I've been measuring a blinky and running the micropython perf bench
#circuitpython-dev
slender iron
blissful pollen
Cool I had never looked at the perf bench before. Thanks
slender iron
I hadn't much either
jaunty juniper
someone understands what's happening here ? It looks like the library is trying to override property accessors in a subclass, and it doesn't work that way:
https://forums.adafruit.com/viewtopic.php?t=199790
this library:
https://github.com/adafruit/Adafruit_CircuitPython_EMC2101
test code:
class Thing:
def __init__(self, i2c):
self.i2c_device = i2c
@property
def i2c(self):
print(self)
return self.i2c_device
class Thing_sub(Thing):
def __init__(self, i2c):
super().__init__(i2c)
@property
def i2c(self):
return super().i2c
c = Thing_sub("I2C")
print(c.i2c)
code.py output:
<super: <class 'Thing_sub'>, <Thing_sub object at 0x3fd8c9a0>>
Traceback (most recent call last):
File "code.py", line 17, in <module>
File "code.py", line 14, in i2c
File "code.py", line 7, in i2c
AttributeError: 'super' object has no attribute 'i2c_device'
onyx hinge
@jaunty juniper probably super objects just don't work that way in CP/MP. (in MP it erroneously results in a <property> object which also isn't right) You're welcome to file an issue about it but I doubt there's an easy and simple fix.
jaunty juniper
yeah but that's a bug in the library
onyx hinge
then the library will need to be changed to work around the core limitation I suppose
jaunty juniper
also the super().thing notation doesn't work with normal properties ???
class Thing:
def __init__(self, i2c):
self.i2c_device = i2c
class Thing_sub(Thing):
def __init__(self, i2c):
super().__init__(i2c)
def test(self):
print(self.i2c_device)
print(super().i2c_device)
c = Thing_sub("I2C")
c.test()
I2C
Traceback (most recent call last):
File "/Volumes/FUNHOUSE/code.py", line 13, in <module>
c.test()
File "/Volumes/FUNHOUSE/code.py", line 10, in test
print(super().i2c_device)
^^^^^^^^^^^^^^^^^^
AttributeError: 'super' object has no attribute 'i2c_device'
that's python 3.11
so... it's an accident that it works with @property in C python or what ?
anyway I'll open an issue in the library
onyx hinge
so the fix in EMC2101 is to just use self.i2c_device in the subclass, rather than super().i2c?
jaunty juniper
hey my internet is back !
there is stuff like that:
@property
def fan_speed(self):
"""The current speed in Revolutions per Minute (RPM).
:return: float speed in RPM.
"""
self._check_status()
return super().fan_speed
which is the same case as my test
also I'm not sure that I understand the actual error with EMC2101
the error is when i2c_struct tries to get obj.i2c_device, and somehow obj is super() ? That might be a wrongly reported error, caused by the complex weave that adafruit_register weaves
chilly jay
How would I go about including the FreeRTOS and task headers from ports/espressif/esp-idf/components/freertos/include/freertos/ correctly with #include so they work flawlessly with VSCode intelli-sense?
It seems there are multiple FreeRTOS.h and task.h files in the project structure so only specifying the file name results in a confused VSCode while using the entire path also seems to confuse VSCode (types like TaskHandle_t are simply not recognised at all while they should be) and basically makes it unable to compile.
I am currently developing a module for usage with circuit python for use on an esp32s2 / feather s2 that has to use these headers since it does some stuff with FreeRTOS tasks and such.
midnight ember
Scott, I have 2 iMXβs if you need someone to run simple beta tests of builds I can do that. I donβt have access to a debugger though. Can offer to be a simple helper if needed.
rugged spindle
hopefully this is the correct place - does someone know why canio is available on the QT Py ESP32-C3 but not on the Seed Xiao ESP32-C3?
jaunty juniper
hopefully this is the correct place - does someone know why `canio` is available...
it is though, where do you see it's not ?
rugged spindle
having looked at my clipboard - it would appear "fat finger syndrome" was involved
you are right - it is shown on the canio page
please ignore me (for this once, not every time) π
proven garnet
Not exactly CircuitPython related, but seems like the relevant audience - but I started converting one of my CPython libraries to a C extension, and boy is it fun (when it works). Is this what core work feels like? π
stuck elbow
I think core work is mostly tracking down obscure off by one errors
proven garnet
There are only two problems in CS - cache invalidation, naming things, and off by one errors.
brazen hatch
I am doing the cptoml module now, it's almost done.
Reading is 100% done, supporting ", ', comments, octal, hex, scientific notations, bool, int (with +/- support).
I have done about 70% of writting new keys.
It's amazing how many edge cases there are.
I don't want to implement dates.. It's such a paaaiiiin.
