#circuitpython-dev
wraith crow
@slender iron Okay, I probably just missed the status line when I first started trouble shooting. I tried for a couple hours last night to get my crash to reoccur with no luck, so I'll grab a new build and move on :).
lone axle
Is it possible for user code to read other values out of .env file? it could be helpful for other credentials or even configuration options maybe if user code can access values from it.
lone axle
Awesome!
slender iron
it does do the file read so only do it once when loading the value
blissful pollen
Off hand I'm sure that would be useful for adafruit.io too
lone axle
Would it make sense in librariers that currently require the use of secrets.py to start adding code that will try to access what it needs from os.getenv() first before falling back to secrets.py?
slender iron
maybe wait until 8.x is stable
7.x doesn't have getenv iirc
@lone axle just pushed a version without the hard coded ip
it should require a password for /fs now too
I should be able to follow your stream
lone axle
CIRCUITPY_WEB_API_PASSWORD inside of .env is how to set the password that will be then used from the browser client?
lone axle
Nothing better comes to mind for me right away. I'll let you know if anything does though.
slender iron
np, we may need more config to enable/disable things too
lone axle
did you mention at some point Console / REPL being available over websocket or something similar for the web workflow?
slender iron
ya, that's my plan
not going to do it before I PR though
I still need to do put and delete for files
I'll do serial in a follow up
I'm already doing a lot in one PR
tulip sleet
@slender iron @idle owl we should start building 8.x bundles, but I can't remember / can't find what to change to enable that
jaunty juniper
<@252717193496756235> <@330227457296957440> we should start building 8.x bundles...
we are
slender iron
tulip sleet
we are
Thanks, it didn't start until last night. I looked yesterday here: https://adafruit-circuit-python.s3.amazonaws.com/index.html?prefix=bundles/adafruit/. Lat night is the first one.
jaunty juniper
fresh from the oven
calm aspen
@blissful pollen hi
I threw this together to help myself note the process. I have a build environment that compiles.
https://github.com/im-redactd/circuitpython-dev
I forked and tree'd the i2cperipheral bits here
https://github.com/im-redactd/circuitpython/tree/7.3.0-i2cperipheral
and I've been spending as much time as I can trying to learn the ins and outs of different ports. but I feel like I'm treading water due to the amount of work I have and little understanding
blissful pollen
I get that. Anything particular that's holding you up right now? Looks like a good start, I'd work on renaming the files (get rid of the copy) and as Scott mentioned in the issue remove any ATMEL specific code
So if you compare your ports/raspberrypi/common-hal/i2cperipheral/ directory to the at-mel ports/atmel-samd/common-hal/i2cperipheral/ the file structure should match
calm aspen
got it. I think what derails me the most is that I'm referencing the micropython implementation and it doesnt match any of the circuitpython nomenclature that I start getting more confused. I think as a start I'll try to figure out how to find the rp2040 equivalent to all the atmel/sercom calls in the construct method
blissful pollen
have you looked at the rp2040 dev guide (forget the exact name)?
calm aspen
only the hardware guide. ultimately this requirement is because I've built some custom hardware and need to i2c peripheral function
do you mean the RP2 SDK specifically ?
blissful pollen
Yes, brain is slow today:
https://datasheets.raspberrypi.com/rp2040/rp2040-datasheet.pdf
also
https://datasheets.raspberrypi.com/pico/raspberry-pi-pico-c-sdk.pdf
may have some information
calm aspen
thank you, I'll check out that C sdk guide
blissful pollen
I'd take a look at the MP SDK calls and try to find them in the SDKs. I can take more of a look later at it myself.
calm aspen
oh ok that's a great idea. walk back from that implementation in MP to the SDK and then rebuild
blissful pollen
Also quick hint
mp_printf(&mp_plat_print, "Testing print\n");
printf debugging. that will print something to your serial console.
grepping the code for similar functionality works great too. If you need more of a discussion on the overall layout between shared-bindings/shared-modules and such can always arrange a time for a quick voice chat if you want
calm aspen
thank you
misty garnet
circup throws an error if I try to install a module that already exists on the board:
File "C:\Users\ktritz\anaconda3\lib\site-packages\circup_init_.py", line 536, in extract_metadata
if loc > -1:
UnboundLocalError: local variable 'loc' referenced before assignment
also, it loaded an incompatible .mpy file
jaunty juniper
yeah that's a bad mpy file
misty garnet
7.3.0
slender iron
@calm aspen it may help to read the api docs for i2cperipheral if you haven't already. the common_hal functions should mostly match it
jaunty juniper
circup throws an error if I try to install a module that already exists on the b...
that's what happens when circup encounters a bad mpy file, I was looking at a fix to get a clear error message
misty garnet
I was trying to install adafruit_pioasm, does this mean that the bundle .mpy is bad?
jaunty juniper
you'll have to delete the mpy file on the board
misty garnet
I did and ran it again, but still got an incompatible mpy error
jaunty juniper
no, it doesn't read the bundle's mpy files
hmmmm
I'm starting to wonder if there might be an issue with the current bundle then ?
oh what version of CP are you running ?
oh
no there's an issue with the bundle
jaunty juniper
the mpy files contain the python source...
calm aspen
@blissful pollen I've cleaned up the i2cperipheral.c in an attempt to build a bare bones library. I'm running into compile errors that reference a lot of the variables defined in the shared-bindings. I've grep'd around trying to find how the atmel version handles these but I'm missing something
../../shared-bindings/i2cperipheral/I2CPeripheral.c: In function 'i2cperipheral_i2c_peripheral_make_new':
../../shared-bindings/i2cperipheral/I2CPeripheral.c:73:11: error: 'MP_QSTR_addresses' undeclared (first use in this function); did you mean 'MP_QSTR_address'?
73 | { MP_QSTR_addresses, MP_ARG_REQUIRED | MP_ARG_OBJ },
| ^~~~~~~~~~~~~~~~~
| MP_QSTR_address
jaunty juniper
did you make clean ?
calm aspen
no
jaunty juniper
it's required if you add strings
blissful pollen
Yup anytime I get a compile error that has MP_QSTR try a make clean
calm aspen
thank you
jaunty juniper
ok we have a bug in the bundle where single file libraries mpy contain the py source instead of the mpy file
and I know why, so I'm gonna fix that
though I still don't fully know why it worked before
calm aspen
code.py output:
running dir(i2cperipheral.I2CPeripheral)
['class', 'enter', 'exit', 'name', 'bases', 'dict', 'deinit', 'request']
done
progress
manic glacierBOT
calm aspen
I'm just stepping down the code. I'm at this line and wondering what to do.
the SDK call is this: https://raspberrypi.github.io/pico-sdk-doxygen/group__hardware__i2c.html#ga42571b2db5d1ed0dc083385c230dc9e8
Do I need to pass a struct containing the 2 pins to that or could I make the call twice. one for each pin?
slender iron
@calm aspen neither
you'll want the i2c_inst_t
looking at common-hal/busio/I2C.c may be helpful too
calm aspen
thank you for that.
I'm sorry if a lot of these questions are basic C.
for this call: i2c_set_slave_mode(self->peripheral, true, *addresses);
is that valid or should I do a check to see how many num_addresses there are and pass addresses[0] addresses[1] instead?
jaunty juniper
fix for the bundle's current issue of having invalid mpy files (for single file libraries)
https://github.com/adafruit/circuitpython-build-tools/pull/88
blissful pollen
thank you for that.
I'm sorry if a lot of these questions are basic C.
for t...
Generally the shared-bindings call will check (as it is common for all ports). I usually test with a few bad values to find out later
calm aspen
would a working construct be enough for a i2c client to see the peripheral on the wire?
blissful pollen
would a working construct be enough for a i2c client to see the peripheral on th...
Not sure on that if its waiting for a response that the rest of the API has to provide maybe not
manic glacierBOT
still rain
Hello,
I am having an issue with CircuitPython. I am interested in viewing the amount of memory I have available using the micropython.mem_info function. However, whenever I try running this function I receive an AttributeError.
While reading through the docs I noticed that it was stated that the micropython module may unavailable on some builds on CircuitPython. I was wondering if there were any alternatives to this function in the standard library in CircuitPython.
Thank you!
I'm on Sony Spresense running CircuitPython 7.3.0
ornate breach
Hello,
I am having an issue with CircuitPython. I am interested in viewing the...
You’ll want to use gc.mem_free
still rain
Oops wrong channel.
ornate breach
gc is garbage collection and you can use it to see used memory and free memory as well as a few other things
still rain
You’ll want to use `gc.mem_free`
This worked
calm aspen
@blissful pollen I2C addresses found: ['0x40'] !
blissful pollen
That's awesome!
calm aspen
this is great. on to the next method!
blissful pollen
I'll be around on and off all weekend so feel free if you're working on it to still ping me if you have questions I'll try to respond
calm aspen
I really appreciate it. and everyone else that's helped so far
languid whale
I really appreciate it. and everyone else that's helped so far
If you have a github site I'd be more than willing to test on the array of boards I have here, as time permits. 👍
wraith crow
I've pulled down CP from the github repository tonight and built it for the ESP32-S3 DEVKITC-1-N8R2 and I'm once again crashing the board into safe mode when I attempt a wifi.radio.connect. The terminal header indicates WiFi: No IP and I've spent some time trying to figure out how I managed to get past this last night with the UM FeatherS3 but haven't had any luck. I flashed an 7.3.0 version and connected it to the WIFI, thinking that if I had a recent connection maybe the microcontroller would somehow pick up the IP address when I re-flashed 8.0.0, but that didn't work either. I also created a .env file with the wifi credentials, but that appeared to hang the board and when I connected to the UART interface it looked like the board was in a boot loop of some sort.
calm aspen
if anyone is up and looking for something to peek at.
https://github.com/im-redactd/circuitpython/blob/7.3.0-i2cperipheral/ports/raspberrypi/common-hal/i2cperipheral/I2CPeripheral.c
I create the object fine. device = i2cperipheral. I then create request = device.request() and the code moves into more methods like the is_addressed and ack. I'm concerned that the request() object doesn't show its properties when I run dir(request). properties like is_read, address are not showing up there.
should they be? I'm trying to step through the code as I think it should work and sort out the library and I'm worried something is broken.
blissful pollen
if anyone is up and looking for something to peek at.
https://github.com/im-re...
Probably won't get a chance to look till a bit later tomorrow but if you can post a snippet of the python code you're running too I will (I'm not that familiar with i2cperipheral so the python code will help)
calm aspen
here's a gist of the code running on a pico to test. https://gist.github.com/im-redactd/1f53ad885fe108bf0bd5f56ba59db879
the other end is just another rp2040 board running an i2c scan for now.
timber mango
Basically, you are right. The only restriction is that ring buffer storage type must be kept unsigned.
Otherwiese this approach fails. See the case #6 in the code below.
