there is only one Adafruit ESP32-S3 Feather guide right ? It should be updated to mention the 2 different models, I'll leave feedback to mention that, but I am not missing a "ESP32-S3 Feather with PSRAM" guide ? (the issue discovered in #help-with-circuitpython is that it links to the CP.org page of the nopsram version only)
idle owl
Yes, you are installing things to the system directories
Ah fair enough. I didn't remember ever typing it without sudo but wanted to make sure we weren't telling folks to do unnecessary sudos.
idle owl
<@330227457296957440> there is only one **Adafruit ESP32-S3 Feather** guide righ...
You are correct, one guide.
I saw that go by in the other channel.
jaunty juniper
ugh I clicked on a sub link in the page and I think I left the feedback on the wrong guide ๐
stuck elbow
I thought the best kind of feedback is when people thank you.
idle owl
General feedback, yes.
jaunty juniper
close second
idle owl
But guide feedback is a whole other beast.
thorny dove
Question about CPY core module i2cslave. In the list of core modules in https://circuitpython-jake.readthedocs.io/en/latest/docs/index.html is listed 'i2cslave'. However in the flashed version of CPY: 'Adafruit CircuitPython 8.0.0-beta.0-49-g14fc4a079 on 2022-09-14; Adafruit PyPortal Titano with samd51j20' there is no module i2cslave, but there are i2cperipheral and i2ctarget, both with the same named attributes. Someone can elaborate me on this?
thorny dove
@onyx hinge thanks to the BOT showing here activity of you. I opened this: https://github.com/adafruit/circuitpython/pull/6933#pullrequestreview-1124233781 and I learned something from your code, especially the 'Test script do_dns_ping.py. I had some code for ping'ing that never worked, however I implemented that part of your example mentioned (especially ```
info = pool.getaddrinfo("google.com", 80)
addr = info[0][4][0]
print(f"Resolved google as {addr}")
ipv4 = ipaddress.ip_address(addr)