#include <stdio.h>
#include <stdint.h>
#include <stdbool.h>
int pop_unsigned(bool error)
{
if (error) return -1;
else return ((uint8_t)255);
}
int pop_signed(bool error)
{
if (error) return -1;
else return ((int8_t)255);
}
int main()
{
uint8_t u8_noerr = pop_unsigned(0); // (1) u8_noerr := 255 - ok
u8_noerr = pop_signed(0); // (2) u8_noerr := 255 - ok
uint8_t u8_err = pop_unsigned(1); // (3) u8_err := 255 - wrong
u8_err = pop_signed(1); // (4) u8_err := 255 - wrong
int i_noerr = pop_unsigned(0); // (5) i_noerr := 255 - ok
i_noerr = pop_signed(0); // (6) i_noerr := -1 - wrong
int i_err = pop_unsigned(1); // (7) i_err := -1 - ok
i_err = pop_signed(1); // (8) i_err := -1 - ok
return 0;
}
I'll try to take a look later today. Busy for a while this morning
proven garnet
Finally got around to finishing the manual cleanup of the .pylintrc patch so all the libraries in the bundle should be good
tall delta
anyone know how i can get only the files required to build for the raspberry pi pico?
calm aspen
I'll try to take a look later today. Busy for a while this morning
no problem! thank you
calm aspen
anyone know how i can get only the files required to build for the raspberry pi ...
that's not a thing. if you're building circuitpython you have to pull down the repo since it's modular.
the official guide is here: https://learn.adafruit.com/building-circuitpython - since I'm doing exactly that right now I put together a TL;DR version of that to quickly spin up a dev environment in a fresh VM. linked here if it helps: https://github.com/im-redactd/circuitpython-dev
jaunty juniper
make fetch-submodules will avoid fetching the bulk of the broadcom SDK, saving a few GB (compared to running git submodule update and truff), but it will still fetch the ESP-IDF for example
tall delta
hmm, is there not a "slimmed down" version or something?
tall delta
`make fetch-submodules` will avoid fetching the bulk of the broadcom SDK, saving...
how do I delete a majority of the submodules and only keep the ones required for rpi pico? is there a list of submodules required for it?
jaunty juniper
you could delete the other ports before fetching the submodules 🤷
tall delta
so just delete the directories?
wraith crow
I'm pretty sure I've deleted all the folders under ports except raspberrypi and built for the Pico without problem. I probably did the delete after the fetch-submodules but I'm guessing it would work (and make the fetch-submodules faster) if you did the delete first
manic glacierBOT
blissful pollen
no problem! thank you
Not sure why the request would not show anything. Did you try creating just like device = I2CPeripheral (board.SCL, board.SDA, (0x40, 0x41)) not in the with and then call r = device.request() and do all this in the REPL?
If you're still having issues i'll find a RP2040 board and build it myself to see
jaunty juniper
hey if there's anyone around, I'd like to bring up again the bundle issue: the Circuitpython bundles are currently unusable. I have a fix, if we could have a look at it ?
https://github.com/adafruit/circuitpython-build-tools/pull/88
blissful pollen
hey if there's anyone around, I'd like to bring up again the bundle issue: the C...
I can look at, do you need /want specific testing done with it? I'm less familiar with the build tools
jaunty juniper
Ideally local testing would be nice, especially if you can do it on linux
considering the changes, check that the bundles contain what they should, in case I missed something
blissful pollen
considering the changes, check that the bundles contain what they should, in cas...
So if I was building all the bundles I'd run this against a directory where I downloaded all the libraries? Just trying to figure out how to run it myself
jaunty juniper
you pip install -e . circuitpython-build-bundles (in a venv), checkout the bundle
run update-submodules.sh and the the same command run by the CI if I'm not mistaken
circuitpython-build-bundles --filename_prefix original --library_location libraries --library_depth 2 --package_folder_prefix "adafruit, asyncio"
blissful pollen
you `pip install -e .` circuitpython-build-bundles (in a venv), checkout the bun...
Where is the update-submodules.sh script from? I think I'm missing some repo to build all this
blissful pollen
Ah okay I was missing the name of the adafruit bundle repo, never used/built it before... cloning
@jaunty juniper Success! at least it built without errors. What was happening before that this was fixing? I'll check it is not happening now
jaunty juniper
if you look at the current bundle from the zip, single file libraries .mpy files actually contain the python source
so for example neopixel.mpy is actually a python file
blissful pollen
Just triple checking but they seem to be compiled mpy files now
jaunty juniper
that impacts all bundles and the learn guides project bundles
blissful pollen
I ran it on windows linux and built fine for me so I'll approve it
jaunty juniper
thanks 👍
lone axle
Thanks for the fix @jaunty juniper and testing @blissful pollen. I tested the PR branch as well and all seems good. I've merged that and made a new release of build tools. As far as I know the bundle build tonight should use the new one and tomorrows bundle should be fixed.
calm aspen
@blissful pollen same if I initiate it that way.
from i2cperipheral import I2CPeripheral
device = I2CPeripheral(board.GP5, board.GP4, (0x41, 0x40))
print(dir(device))
request = device.request()
print(dir(request))
only thing that request prints is class
blissful pollen
That's strange cause as far as I see request is fully within the shared-bindings/i2cperipherial directory. So assuming the request isn't None it should have the dictionary with it
calm aspen
I'm building a samd21 board now just to see what it looks like
blissful pollen
If you want to try to trace it the shared-bindings/i2cperipherial/I2CPeripherial.c file defines both I2CPeripherial and I2CPeripherialRequest
I'm looking on an ESP32S2 that I just happened to have hooked up
calm aspen
samd library has the same result.
calm aspen
@blissful pollen looking at the ESP32 implementation. it differs a bit from the samd in the is_address method.
I'm not entering that for loop so it never returns true. in the esp version they just return true since it supports only 1 address.
if I return true then I get the properties
misty garnet
I miss the RP2040 PIO channel 🙂
tall delta
I'm trying to modify this function in supervisor/shared/filesystem.c to change the code.py to a custom code.py (supervisor/shared/code.py created by me). anyone know how I could do this?
static void make_sample_code_file(FATFS *fatfs) {
#if CIRCUITPY_FULL_BUILD
FIL fs;
UINT char_written = 0;
const byte buffer[] = "print(\"Worked!\")\n";
// Create or modify existing code.py file
f_open(fatfs, &fs, "/code.py", FA_WRITE | FA_CREATE_ALWAYS);
f_write(&fs, buffer, sizeof(buffer) - 1, &char_written);
f_close(&fs);
#else
make_empty_file(fatfs, "/code.py");
#endif
}
in the same directory there is a code.py
i want to read the contents of it into buffer
FILE *f = fopen("a.txt", "rb");
fseek(f, 0, SEEK_END);
long fsize = ftell(f);
fseek(f, 0, SEEK_SET);
char *buffer = malloc(fsize + 1);
fread(buffer, fsize, 1, f);
fclose(f);
buffer[fsize] = 0;
this is my current implementation
but how do I port it to the rpi pico so I can add it to the filesystem.c?
jaunty juniper
you would need to make your file into a C structure that you can include, or have some pre-processing that does that
jaunty juniper
remember the C code is compiled it is not run here, there is no such thing as "supervisor/shared/code.py" when the code is later run on the board
tall delta
oh yeah.. you're right, what preprocessing would I need to do then?
jaunty juniper
hmmm I don't know
the simplest is to put the whole content of the file into a string
but what do you want to do exactly ?
tall delta
is there a python or c library that converts a file into a "c compatible string"? like for example,
print("test")
hello()
to
char buffer[] = "print(\"test\")\nhello()";
it would be easy to write such a python converter
tall delta
but what do you want to do exactly ?
well I'm trying to make a framework and I require a final uf2 to be compiled and generated with a custom code.py
tall delta
it would be easy to write such a python converter
does anything exist as of now? or would i have to create one?
jaunty juniper
one thing you can do is add your script to the frozen modules and just have a simple code.py that imports it
tall delta
where do i add the script in the frozen modules?
jaunty juniper
you make a subdirectory in frozen, and look at the mpconfigboard.mk of the Circuitplayground for example to see how to add a frozen module
tall delta
alright!
tall delta
you make a subdirectory in frozen, and look at the mpconfigboard.mk of the Circu...
where exactly is this file?
tall delta
ah, thank you!
jaunty juniper
by that I mean FROZEN_MPY_DIRS etc.
tall delta
yeah, got it! thanks!
WARNING:root:Unexpected echo. Expected b'%Run -c $EDITOR_CONTENT\r\n', got b'raw REPL; CTRL-B to exit\r'
know why I'm getting this error in thonny?
tall delta
WARNING:root:Unexpected echo. Expected b'%Run -c $EDITOR_CONTENT\r\n', got b''
i get this error after compiling it for the raspberry pi pico, the raspberrypi pico build that i downloaded worked fine
jaunty juniper
the Adafruit bundle is fixed, is there a way to do a manual release of the community bundle ?
misty garnet
will circup pick it up automatically?
jaunty juniper
yup
circup would still need a fix if you already have a bad mpy on the board
(but you can just delete it, then reinstall with circup)
manic glacierBOT
misty garnet
circup would still need a fix if you already have a bad mpy on the board
Is there a way for curcup to install the py version instead? It was actually useful having quick access to the source
misty garnet
👍
lone axle
the Adafruit bundle is fixed, is there a way to do a manual release of the commu...
I'm not sure if possible / how to trigger it manually. But I've updated and released NVM_Helper which is in the bundle so it should auto generate one tonight.
jaunty juniper
I'm not sure if possible / how to trigger it manually. But I've updated and rele...
nice, we'll be good then 👍
manic glacierBOT
manic glacierBOT
manic glacierBOT
jaunty juniper
Community bundle released, now all that remains is single-file libraries that were released during the bug to their own repo I assume
onyx hinge
@jaunty juniper if it's convenient for you please file a bug saying we need tests of the script that led to these problems and assign it to me. It should be doable. (I'm mobile right now or I'd do it myself)
manic glacierBOT
lone axle
@slender iron Curious if you have thoughts about how best to store the pages that will be served for the web workflow? I've got a local project with the /edit/ page HTML and a python build script that minifies it. I'm copy/pasting it from there into the string in the core code.
Should I make a repo from that project to have a central place to house the served pages? Or those could go in the core repo somewhere, or elsewhere?
I think there is some opportunity for an index page served at the root to contain some welcome message and link to the /fs/ page as well as links to documentation and other helpful resources.
jaunty juniper
<@199811382793863168> if it's convenient for you please file a bug saying we nee...
like an issue on the built tools repo to add tests to the CI ?
onyx hinge
The tools repo is where the test would need to be I think
So that it can find a problem before the pull request is merged
slender iron
I've pulled down CP from the github repository tonight and built it for the ESP3...
I've seen this some too. It's a TG1 WDT reset. I haven't figured out why.
slender iron
<@252717193496756235> Curious if you have thoughts about how best to store the ...
I'd have a python script generate a c file that gets compiled. Kinda like get_display_resources.py iirc
in tools
I was thinking we could do a new repo for the "online" files and we can direct a subdomain of circuitpython.org to it
maybe local.circuitpython.org
<@&356864093652516868> Just a reminder that the meeting is tomorrow. 25 hours from now. Today is the observation of Juneteenth here in the US.
lone axle
I see. So a function like this https://github.com/FoamyGuy/circuitpython/blob/6eaa1e5adbc74ff07b63678a555c1c370ef69e9c/supervisor/shared/web_workflow/web_workflow.c#L559-L566 could get refactored to it's own C file and that file can be generated by a script that will minify and/or gzip the page and embed it within the function.
lone axle
should the un-minified .html file live inside of tools/ with the script or somewhere else?
lone axle
okay, I'll keep it in it's own place for now. I will tinker with it some more later on today.
slender iron
thanks!
digital shoreBOT
turbid radish
<@&356864093652516868> Just a reminder that the meeting is tomorrow. 24 hours from now. Today is the observation of Juneteenth here in the US.
frozen maple
Hi @tulip sleet,
I'm the one trying to wrap my head around the BLE central implementation for HCI. 🙂 I have a quick question about CircuitPython code in general: how do the C function names get translated into Python methods when CircuitPython is compiled? For example, in devices/ble_hci/common-hal/_bleio/Adapter.c I was able to alter the message that is raised with mp_raise_NotImplementedError on line 467, and compile the code, put it on my Feather RP2040, and see the message returned when I try to call adapter.start_scan() from the REPL. But I'm still unsure of how the C function common_hal_bleio_adapter_start_scan gets turned into the Python method start_scan().
Sorry if my questions are rather basic.. but I really want to dive into this stuff more deeply. Thanks for any guidance you can give me!
tulip sleet
Hi <@329766224093249548>,
I'm the one trying to wrap my head around the BLE cent...
The Python interface is in shared-bindings/. The routines there call the common-hal routines, which are either in port/*/common-hal, or shared-modules/, or in the case of HCI, in devices/.
frozen maple
The Python interface is in `shared-bindings/`. The routines there call the `comm...
Thanks! I'll study that
tulip sleet
Thanks! I'll study that
also check out https://docs.circuitpython.org/en/latest/, and these subcategories in the left sidebar
onyx hinge
hey y'all! I was hoping to join the meeting today (tonight for me) but the wifi at this place we're staying isn't up to the task of voice calling. I've had a great big vacation with my spouse over the last few weeks but now I'm looking forward to picking up my Adafruit work again next week. Group hug, as I also wrote in the meeting notes.
tulip sleet
hey y'all! I was hoping to join the meeting today (tonight for me) but the wifi ...
get that last piece of vacation in! see ya!
onyx hinge
the fun's all done, it's just packing & catching flights now
tulip sleet
the last pastry 🙂
onyx hinge
they do make a good pastry all over europe
thorny jay
There was a meeting reminder 24H ago, but not recently (AFAIK).
idle owl
<@&356864093652516868> Weekly meeting in about 7 minutes! Please add any hugs and updates to the notes doc before we get started! Thanks! https://docs.google.com/document/d/18sH6UlmzVw8ObNTyvUt9biDsBoT6tYAAZ9LwRwhQXow/edit?usp=sharing
modern wing
Happily lurking again this week! Hope everyone is doing well. 🥧
candid sun
text only this week
thorny jay
@slender iron I need to catchup on the Web workflow... at one point I was looking for a CP web server (to host a single file like this: https://learn.adafruit.com/wordle-personal-esp32-s2-web-server) but failed. Will it be possible to have a web server for CP code purpose? Any way to interface, like CGI?
slender iron
@thorny jay I haven't thought about hooking user code into it
it will serve html from the filesystem by default
thorny jay
<@681528999520108564> I haven't thought about hooking user code into it
It could be "fun" to click the web page and change a GPIO value. 🙂
slender iron
it's been june-uary here 🙂 🌧️
old smelt
I was just going to roll with January, @idle owl
modern wing
27th month of March 2020 😕
errant grail
The Hallmark channel is preparing for it to be December, so it's all okay.
modern wing
The Hallmark channel is preparing for it to be December, so it's all okay.
No joke, I saw Christmas holiday decorations in my local arts & crafts store.
ornate breach
No joke, I saw Christmas holiday decorations in my local arts & crafts store.
And I’m here for it because I love Christmas 🎅
But I do enjoy Halloween 🎃
And Thanksgiving a lot
So many great days for great circuitpython powered hacks
errant grail
Santa will sunburn pretty badly making the rounds in July.
modern wing
But he'll love the southern hemisphere this time of year.
ornate breach
Which always made me wonder why Santa did the whole earth rather than just waiting until winter in each hemisphere 🙂
Save time delivering presents, especially since like 80% of the earth’s population lives north of the equator
thorny jay
This seems to be text only day. 😦
blissful pollen
This seems to be text only day. 😦
I was going to try to listen but forgot my headphones 😢
thorny jay
Octopus, I had Octopus G&W!
slender iron
🦷 time for me! thanks for running the meeting kattni!
modern wing
Thanks everyone -- another meeting in the bag 🙂
candid sun
thanks for hosting Kattni
thorny jay
Is that a record time?
modern wing
Is that a record time?
CircuitPython meeting speedrun, 100% no-glitch.
time is wibbly-wobbly anyways
proven garnet
@idle owl just finished listening to the meeting! I'm happy to own the Dependabot introduction. I think the best implementation with the fewest hiccups is to pin major revision as I put it in the notes.
Changing a library to use features added in minor revisions would mean changing the requirement manually anyways like it does now.
But I'm definitely happy to take this on
idle owl
But I'm definitely happy to take this on
Great! Then I'm ok with it. Let's start with a small set, as usual, and see how it goes. Maybe logging and the libraries involved with logging would be a good place to start.
slender iron
@lone axle you should check out code.circuitpython.org too. It uses codemirror 6 for editing
lone axle
I will check it out. I looked into codemirror first but it seemed to need a build step. Ace was quicker to jump in and use, but def open to any of them.
Do you know what the effective limit of the size of thing that can be sent to _send_chunk() could be? As I added more features it eventually got to a point where the page doesn't load fully any more in the browser. I need to figure out how to troubleshoot it better, but I'm thinking it's not receiving the full page if it's too large and it gets sent all in one chunk. Does that seem like a reasonable assumption? I tried briefly splitting to multiple _send_chunk() without success but intend to tinker with it some more still.
proven garnet
Also, @tulip sleet to your point we could use Dependabot for checking if Blinka updates which I assume is synonymous with CPY changes
So like pin libraries to the major version
e.g., Adafruit_Blinka==7.*
tulip sleet
I don't understand what you mean if Blinka updates. I was thinking about building separate 7.x and 8.x bundles, for instance, based on certain versions, but I didn't think Dependabot would help that situation. E.g., we add or remove some API, and want to hold back the change from the 7.x bundle but not from 8.x. For example, I made Pin hashable in 8.x. That could be used in the CP library, but wouldn't work on 7.x. Would also have to hold it back when we update the frozen libs in, say, 7.3.x builds
proven garnet
Not sure what value that would give compared to the dependent libraries (my thoughts and intention were what @idle owl described, sorry I wasn't at the meeting!)
Ah I see, got it
tulip sleet
i was just confused during the meeting about the purpose
proven garnet
Actually, on that note, the only remaining thing to think about is whether we want to pin min versions of Adafruit_Blinka. There's definitely value in it, but it also means a looooot of PRs to sift through
tulip sleet
we could do that on demand when we make some Blinka change that requires it
slender iron
@lone axle I don't know of a chunk size limit
proven garnet
That's true, I could explicitly ignore it in the config file
tulip sleet
but blinka versions can be required by the pip mechanism, hmm, hmm. I'd say let's wait for a use case
proven garnet
Since they all rely on Blinka, it should be easy enough to change it via a patch so I don't think it's important to have to get it perfect at initial introduction either.
manic glacierBOT
lone axle
Did the serial output that shows on built-in displays by default get scaled up for some builds recently in main branch? The text on my Titano is larger than I recall it I think.
proven garnet
Oh neat, you can actually use Dependabot for tracking revisions of things in the YAML workflow files (I think)
proven garnet
Probably not what we want, though. Probably would want to patch it as opposed to sift through hundreds of PRs
tulip sleet
@slender iron I made a bunch of 7.3.x PR's (mostly backports) which I think are everything we would want for a 7.3.1 release.
proven garnet
What are people's thoughts on getting a PR for a minor revision update? Might be nice to trigger optimize/update a library when new additions are made, but I feel like those dependents rarely change when it happens. My vote is to pin to major versions, and only pin to minor versions (manually, no auto-PR) when updates occur that require them like we already do (i.e., using a feature that was introduced in a dep_module v4.1.0 would go from dep_module~=4.0 to dep_module~=4.1
tulip sleet
What are people's thoughts on getting a PR for a minor revision update? Might b...
LIke you say, a minor revision update would not force code that used the library to change. If a specific enhancement was done, then we probably already have the use case in mind, and would know the reason for the update. But finding other forgotten possible updates is something that would have to be done manually. But I think minor version updates happen so often, there would be too many PR's.
slender iron
Did the serial output that shows on built-in displays by default get scaled up f...
The font should have gotten denser. There were two extra rows of pixels in the font that we don't use. It was also rearranged.
slender iron
<@252717193496756235> I made a bunch of 7.3.x PR's (mostly backports) which I th...
k, starting to look now
tulip sleet
#6512, #6513, #6514, #6517. I wrote release notes already.
idle owl
@proven garnet Thank you for looking at the duplicate code check failures! I am unsure whether anyone has the cycles to do a full refactor on those libs. I think for now, we should rename these issues to "Code fails duplicate-code check, consider refactor" and then add the duplicate-code back to the pre-commit-config.yaml file. And, add a note to each of the issues saying something like, "A refactor is needed here. When complete, remove the duplicate-code disable from the pre-commit-config.yaml file."
Add duplicate-code disable only to these four libraries.
proven garnet
Most of my stuff I'm doing now is feature additions, so since this is probably higher priority I can shift over. I don't think it'll take me too long, but definitely a little bit to do.
idle owl
Refactoring can be pretty beastly. I'm not sure that's where we want you to focus your time. I'm almost certain if I look into it, the answer will be to punt it.
I appreciate the offer!
idle owl
Most of my stuff I'm doing now is feature additions, so since this is probably h...
Throwing a 🔧 into it.... Also they would all need to be backwards compatible. Heh.
idle owl
<@135415127879974912> Thank you for looking at the duplicate code check failures...
Either way, I think continue doing what you've been doing. And if you could do the stuff I suggested ^^, that would be amazing. With the issues updated, we at least know where it needs to happen, etc, once we do have cycles.
proven garnet
For sure, I'm on it!
idle owl
For sure, I'm on it!
Thank you!
slender iron
@lone axle did you start a landing page for the CP web workflow yet?
proven garnet
@idle owl can I just force push the previous commit to main? Or do you want to preserve the history of removing and readding?
idle owl
<@330227457296957440> can I just force push the previous commit to main? Or do ...
I mean, force pushing will still leave a history trail. I think that's fine. Easier than the alternative. As long as that was all that was in the previous commit.
proven garnet
Sorry I think I meant reset main to the previous
lone axle
<@382939733107408897> did you start a landing page for the CP web workflow yet?
Nope, not a landing page. Only edit.
idle owl
Sorry I think I meant reset main to the previous
Oh right. Either way.
proven garnet
My git terminology is limited to Commit, Push, and There Be Dragons
idle owl
My git terminology is limited to Commit, Push, and There Be Dragons
I remember being there. My list of what I know is longer, but I still have a lot of Dragons.
lone axle
I need to try out code.circuitpython.org still. I haven't used that on a BLE device yet. On the edit page I was working on I added a "help" button that will open up user to a docs page searching for whatever they have highlighted in the code. It seems like a nifty integration to get help with a specific part of your code quickly. Maybe eventually code.circuitpython.org can have a similar helper feature.
jaunty juniper
speaking of web workflow, I've been trying to hide blinka and the status bar with code like that and I get problems. It looks like supervisor_start_terminal called by reset_terminal doesn't resize to the dimensions passed, but reserves a line for the status bar
code like that works in 7.3.0
for pos in range(len(board.DISPLAY.root_group) - 1):
board.DISPLAY.root_group[pos].hidden = True
tile_height = board.DISPLAY.root_group[-1].tile_height
delta = board.DISPLAY.height % tile_height
supervisor.reset_terminal(board.DISPLAY.width, board.DISPLAY.height)
board.DISPLAY.root_group[-1].x = 0
board.DISPLAY.root_group[-1].y = - delta
@idle owl repos reset, and I templated what I'll put for the issue comment in the EPD repos issue
jaunty juniper
in 8 there's an empty line or when playing with possible values for height or y, other kind of issues like the live typing not showing up (screen updates only on enter)
idle owl
<@330227457296957440> repos reset, and I templated what I'll put for the issue c...
Great!
slender iron
@jaunty juniper ya, it will always have the title bar afaik. I didn't test the reset
trim elm
Hey @proven garnet Is there any chance that you may have deleted the commits removing the duplicate-code check from the libraries that it failed on? I recognize this is a kinda weird question but something really odd happened with some of the libraries that we opened issues on (the commits kinda just disappeared from the main branch)
Also @idle owl ^^
tulip sleet
@slender iron when you have time, I have a q about the CIRCUITPY_MODULE espressif wroom.c and wrover.c files. I was making something similar for the ESP32, but the pins mentioned in never_reset_module_internal_pins() don't seem to correspond to the pins mentioned in the module datasheets.
idle owl
Hey <@135415127879974912> Is there any chance that you may have deleted the comm...
Yes.
That is basically what happened.
There's a discussion above.
trim elm
Ah ok. That confused the heck out of me
slender iron
@tulip sleet I'm around now but not sure I'll be able to remember 🙂
tulip sleet
like for the WROVER
void never_reset_module_internal_pins(void) {
// SPI Flash
common_hal_never_reset_pin(&pin_GPIO27);
common_hal_never_reset_pin(&pin_GPIO28);
common_hal_never_reset_pin(&pin_GPIO29);
common_hal_never_reset_pin(&pin_GPIO30);
common_hal_never_reset_pin(&pin_GPIO31);
common_hal_never_reset_pin(&pin_GPIO32);
}
but the datasheet says
trim elm
There's a discussion above.
Ah ok now I see it. Thanks for clearing that up for me
tulip sleet
maybe the module pin number got mixed up with the GPIO number??
if so, I wonder why it works anyway, maybe it's not necessary
I will go back in time with git to see if it was ever changed
idle owl
@proven garnet Please add your template reply to the PyBadger issue as well. Thanks!
slender iron
@tulip sleet what datasheet are you looking at?
tulip sleet
<@329766224093249548> what datasheet are you looking at?
the wrong one :), ha, hold on
slender iron
wrover => esp32s2 wrover
tulip sleet
I was looking at ESP32 wrover (bad googling)
so 26-32 are the module pin numbers, corresponding to GPIO33-38
so the q still holds, I think
proven garnet
<@135415127879974912> Please add your template reply to the PyBadger issue as we...
Done!
tulip sleet
@slender iron I think the pin labels are not the QSPI pins, despite FSPIHD, etc. labels. Pins GPIO27-GPIO32 are not exposed. OK.... sorry, this was really confusing. I think I understand now.
The ESP32 modules are quite different. The internal schematics are sort of incomplete
slender iron
np
dusk gull
Due to supply chain issues I am changing from the MX25L51245G 64MB flash to Winbond W25Q512JVEIQ, and I think I wrapped my head around editing the definition files for the flash chips... but I am having troubling knowing where to actually add this file so I can compile it. It looks like these use to be in the devices.h file under supervisor, now it seems like it is under nvm.toml
What seems unclear is when defining this in the board file, I do not see how this references the toml files because naming does not match.
jaunty juniper
Due to supply chain issues I am changing from the MX25L51245G 64MB flash to Winb...
you would put it right next to nvm.toml/flash/winbond/W25Q128JVxQ.toml like "W25Q512JVxQ.toml" I assume ?
and then EXTERNAL_FLASH_DEVICES = "MX25L51245G, W25Q512JVxQ"
dusk gull
you would put it right next to `nvm.toml/flash/winbond/W25Q128JVxQ.toml` like "W...
One reason in a recent build "EXTERNAL_FLASH_DEVICES = GD25S512MD" worked but I do not see where that directly references with toml files. And it does not exist in devices.h
dusk gull
and then `EXTERNAL_FLASH_DEVICES = "MX25L51245G, W25Q512JVxQ"`
So you can put multiple flash options?
jaunty juniper
it's there: nvm.toml/flash/gigadevice/GD25S512MD.toml
yup, many boards have different options for flash, especially with the chip shortage
it finds the chip by SKU inside the {technology}/{manufacturer}/{sku}.toml path
did you type 3 x or something ? bot's a little trigger happy
dusk gull
did you type 3 `x` or something ? bot's a little trigger happy
Not sure what happened.
When I referenced the flash in the past it did not need quotes, is quotes needed only when mulitple are suggested?
jaunty juniper
I don't think the quotes are needed, but don't take me as a reference on makefiles, I just look at the other boards, what's there should work
❯ ag "EXTERNAL_FLASH_DEVICES = .*,"
ports/stm/boards/stm32f411ce_blackpill_with_flash/mpconfigboard.mk
8:EXTERNAL_FLASH_DEVICES = GD25Q16C,W25Q64FV,W25Q32JVxQ,W25Q64JVxQ
ports/nrf/boards/challenger_840/mpconfigboard.mk
9:EXTERNAL_FLASH_DEVICES = "W25Q16JVxQ,W25Q32FV"
ports/raspberrypi/boards/arduino_nano_rp2040_connect/mpconfigboard.mk
10:EXTERNAL_FLASH_DEVICES = "AT25SF128A, IS25LP128F"
...
one of them doesn't use quotes, some don't have spaces 🤷
dusk gull
I don't think the quotes are needed, but don't take me as a reference on makefil...
Thanks, I will look at some for reference. I appreciate you clarifying, trying to make this update quick so I can test a new build and needing to quickly dump a new memory chip on has been a big distraction.
Just need to figure out where to modify this in my build cycle. Looks like nvm.toml is sub module'ed... I need to dump an edited file into the cloned repo before the build
jaunty juniper
yeah the downside of nvm.toml is having to PR two repos
dusk gull
yeah the downside of nvm.toml is having to PR two repos
Yeah, I need to replace a file in my dockerized build system
jaunty juniper
@slender iron @lone axle hey I might look into making a circup-like web interface to plug into the web workflow, choosing a library or reading code.py (or others), finding dependencies, and offering files to install. I already made a "bundler" that does most of that but makes a zip https://www.neradoc.me/bundler/ (though it required hosting the bundle zip on the same server, due to CORS restrictions)
lone axle
<@252717193496756235> <@382939733107408897> hey I might look into making a circu...
Nice! I haven't tried it yet but I think you could PUT request .mpy library files into the lib dir similar to the way the plain text files are saved.
dusk gull
Anybody know where the NVM toml files are when cloning and getting sub modules? I need to dump a custom toml for testing and not sure where to put it
jaunty juniper
in data/nvm.toml
oh I cut the data/ from the paths I gave earlier, I meant to give them relative to the CP root
dusk gull
oh I cut the `data/` from the paths I gave earlier, I meant to give them relativ...
NP thanks for being here, saving me a lot of grief.
orchid basinBOT
wraith crow
I tried building CP with the DEBUG=1 flag for the ESP32S3 Devkit and got a Too Little Flash message. Is there something else I need to do?
`Successfully created esp32s3 image.
5665904 bytes used, -3568752 bytes free in flash firmware space out of 2097152 bytes (2048.0kB).
70 bytes used, 8122 bytes free in 'RTC Fast Memory' out of 8192 bytes (8.0kB).
4112 bytes used, 4080 bytes free in 'RTC Slow Memory' out of 8192 bytes (8.0kB).
16383 bytes used, 16385 bytes free in 'Internal SRAM 0' out of 32768 bytes (32.0kB).
212148 bytes used, 213836 bytes free in 'Internal SRAM 1' out of 425984 bytes (416.0kB).
0 bytes used, 65536 bytes free in 'Internal SRAM 2' out of 65536 bytes (64.0kB).
Too little flash!!!`
jaunty juniper
you can disable modules to reduce the size of the build
wraith crow
@jaunty juniper Okay, I'll give that a try, thanks
lone sandalBOT
carmine girder
@tulip sleet + @onyx hinge + @slender iron : I am catching up on the fixes you have just done for https://github.com/adafruit/circuitpython/issues/6205 + https://github.com/adafruit/circuitpython/issues/6489 and just wanted to say THANK YOU!!!! I have been having sporadic issues with the Matrix Portal but always blamed myself for it. I can't wait to try out 7.3.1 and see how it behaves going forward. The best uptime I managed to see on it so far had been: ~19 days (27769 minutes)
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
proven garnet
Looks like Dependabot has a bug handling comparable releases (~=) which is the best way to define requirements, so aside from building our own mini Dependabot it may be worth just holding off until it's patched.
Though it did occur to me during the search we could probably reduce our requirements manifest to one file that setup.py reads and uses similar to how it does with the readme
I have an example here
That would prevent instances where requirements.txt and setup.py get out of sync regarding dependencies, which I've seen occasionally
daring pumice
Morning everyone,
Having so difficulty getting my Custom PCB to boot using CPy. It uses an nrf52840 SOC by Raytac (the same one as the feather sense).
I've been following the guide on Adafruit's site: https://learn.adafruit.com/circuitpython-on-the-nrf52/nrf52840-bootloader?gclid=Cj0KCQjw2MWVBhCQARIsAIjbwoPe63jZj8Boe6g0EKl_kvhiUnNlHDDgfDsS53FjktVWap3YRAXn4KQaAqOZEALw_wcB
I'm absolutely sure I'm doing something wrong, but not sure what.
Can anyone offer some counsel and/or guidance on how to proceed?
onyx hinge
2 "slice step can't be zero" | 'slice step cannot be zero'
I wrote a little script to find "similar" translation strings, based on "edit distance". Some of what it finds is not useful, but some is. Like this pair of message, which are just 2 edits away from identical.
onyx hinge
(apparently one is in ulab, and follows the "can't" convention of micropython, while the in cp we changed it to match desktop)
slender iron
I tried building CP with the DEBUG=1 flag for the ESP32S3 Devkit and got a Too L...
Did you get this working? I think I have a local change to fix this for myself
slender iron
Morning everyone,
Having so difficulty getting my Custom PCB to boot using CPy....
Please explain what you've tried and what is happening now
tulip sleet
@daring pumice do you have a regular nRF52840 Feather? If you could erase it and then add the bootloader and SD, and vet that, I think you can create a recipe that should work for your module.
daring pumice
Please explain what you've tried and what is happening now
Hi Tannewt - I will do that, much appreciated.
Just on a call right now but will put that together today. Would you prefer it be posted here or on GitHub?
daring pumice
<@447309542863273985> do you have a regular nRF52840 Feather? If you could erase...
I do, and that's a good idea. I'll try that after my call and get back.
Thanks again guys - sorry for the bad timing.
tulip sleet
@slender iron how did you generate the various config files in esp-idf-config/ I have a start for an esp32, but it has other sections, and I'm not sure where they came from originally
tulip sleet
thanks, I was doing some by-hand dedupe,
slender iron
python tools/update_sdkconfig.py --board=espressif_esp32s3_devkitc_1_n8 --update_all
without update_all it'll just do the board's sdkconfig
that's tools in ports/espressif
there is another script next to it too
tulip sleet
@slender iron thanks! got it
@slender iron one more thing: where is the repo to assign creation ids for adafruit? The link in https://github.com/creationid/creators/blob/main/README.md to https://github.com/adafruit/creations doesn't exist
tulip sleet
Our C3 board has CIRCUITPY_CREATION_ID = 0x00010001. But I like the numbering scheme you used in the Espressif repo: https://github.com/creationid/espressif-creations
slender iron
you can change it then 🙂
slender iron
just can't be all zeros because those are USB
tulip sleet
I'll make the repo, then, do you want it in adafruit org or creationid org?
(I don't have permish for the latter0
slender iron
anyone else have an esp32s2 tft feather?
none of my flashing is working
either direct or indirect with uf2
jaunty juniper
I don't have any particular issue with mine
slender iron
weird!
jaunty juniper
I did flash an S2 earlier and discovered I should not use a 921600 baud rate with esptool (that I somehow had in a copy pasted line), but that's all
tulip sleet
anyone else have an esp32s2 tft feather?
If i leave tio connected to the ACMx port, it interferes with esptool.py
slender iron
UF2 doesn't work for it either though
tulip sleet
but that is not true with UF2
slender iron
I tried the UF2 upgrade too
tulip sleet
did you start with an erase?
slender iron
I've got another board working so I'm not blocked
I started with a make flash
it did have the arduino example but now just goes into uf2
tulip sleet
I started with a make flash
if that does not do a full erase first, I'd try that; that's my only suggestion. (Also making sure the board is the board you think it is)
blissful pollen
anyone else have an esp32s2 tft feather?
I flashed one to whippersnapper and back last night no issues. To version 7.3.0
idle owl
anyone else have an esp32s2 tft feather?
I have one on my desk.
slender iron
I'll take a look later.
slender iron
yup, full erase fixed it
manic glacierBOT
proven garnet
Looks like Dependabot has a bug handling comparable releases (`~=`) which is the...
@idle owl not sure if you saw this but wanted to get your thoughts on a patch like this
proven garnet
It would probably have to be done with my tools but would consolidate everything into one file. Also still sad about Dependabot.
proven garnet
We're not building our own 😄
Haha figured
I'll stay posted to see if it ever gets fixed
idle owl
So we're also eventually planning to move away from setup.py which was apparently always legacy to the pyproject.toml. But there's no ETA on that, and it's apparently super involved. Someone at PyCon this year was looking into it, but I haven't heard anything since.
proven garnet
To the toml file?
idle owl
Yes.
proven garnet
Well Dependabot is currently a bust so happy to pick it up. Otherwise, I'll just work on some of the stuff talked about in other meetings
idle owl
@proven garnet I think it's kind of a bonkers update. If you want to look into the process and report back what you find, that would be an acceptable place to start. I don't want to send you off on a massive quest without knowing what's involved with it, because I know Limor has other plans for you. However, I don't know when she'll get to that since it will involve some guidance.
And it's also based on parts availability. So... very uncertain on timing.
devout jolt
Hi! I have a question about displayio.Display (and its subclasses like adafruit_st7789). I made the gist above for an ST7789 display on a RP2040, and the display seems configured as a 135x240 display instead of the 240x135. If I create the display with no rotation parameter and do continuous 135x135 sprite writes, my framerate is pretty good (about 225 msec per frame) but if I set display.rotation=90, then framerate drops to about 550 msec per frame. Is this a known attribute of Display?
idle owl
@lone axle You might have an idea here. ^^
lone axle
I'm not sure about performance impact of the rotation property. I don't recall hearing anything about it being limited, but I don't think I've ever measured it in different rotations.
The times I have used different rotations I didn't go back and forth much either so I don't really have a comparison between them mentally.
devout jolt
The times I have used different rotations I didn't go back and forth much either...
I just verified the same effect on FunHouse. By default it seems it is rotation=270 but if you set rotation=180 or rotation=0 it is MUCH faster
blissful pollen
I’m trying to recall but I think it does the rotation calculation which could be the slowdown. I’ve never timed it either. I currently don’t have easy access to my computer to check either (yay house repairs).
slender iron
@devout jolt rotation impacts the order pixels are read out of files
what are you displaying?
devout jolt
what are you displaying?
sprite sheet from OnDiskBitmap
I will try with vectorio stuff
slender iron
oh ya, that'll do it
displayio tries to always align its output into rows to the display
it's very slow if a row contains a column of pixels from a source bitmap
stuck elbow
you can rotate the display in hardware by changing the init sequence
slender iron
I tried not to do that so that the rows we write would align with the refresh order of the display
stuck elbow
specifically the MADCTL register
slender iron
I don't think displayio is fast enough for that to matter though
stuck elbow
well, if rotating 90 degrees makes it noticeably slower, that should fix it
(it will then be slower at 0 and 180 degrees, though)
slender iron
ya, agreed
stuck elbow
@slender iron would it make sense for the display.rotation to use the MADCTL register on the displays that have it?
devout jolt
okay nevermind I guess. I just verified on FunHouse (which has a square LCD to remove that as a variable) and using only in-memory vectorio shapes, and framerate is not dependent on rotation. It is the file read time. Here's the test sketch I used:
import time, random
import board, displayio
import vectorio, rainbowio
display = board.DISPLAY
maingroup = displayio.Group()
display.show(maingroup)
for i in range(20): # demo with a bunch of vectorio circles
palette = displayio.Palette(1)
palette[0] = rainbowio.colorwheel(random.randint(0,255))
x,y = random.randint(0,display.width), random.randint(0,display.height)
ball = vectorio.Circle(pixel_shader=palette, radius=20, x=x, y=y)
maingroup.append(ball)
display.rotation = 0
while True:
t = time.monotonic()
for ball in maingroup:
ball.x = (ball.x + 1) % display.width
display.refresh(target_frames_per_second=60)
print( int( (time.monotonic()-t)*1000) ) # ~210 msec/ frame on FunHouse
@devout jolt the file read time may depend on the order in which the pixels are being read from it
devout jolt
<@352910176736772096> the file read time may depend on the order in which the pi...
yeah exactly
slender iron
ya, ondiskbitmap ends up seeking through the file and loading a sector for every pixel potentially
devout jolt
so lesson learned: if you want fast sprites and you can load them in RAM, load them in RAM. thanks sorry for bugging everyone about this
slender iron
np
lone axle
If you do still want to save the RAM and use OnDiskBitmap maybe you could rotate the source image inside the file so that it's "sideways" when viewed normally, but correct when viewed on the display with the fastest rotation setting.
idle owl
so lesson learned: if you want fast sprites and you can load them in RAM, load t...
Sometimes learning involves bugging folks. 🙂
devout jolt
If you do still want to save the RAM and use OnDiskBitmap maybe you could rotate...
totes. but it makes it harder to supply one spritesheet that is cross-platform, since everyone's laying out their displays at all the angles 🙂
lone axle
yep, true that.
devout jolt
Look at them RAM sprites fly!
slender iron
welcome page:
ornate breach
That's a lot of sprites.
Speaking of sprites, I had an idea for blinka! Blinka slithering after “AdaBugs” to represent squashing bugs in circuitpython 🙂
Seemed like a cute way to visualize Circuitpython development
stuck elbow
welcome page:
is that a <dl> in there?
stuck elbow
well, in html you usually display lists with <ul> or <ol>, and dicts with <dl>
though the default style for it is really ugly
tulip sleet
@slender iron did you have some trouble with CONFIG_ETH_ENABLED when doing the mdns stuff? It's turning on mysteriously in the ESP32 build, but I have not set it anywhere. It's causing an #include file error.
btw, running the sdkconfig script generates some significantly different files than what is checked in
manic glacierBOT
manic glacierBOT
proven garnet
Got pyproject.toml working on the BME680 library!
lone sandalBOT
manic glacierBOT
manic glacierBOT
tulip sleet
@onyx hinge I think you will see this in your email, but in case not: https://github.com/micropython/micropython/pull/8813 re mp_obj_type_t size
onyx hinge
excellent news! should bring us some additional size savings when we can merge it in
tall delta
anyone have any idea how i can add https://github.com/raspberrypi/pico-examples/tree/master/flash/nuke to the raspberry pi pico build?
so that it'll first nuke the flash memory before flashing the real firmware
proven garnet
Got `pyproject.toml` working on the BME680 library!
@idle owl wanted to ping you on this but figured the AM was better 🥱
blissful pollen
so that it'll first nuke the flash memory before flashing the real firmware
Generally the CP builds are loaded by the UF2 boot loader so I would think it needs to go there (assuming it wouldn’t remove the boot loader too). That is found in the tinyuf2 repo.
tulip sleet
anyone have any idea how i can add `https://github.com/raspberrypi/pico-examples...
What's your use case for wanting this? Do you want to clear CIRCUITPY?
tall delta
What's your use case for wanting this? Do you want to clear CIRCUITPY?
before flashing a file, you usually flash nuke.uf2 to remove the old firmware
tall delta
Generally the CP builds are loaded by the UF2 boot loader so I would think it ne...
any idea how?
tulip sleet
you don't need to do that
flashing new firmware does an erase first anyway, of the blocks it is going to write
for CircuitPython, you generally don't want to do this, because it will erase what you have in CIRCUITPY
tall delta
flashing new firmware does an erase first anyway, of the blocks it is going to w...
yeah but if the firmware is smaller than the older firmware the other blocks won't be deleted
tulip sleet
you can flash on top of the old one
it doens't matter. Those blocks won't be used or read or anything
tall delta
for CircuitPython, you generally don't want to do this, because it will erase wh...
no, this is a project that generates the uf2 file for you with the code already in it so clearing CIRCUITPY is also a need
tulip sleet
if you are flashing CIRCUITPY at the same time as the CircuitPython firmware, then again you won't have to erase anything, because you'll be writing blocks that are part of CIRCUITPY
tall delta
i want to clear CIRCUITPY awell
tulip sleet
when you say "with the code already in it", do you mean code in CIRCUITPY, or frozen?
tall delta
when you say "with the code already in it", do you mean code in CIRCUITPY, or fr...
code in frozen
tulip sleet
frozen is part of the firmware
tall delta
yeah, so nuke erases the firmware
tulip sleet
you don't need to erase the firmware; it gets replaced when you load the new UF2
tall delta
oh
tulip sleet
there's no reason to use nuke unless there's some bad code.py in CIRCUITPY that is causing an immediate crash or something like that
and even then you can get into safe mode with a slow double click
every flash block that is part of the UF2 is erased first by the bootloader, and then written
all the frozen code is part of the CircuitPython UF2 - it is just more data that is part of the firmware. The frozen code is not in CIRCUITPY. I hope this makes your life easier.
tall delta
alright, thanks!
what if i needed to erase the code.py also? cause that is also being auto generated..
tulip sleet
how is it being auto generated?
tall delta
how is it being auto generated?
it converts from duckyscript to python.. this project is to turn the pico into a hid device..
tall delta
by modifying the buffer in filesystem.c
tall delta
on every hard restart?
before the firmware is compiled, the filesystem.c file is modified.
tulip sleet
do you have a pointer to the code? in your forked repo?
tall delta
do you have a pointer to the code? in your forked repo?
not really, all the script does is replace the text
tulip sleet
so you are replacing the "hello, world"
tall delta
https://github.com/compromyse/raspi-ducky/blob/main/circuitpython/supervisor/shared/filesystem.c#L77
it replaces this line
tall delta
so you are replacing the "hello, world"
yeah but in this case I have a different default
rather than print(hello, world), this is my default
import picoducky
main()
are you turning off visibility of CIRCUITPY?
tulip sleet
tall delta
yes, see <https://learn.adafruit.com/customizing-usb-devices-in-circuitpython#ca...
ah, thanks!
tulip sleet
a UF2 is just a bunch of blocks: it could include a CIRCUITPY
tall delta
where is boot.py located, from this article?
tall delta
is it possible to use the nuke?
tall delta
boot.py is in CIRCUITPY, just like code.py
alright!
tulip sleet
there is still no reason to use the nuke
you could check always rewrite code.py on restart, or you could check to see whether it's what you want, and rewrite only if it's wrong. See main.c, which checks to see if boot_out.txt needs to be rewritten or not
tulip sleet
it's not there by default, you create it
tall delta
you could check always rewrite code.py on restart, or you could check to see whe...
alright!
tulip sleet
it's just another file
tall delta
it's not there by default, you create it
ah, alright! thanks!
jaunty juniper
no, this is a project that generates the uf2 file for you with the code already ...
hey, you are using an RP2040 board right ?
tall delta
hey, you are using an RP2040 board right ?
yep
jaunty juniper
you can make a board with a normal version of Circuitpython and put what you want in CIRCUITPY and then create a UF2 file from that, that contains the entire flash, that you should then be able to flash as a whole on any identical board
you can do that with picotool:
picotool save --all my_full_board.uf2
(the board must be in RPI-RP2 mode)
tulip sleet
@jaunty juniper thanks for pointing out the ease of doing this. I had forgotten that
jaunty juniper
it's really nice for deploying on multiple boards
slender iron
@tulip sleet I don't remember any ETH issues. I do have to do some manual merging of the sdkconfig files
slender iron
well, in html you usually display lists with <ul> or <ol>, and dicts with <dl>
Ah, it's a <ul>. The file is here: https://github.com/tannewt/circuitpython/blob/cp_webserver/supervisor/shared/web_workflow/static/welcome.html I'd love a small html file with a small css to make it look good (especially on mobile)
lone sandalBOT
tulip sleet
@slender iron could i video with you today about the changes I'm seeing made by the tools/ scripts, and the different default values?
tulip sleet
thanks!
manic glacierBOT
idle owl
Got `pyproject.toml` working on the BME680 library!
Oh nice!
idle owl
<@330227457296957440> wanted to ping you on this but figured the AM was better �...
You can ping me anytime. I have notifications off when I don't want to get them, and they'll be here whenever I return. 🙂
@proven garnet When we find some overlapping time, I would like to chat about the process and what's involved.
manic glacierBOT
hidden rain
Using the Pico(rp2040) with CPy SPI to generate a sequence of alternating bits at 40KHZ. The configure interface allows one to specify bits. Can I assume that I might be able to set this to bits=16 with value bytes([0xAA,0xAA]). ???
stuck elbow
@hidden rain why aren't you using pwm?
the spi peripheral is not going to give you a consistent frequency
hidden rain
For my application, I need near Square Waves. The application was previously implemented on a 16bit TI-MSP430. So, I only need 16bits of alternating squares waves at 40KHZ. If there is a better way, I am open...
ornate breach
Isn’t pwm square waves?
stuck elbow
if you need exactly 16 bits at exactly 40kHz, then I suppose pulseio is your best bet
on the rp2040 you can also use pio directly
hidden rain
I will look into PWM. Thanks. However, since you mentioned it, what is the freq accuracy of SPI? is it that its not consistent each time its used?
stuck elbow
can you tell us more about what you are trying to do?
manic glacierBOT
hidden rain
As i've indicated, I am porting an application which previously used SPI to generate a square wave at 40KHZ
stuck elbow
since SPI is a clock-based protocol, it's the master that controls the clock, and there is no need for a consistent frequency -- so there may be gaps between sending bytes
but why 16 bits specifically?
hidden rain
Its a previously determined pulse width to ring a transducer. Our previous implementation is very accurate.
stuck elbow
if you need to send modulated pulses, then pulseio.PulseOut is for you
hidden rain
It sounds like I should in to that ...seriously. But I need some closure on SPI. While there might not be a need for consistency on the SPI clock, wouldn't it be easier for HW implementers create a consistent clock rather than a wondering clock? Just trying to complete my understanding here...
tame creek
it probably depends on a lot of things. if the SPI module is in some interrupt-driven mode, if you don't service the interrupt within the right window, it might pause the sending and the clock until you have serviced it. depending on what else your firmware is doing, you might not always get the timing right. SPI is intended for sending data, not waveform generation; if you want to use it for the latter, it's going to take a lot of work to get it consistently right, like maybe writing stuff in assembly language level of work
stuck elbow
the spi send in circuitpython is blocking, it means that you call it, it sends the bytes you have given it, blocking until they are all sent, and then goes back to executing python code
so you will have inevitable gaps for when the python code is being executed
the only guarantee SPI gives you is that it will send the data you give it at a clock speed not higher than configured
hidden rain
I get it. So only if the app is in a non-interruptible mode, blocks for 0.4 milliseconds (our app) or so, then I might expect "good" behavior. Ok, all good reasons to use PulseIO facility. Hopefully there are no "gotchas" there equally as bad. Just so I am clear in the SPI interface, the bits argument in configure CAN specify 16 bits, right?
stuck elbow
I get it. So only if the app is in a non-interruptible mode, blocks for 0.4 mill...
I think it depends on the port, not sure if it works on rp2040
hidden rain
The RP2040 Datasheet indicates its capable, but CPy might not acknowledge it. (??) Not enough documentation there...!
stuck elbow
the easiest way is to try and see
hidden rain
Sure enough! Looking at Pulseio. Hoping I can control if the first part of the signal is high rather than low...
stuck elbow
and if pulseio doesn't fit your needs, the pio lets you make a pin output pretty much anything you want, but it's a bit of a learning curve for the pio assembly
I'm afraid it always starts with on:
Pulse alternating on and off durations in microseconds starting with on.
but it's also blocking, doesn't work in background
for background operation it's pwm or pio
hidden rain
I've programmed 6502 and 68000 and old IBM 360 assembly. PIO only has 6 (or 8) commands? Blocking is OK for our app. We had some issues on the older SPI anyway, like Beginning and trailing power spikes, etc....However, its seems I'll need a scope....
manic glacierBOT
hidden rain
Thanks for your help!
onyx hinge
we have a guide on pio https://learn.adafruit.com/intro-to-rp2040-pio-with-circuitpython in addition to the good docs from raspberry pi themselves
stuck elbow
I've programmed 6502 and 68000 and old IBM 360 assembly. PIO only has 6 (or 8) c...
you can use an rp2040 as a logic analyzer, if that helps...
hidden rain
So use one as a Generator, and one as an Analyzer (using ADC?)
stuck elbow
well, it's a digital signal, so not even adc
onyx hinge
you'd want to have a loop where you alternately make a pin high or low (possibly with the "side set"), and use the decrement-and-jmp instruction to do the high and low pulse the desired number of times. By using delays ([1]) or NOPs you can ensure the high and low 'parts' are the same in PIO cycles...
jaunty juniper
https://hackaday.com/2022/03/02/need-a-logic-analyzer-use-your-pico/
does that work ? I remember a project like that but it was in its infancy
hidden rain
OK.. so that algorithm would generate the pulses... so then I would monitor a GPIO pin and slam the results into a buffer at some read speed..?
oh COOL!
onyx hinge
you'll need a PIO clock related to your 40kHz output frequency (at least 2x but maybe 4x), but unfortunately the 125MHz base clock doesn't evenly divide down to 80kHz or 160kHz. If you request a frequency that's not an exact division, then there'll be some variation (+-8ns, one clock at 125MHz) of PIO cycle lengths
hidden rain
I can live with that...
onyx hinge
pio can read bits in too, there are examples for UART and SPI that show how to get bits in to the microcontroller (in the pico docs if not on learn)
stuck elbow
does that work ? I remember a project like that but it was in its infancy
it's hackaday, so who knows, but Arsenijs usually knows what he's talking about
hidden rain
quite exciting...now I just need another 48 hours ... so many projects ... not enough time... can't wait to share with my local micro-controller group
Is there a way to capture the last 30 minutes to a file?
stuck elbow
copy-paste, I suppose
hidden rain
Done. Thanks Again...I have a lot of reading to do...(later!)🤓
blissful pollen
does that work ? I remember a project like that but it was in its infancy
I wrote up and have a repo on how to use it. a few others tried to make it more robust but my hackster.io write up covered the basics.
Im stuck away from my PC for a bit so can’t find the link.
jaunty juniper
I know of yours, but that hackaday post seems to be something else
idle owl
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1ZQVrnwy5uTp5YkvsJUs9dr-UjYWejaTNT2OzaN27UhQ/edit?usp=sharing
manic glacierBOT
tall delta
you can make a board with a normal version of Circuitpython and put what you wan...
but then the picotool is a dependency to the program.. I was hoping to have minimal dependencies..
jaunty juniper
a dependency to your creation of the initial image, once you have the UF2 you're good
it's not more a dependency than the environment to build Circuitpython
I guess it depends what you want to do, I don't know your whole context
warm stump
some unlikely progress on a custom node-red python node called node-red-contrib-empython
which is the micropython javascript emscripten port built and heavily modified
but it's too messy to be able to upload in any meaningful way in its current state and requires httpStatic in settings.js to host firmware.wasm
all said and done, node-red-node-circuitpython? update: I have just made it into a dev branch: https://github.com/TheMindVirus/helpnodered/tree/empython
spiral elk
@idle wharf I'll update my build environment and spin up a build tomorrow for the Lolin board to try on real hardware.
Unless you have a binary at hand.
mental nexus
I made some changes to the ESP-IDF for my circuitpython fork. How can I upload to github my version of the ports/espressif/esp-idf into my circuitpython github fork so I can save it for posterity?
stuck elbow
well, you fork idf as normal, then change the url in the submodules
mental nexus
Ok, I now see the .gitmodules file that you are referencing.
I made my changes to the esp-idf in my local branch of circuitpython. If I fork the esp-idf, will I have to go find my changes and then somehow merge them into my esp-idf fork. Is there any easy way?
stuck elbow
you can push your changes to your new repo, since they will be related
or you can cherry-pick from your local repo
mental nexus
Ok, I think I've got it. Thanks @stuck elbow !
manic glacierBOT
glossy pecan
Not a CP issue, but closely related. Can’t install or run Mu Editor on Apple Silicon (M1X) MacBook Pro. Finder won’t copy app bundle (image) to Application nor(!) to data folder. Attempting to run app from distribution virtual disk (.dmg file) starts up, but quickly crashes (trying to access a file?) Changing System Prefs to give Mu file access didn’t help. Recommendations? Better channel? 🔥
jaunty juniper
Not a CP issue, but closely related. Can’t install or run Mu Editor on Apple Sil...
what do you mean by the Finder won't copy ? does it show an error message ? silently do nothing ? what happens ? Is it the same as here ? https://github.com/mu-editor/mu/issues/2274
glossy pecan
@jaunty juniper Finder does nothing. Yes, the mu issue 2274 is related -- error report from my colleague, directly on the mu github page.
jaunty juniper
that issue mentions an error message
glossy pecan
Mu issue 2274 is correct, did get an error message in that specific case. The report was from my colleague. We're encountering a number of file io/security/etc. issues that are specific to the following configuration: macOS Monterey, current release (not dev beta), and Apple Silicon. We're also seeing. bunch of serial-over-USB issues specific to this MacBook Pro, but haven't eliminated the USB-C to USB Micro-B cable as a cause.
jaunty juniper
that kind of looks like a Finder bug, I would try copying to another drive first if you can, or maybe using the command line to copy it ?
or a bug in the APFS implementation on M1 ? 🤷
glossy pecan
To avoid APFS security issues, I've asked Mark to try a thumb drive. Will try to do a specific device <-> cable <-> MBP testing matrix within the next day. One issue I'm not an expert on is: What happens in a cable/adapter between Thunderbolt 4/Type-C socket and the Micro-B plug. Just wiring, or is some logic HW involved?
To be specific, one definite serial IO issue is: We have a "simple" [Thunderbolt4/Type-C-socket] <==========>[Micro-B plug] cable. Using it between the MBP and a RPi Pico doesn't work. Pico is not even seen in the Mac System Report | USB list. If, however a different cable setup is used, with a USB hub in the middle, the Pico is seen and serial IO works. ==> i distrust the first cable.
timber mango
Hello everybody. I'm trying to play wave sample at different pitch, but i don't kwon how to do that. Does anyone kwnows how to make different pitch with audiomixer or other library. Thanks !
I use à Pi Pico code with CircuitPython and I plug a neotrellis to the pico pico . My goal is to play the same sample at different pitch when i press a pad on the neotrellis. My first idea was to make different samples at different pitch with audacity but I got lot of bug when I play and memory is very short on the pico.
onyx hinge
@glossy pecan looks like there are several issues related to this filed with mu-editor, you may find it useful to read for suggestions or if you have additional info to add. https://github.com/mu-editor/mu/issues/1847 has the most info [but they believed it was resolved as of December] and https://github.com/mu-editor/mu/issues/2274 which is still open but doesn't have much activity as yet.
I see you already found 2274
tulip sleet
Hello everybody. I'm trying to play wave sample at different pitch, but i don't ...
Are these just sine-wave tones or are they complex waveforms? See https://docs.circuitpython.org/en/latest/shared-bindings/audiocore/index.html#audiocore.RawSample for some ideas
onyx hinge
the sample_rate property of a wavefile is settable, it looks like
manic glacierBOT
@kstillson Just had a little time to build for my RP2040 and it looks like something is being inverted on the reading address. Reading 0x3D 0b0011 1101 instead of 0x42 0b0100 0010 attached below are the scope plot and screen shot.
Client_Transmission.txt
[Client_Transmission.csv](ht...
proven garnet
<@135415127879974912> When we find some overlapping time, I would like to chat a...
Sounds good!
native sandal
I'm interested in adding TLS support to the adafruit/Adafruit_CircuitPython_Wiznet5k driver. I’m an expert software developer with decades of experience on a multitude of platforms, but my Python knowledge is fairly basic (I’ve written a few scripts to help with iOS development, and dabbled in CircuitPython). So I decided it would be good to take on this "good first issue", in order to make sure I can update an existing CP library. I'm on an M1 MacBook Pro, I've forked the repo and cloned it, and I installed mypy to see if I could get it to show me the type annotation warnings mentioned in the bug. I can, so all's good there.
Now I’m merrily annotating the file, learning about Python type annotation as I go. My concern is this. The bug says to import from typing conditionally in case the module doesn’t exist. In that case, what happens in the code with all the undefined imports? Are they just None and ignored by CP? @lone axle @idle owl
stuck elbow
@native sandal the type annotations are not getting executed when running the code normally, without mypy, they are only annotations
native sandal
<@435905586203459624> the type annotations are not getting executed when running...
Thanks for confirming that.
I'm running into a new issue. In some instances, if I annotate a method in a class, then references to an import within that method generate an error:
@property
def connected(self) -> bool:
"""Returns whether or not we are connected to the socket."""
if self.socknum >= _the_interface.max_sockets:
return False
status = _the_interface.socket_status(self.socknum)[0]
if (
status == wiznet5k.adafruit_wiznet5k.SNSR_SOCK_CLOSE_WAIT
and self.available() == 0
):
result = False
else:
result = status not in (
wiznet5k.adafruit_wiznet5k.SNSR_SOCK_CLOSED,
wiznet5k.adafruit_wiznet5k.SNSR_SOCK_LISTEN,
wiznet5k.adafruit_wiznet5k.SNSR_SOCK_TIME_WAIT,
wiznet5k.adafruit_wiznet5k.SNSR_SOCK_FIN_WAIT,
)
if not result and status != wiznet5k.adafruit_wiznet5k.SNSR_SOCK_LISTEN:
self.close()
return result
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:151: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:157: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:158: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:159: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:160: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:162: error: Module has no attribute "adafruit_wiznet5k"
If I remove the -> bool annotation, it checks fine.
(but I get an error that the method isn't annotated)
stuck elbow
do you have a link to the original file on github?
stuck elbow
thanks
I would say this is a legitimate error, there is no adafruit_wiznet5k inside https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/blob/main/adafruit_wiznet5k/adafruit_wiznet5k.py
native sandal
I agree, but that's how the code is written, and it runs, afaik
stuck elbow
it might be some quirk of micropython/circuitpython
it should probably be wiznet5k.SNSR_SOCK_FIN_WAIT instead of wiznet5k.adafruit_wiznet5k.SNSR_SOCK_FIN_WAIT
native sandal
I thought it was doing something I just don't understand. I haven't been able to figure out how to get at the WIZNET5K defined inside to use as a type annotation in this file (for _the_interface)
I tried that, and it couldn't find that either.
native sandal
This file is also inside that package.
well, assuming I know what a "module" and "package" are
stuck elbow
what path are you running the mypy from?
native sandal
$ mypy adafruit_wiznet5k/adafruit_wiznet5k_socket.py
native sandal
sorry, thought it was implicit in that. the top level of this repo
stuck elbow
yeah, sorry, a bit dense today
native sandal
Haha no worries
stuck elbow
it should work like that, sorry, I can't see where the problem lies
the problem itself is that it is importing the module instead of the package
but why, I don't know
native sandal
The module is the .py file inside the directory adafruit_wiznet5k, right?
stuck elbow
right
native sandal
That module definitely defines all the things it's trying to use
stuck elbow
but at one level lower
native sandal
I just don't know if import adafruit_wiznet5k as wiznet5k is importing that module or the package
native sandal
like, start the Python repl, import like that, and execute the print?
stuck elbow
or just add it to that code, right after the import, and run mypy on it
native sandal
Nothing prints, just the errors:
23522 02:28: Adafruit_CircuitPython_Wiznet5k/ (rmann/issue-45-missing-type-annotations) $ /Users/rmann/Library/Python/3.8/bin/mypy adafruit_wiznet5k/adafruit_wiznet5k_socket.py --disallow-untyped-defs
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:28: error: Function is missing a type annotation for one or more arguments
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:75: error: "object" has no attribute "get_host_by_name"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:114: error: "object" has no attribute "get_socket"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:121: error: Function is missing a type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:142: error: "object" has no attribute "socket_status"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:147: error: "object" has no attribute "max_sockets"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:149: error: "object" has no attribute "socket_status"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:151: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:157: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:158: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:159: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:160: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:162: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:168: error: "object" has no attribute "remote_ip"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:176: error: Incompatible types in assignment (expression has type "List[int]", variable has type "bytes")
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:180: error: Function is missing a type annotation for one or more arguments
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:187: error: "object" has no attribute "unpretty_ip"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:188: error: "object" has no attribute "ifconfig"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:190: error: "object" has no attribute "ifconfig"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:195: error: "object" has no attribute "socket_listen"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:196: error: Module has no attribute "adafruit_wiznet5k"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:200: error: Function is missing a type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:209: error: Function is missing a return type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:238: error: Function is missing a type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:263: error: Function is missing a type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:272: error: Function is missing a type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:282: error: Function is missing a type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:341: error: Function is missing a type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:356: error: Function is missing a type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:371: error: Function is missing a type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:387: error: Function is missing a return type annotation
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:413: error: "object" has no attribute "socket_disconnect"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:417: error: "object" has no attribute "socket_close"
adafruit_wiznet5k/adafruit_wiznet5k_socket.py:421: error: "object" has no attribute "socket_available"
Found 34 errors in 1 file (checked 1 source file)
Perhaps because I'm in the middle of annotating and stuff is breaking? Not sure.
stuck elbow
no, I assumed mypy imports the file, but it doesn't, it just parses it, and doesn't execute it, which is correct way to do it
stuck elbow
it really feels like the way mypy is resolving imports is somehow different to how circuitpython does it
maybe someone else will have better insights once they wake up
native sandal
Hmm. @lone axle wrote the bug, perhaps he can shed light on it.
Thanks for trying!
native sandal
@lone axle My effort so far to address Issue 45 (note I still haven't figured out the import issue, nor how to properly reference the WIZNET5K class type): https://github.com/JetForMe/Adafruit_CircuitPython_Wiznet5k/blob/rmann/issue-45-missing-type-annotations/adafruit_wiznet5k/adafruit_wiznet5k_socket.py
violet thistle
Anyone knows if there is a plan to implement BLE on the ESP32 S3?
lone axle
I'm interested in adding TLS support to the `adafruit/Adafruit_CircuitPython_Wiz...
I haven't caught up on the full conversation yet, but yes all of the type annotations get ignored by the interpreter on the microcontroller. So it failing the import of typing does not cause any issues since it doesn't ever try to do anything with the types.
lone axle
If I remove the `-> bool` annotation, it checks fine.
what is running the check that is outputting these error messages? Is that pylint or mypy or something different?
violet thistle
https://github.com/adafruit/circuitpython/issues/5926
Thanks, but the last update is months ago, and I don't understand the feasibility or the issues described there.
stuck elbow
Thanks, but the last update is months ago, and I don't understand the feasibilit...
well, you were asking about plans, so here you are, it's in the "long term" milestone
lone axle
Feel free to ping me when you get back to it. I should around most of today. I don't have much mypy experience beyond when I used it to create those issues on the library repos. I'm not sure about any of the specificities of how it resolves imports.
The script that created the issues also checked for only specific types of output from mypy as well, so it's possible that you are seeing warnings and/or errors that we also would have seen when we created it, but wasn't one of the specific lines we were interested in (missing annotations for a function or return one primarily)
violet thistle
Thanks. OK, so I guess not planned anytime soon.
lone axle
Is planned, but not on a specific timeline is how I interpret the messages in the issue.
lone axle
<@382939733107408897> My effort so far to address Issue 45 (note I still haven't...
classes can be referenced as Strings I think as well. So "WIZNET5K" may work as a type for something that is of that class type.
tulip sleet
@slender iron I have a q re espressif never-reset logic when you are free
digital shoreBOT
slender iron
@violet thistle The issue is that the NimBLE stack doesn't have an API for adding services after the stack is started
we're hoping to find someone with more nimble experience to help us but our own priorities have been on web workflow
onyx hinge
<@&356864093652516868> It's about 15 minutes until our weekly meeting! Remember to add your notes to the document. See you soon!
Has anyone tried to get Github CoPilot to generate CircuitPython code?
tulip sleet
Has anyone tried to get Github CoPilot to generate CircuitPython code?
i wanted to, but I have to switch editors 😦
onyx hinge
yeah I'd be grumpy about that too.
thorny jay
Sorry, skipping the meeting with no notes.
tulip sleet
np, it's summer!
stuck elbow
on one of the hemispheres, anyways
onyx hinge
there's more than one hemisphere? 🤯
stuck elbow
there are infinitely many
onyx hinge
It's happened to us all, Dan
errant grail
<Lord of the Rings>
onyx hinge
typical screenshot of mines of moria
tulip sleet
after my time
errant grail
typical screenshot of mines of moria
Ah a map! Thanks!
stuck elbow
the orignal rogue is actually still pretty good, I play it from time to time
onyx hinge
"roguelike" games were a big part of how I flunked out of university repeatedly from 1994-1998 🤣
slender iron
@onyx hinge you could pick up usb host if esp32 isn't ready
onyx hinge
@slender iron I feel intimidated by the very idea, thanks for your confidence
slender iron
thach has done most of the work and the API is laid out already
idle owl
setup.py has been legacy since before we started using it.
onyx hinge
python packaging is a land of many contrasts
mental nexus
Thanks all!
lone axle
Yep, Will send a link shortly.
slender iron
thanks all!
idle owl
Thank you, everyone!
errant grail
Thanks!
onyx hinge
👋
lone axle
@tulip sleet https://www.dropbox.com/s/trls34ui2r5r3j2/circuitpython_weekly_meeting_06_27_22.mkv?dl=0 download button top left corner should give you a copy of it.
tulip sleet
<@329766224093249548> https://www.dropbox.com/s/trls34ui2r5r3j2/circuitpython_we...
downloaded! TYVM!!
lone axle
The one bit that is not in the recording is actually me saying that I have the backup recording running. I keep my local mic muted in OBS when I do the backup recordings outside of my turns to speak and didn't think to unmute when I responded.
tulip sleet
(thanks to 300Mb up/down)
lone axle
❤️ fiber for upload
tulip sleet
yes!
manic glacierBOT
tulip sleet
Here is the notes document for next week's CircuitPython Weekly meeting. It is on Tuesday, not Monday (due to a US holiday) 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/1__eWF8gXJ7qx_BpZiPlvGGkT_jtalZ72YbSqG_5ZEM0/edit?usp=sharing
idle owl
@lone axle I have a couple of questions about the status light code.
What exactly is the POLL_DELAY and why is it set to 3 minutes? Is that how often it's querying the API?
lone axle
What exactly is the `POLL_DELAY` and why is it set to 3 minutes? Is that how of...
Yep, exactly right, frequency that we pull the API. and I tried to choose something not too long, but not too short. There is lots of room to move either way and different values could make sense depending on the project and frequency / duration of their actions runs.
idle owl
OK, got it. So leave it as is, but document it.
Second question is, what scope do I need to put on this API token?
lone axle
let me see if I can tell what I put on mine. It should be fairly minimal I think since it's only reading and the data is from the public repo.
idle owl
Right. Agreed, but I'm rarely clear on what that means in terms of check boxes. Heh.
lone axle
I will find out the minimal requirements. I found mine, but almost certain it doesn't need everything (I re-used one from my IDE)
lone axle
For circuitpython core repo you can make the token with no scope boxes checked.
idle owl
Oh noticed you didn't add the buzzer to the code except in setup. Once I get it running, I'll see if I can add it successfully. I'll timebox myself and seek help from you if I reach the limit.
idle owl
For circuitpython core repo you can make the token with no scope boxes checked.
Nice! Thank you.
lone axle
Sounds good. happy to help if so.
lone axle
Yep shows green on my end as well atm with prints only, no light. So now it'll just run on a loop and update if/when it sees a new workflow run that isn't the same one that it already showed.
idle owl
OK, trying to add buzzer before leaving it running. 🙂
@lone axle Ok, I get the idea of how to do it, and where it should go, but I'm going to end up iterating all over the place to figure it out. So I'm tagging you in. (I'm not on the same line numbers as you, I did a bit of linting.) It obviously goes within the ENABLE_USB_LIGHT_MESSAGES at the end following if conclusion ==. But with the existing time.sleep() I'm not sure how to make it happen without buzzing for 30 seconds which is a big nope. I believe the code is send_command(mSerial, BUZZER_ON) and BUZZER_OFF.
lone axle
I would make a CONCLUSION_BUZZER_TIME I think and then use that subtracted with the conclusion light time to play it. I'll make a new commit in a moment that shows what I am thinking
idle owl
Ok, thank you!
lone axle
@idle owl do you want buzzer for failed workflow only? or buzzer on successful completion also?
idle owl
<@330227457296957440> do you want buzzer for failed workflow only? or buzzer on ...
Any completion. So, failed and successful. Not when yellow.
proven garnet
@idle owl I'm good to meet this week with a list of what's needed for the pyproject.toml switch. I think we have the tools needed to do it. Any time that work well for you?
idle owl
<@330227457296957440> I'm good to meet this week with a list of what's needed fo...
Anytime works for you? Or does it need to be after work?
lone axle
@idle owl this commit adds a new buzzer time variable and should turn it own for that many seconds when the light comes on. https://github.com/FoamyGuy/Github_Actions_Status_Light/commit/8ae628a66ba6a987e0cb7518f5a53664d331f576
It can be set to 0 to disable the buzzer. If it does player the buzzer sound it will subtract that time from the light sleep time so that it still turns back off after the configured delay.
proven garnet
Lunchtime or after 6pm works for me. Both are generally subpar I assume so you can choose 😅
idle owl
WOW that buzzer is loud. 2 seconds is super long.
idle owl
Lunchtime or after 6pm works for me. Both are generally subpar I assume so you ...
6pm tonight work for you?
The name "buzzer" is a vast underestimation of what you're actually dealing with.
@lone axle It appears to go green again even though there isn't a new workflow.
Opening serial port.
Starting Github Actions Status Watcher
Press Ctrl-C to Exit
Fetching workflow run status.
completed - success
GREEN ON
GREEN OFF
Fetching workflow run status.
completed - success
GREEN ON
GREEN OFF```It shouldn't do that right? It's supposed to know it already showed the status of that run?
lone axle
It shouldn't do that right? It's supposed to know it already showed the status o...
correct, it should not show the same one twice. It will start printing the already showed message like this:
idle owl
Mine is not doing that :/
lone axle
However it does store the already_shown ones in memory so if you stop and then start again it will show the most recent one again.
lone axle
Did it print something like: "adding 2570019400 to shown IDs"
idle owl
I stopped it after three times.
This is everything Fetching workflow run status. completed - success GREEN ON GREEN OFF Fetching workflow run status. completed - success GREEN ON GREEN OFF Fetching workflow run status. completed - success GREEN ON GREEN OFF
No other prints happened.
Did I mess something up maybe?
lone axle
can you throw your current code in gist or somewhere and I can take a look?
idle owl
I think I messed something up.
Let me compare it to the original code before I send it to you
I think something got removed.
Letting it loop to test. I accidentally deleted somewhere in my copying and pasting the relevant code.
I think.
Ok, so far so good: adding x to shown IDs
3 minutes is a long time when you're waiting to test. I should have shortened it for testing purposes, heh.
Yes!
@lone axle Turns out I had removed 2 crucial lines of code.
Thank you!
lone axle
Yep you're welcome. Let me know if anything else arises.
idle owl
I will! This code is great. And you didn't even have hardware to test it with! Worked on the first try.
Thanks again and I'll keep you posted. Guide for this project is up next. I'll put your name on the code 🙂
lone axle
Nice! All these years programming and it's still definitely a thrill to get things first try. 🎉
I may adapt it into a matrix portal version. It's a neat idea to have it visible some where at a glance.
idle owl
Nice!
idle owl
@onyx hinge I'm getting this when trying to update my local copy of CP, using make fetch-submodules. ```
From https://github.com/raspberrypi/pico-sdk
- branch 426e46126b5a1efaea4544cdb71ab81b61983034 -> FETCH_HEAD
! [rejected] 1.3.0 -> 1.3.0 (would clobber existing tag)
fatal: run_command returned non-zero status for ports/raspberrypi/sdk
.
make: *** [fetch-submodules] Error 128```
Haven't seen that before.
onyx hinge
<@426200547024961539> I'm getting this when trying to update my local copy of CP...
yeah I know it's been discussed before but I didn't see a resolution when I searched
idle owl
Hmm.