It's the only thing left to do for reading though.
It's almost 200 lines of code now huh. And I am optimising it as much as possible..
It will be an all-in-one for settings.toml though.
It also applies basic formatting.
slender iron
Scott, I have 2 iMXβs if you need someone to run simple beta tests of builds I c...
Thanks for the offer. This optimization stuff is on the imx evk. Definitely use the imx port when prototyping though. Iβm sure there are still general issues
slender iron
How would I go about including the FreeRTOS and task headers from `ports/espres...
I donβt think any of us core devs use vscode so we wonβt know how to get it setup well
jaunty juniper
hmmmm I have an ESP32-S3 boards with builtin I2C oled here, but there doesn't seem to be on-board pull ups π
also the code seems to halt when running board_init trying to setup the display, dunno if it's related, it seems to boot fine without it
tulip sleet
@jaunty juniper I am fixing up adafruit_hid in a few ways. I'm thinking of adding a longer timeout to the Keyboard() etc constructors, and making the timeout optionally specifiable in the constructor. I think you helped people with this: on some host computers (maybe especially after host sleep?), it may take a while to re-enumerate. Also I can use supervisor.usb_connected to check -- it didn't exist when this library was first written.
Do you have some advice you give people ("insert a time.sleep(2)" or whatever)?
jaunty juniper
<@199811382793863168> I am fixing up `adafruit_hid` in a few ways. I'm thinking ...
something like that to handle PC reboots and sleeps:
#help-with-circuitpython message
I should write that down somewhere
tulip sleet
maybe it should be an example in the library, and it should be in a guide as well
jaunty juniper
I would think a default timeout waiting for USB would be nice, and I was wondering if we wanted to make the library itself resilient to USB going away, like print() is to the absence of a CDC connection, but I don't know if ignoring the errors is enough
tulip sleet
I would think a default timeout waiting for USB would be nice, and I was wonderi...
i think there are use cases where ignoring errors is good, and others where the code wants to know
the trouble is that the host and CircuitPython might get out of sync (e.g. caps lock) if there's a temporary disconnection
or the code just wants to flag "hey, your keystrokes are not getting through"
the reboot/sleep issue isn't helped that much by a timeout, because it could be an indefinite period of time. But we have seen cases where it doesn't come back properly. I have a Trinket M0 HID volume control I have to reset sometimes after sleep
jaunty juniper
yeah if we ignore the errors we need to return a failure flag, which might not be better
tulip sleet
we should also document the exceptions that can be raised
jaunty juniper
@tulip sleet with a (ESP32S3) board that has a builtin oled on I2C but no pull ups, can we hack something to support it in board.c with internal pull ups ?
tulip sleet
<@329766224093249548> with a (ESP32S3) board that has a builtin oled on I2C but ...
Yes, you can do a build with CIRCUITPY_I2C_ALLOW_INTERNAL_PULL_UP = 1
jaunty juniper
Yes, you can do a build with `CIRCUITPY_I2C_ALLOW_INTERNAL_PULL_UP = 1`
nice, I'll try that
manic glacierBOT
fossil turtle
Does anyone know of a good link that describes how to bring up SD Card support for mkrzero running circuitpython?
still zephyr
Neradoc. let me know if I can contact you by DM. Thanks π
tulip sleet
@onyx hinge I added a commit to your PR to document that mkfs() is a staticmethod.
that was the original "bug" from the issue
tulip sleet
@onyx hinge I will try to finish #7739 off unless you want to.
onyx hinge
@tulip sleet that would be kind, thank you. I don't have a lot more working time today.
jaunty juniper
well setting CIRCUITPY_I2C_ALLOW_INTERNAL_PULL_UP = 1 didn't work, it still complains about "no pullups", I wonder if there is a setting on ESP32S3 to change the strength of the internal pull up. It came with an Arduino/u8g2 demo, maybe I could look at how Arduino does its thing
tulip sleet
well setting `CIRCUITPY_I2C_ALLOW_INTERNAL_PULL_UP = 1` didn't work, it still co...
you can turn off the check for pullups, and then ALLOW_INTERNAL_PULL)UP will set the pullups
i am trying to remember why allow_internal_pullup was added. let me look
jaunty juniper
I tried that too, so far I was not able to scan or find the display (scanning takes an eternity)
I vaguely remember somebody needing/wanting it for their NRF board ?
jaunty juniper
yeah I'll just give up on this board for now, too bad the other two have pullups and it works nicely with board.DISPLAY
tulip sleet
I think the "ALLOW" was for a different reason
You could try replaceing this:
#if CIRCUITPY_REQUIRE_I2C_PULLUPS
// Test that the pins are in a high state. (Hopefully indicating they are pulled up.)
gpio_set_direction(sda->number, GPIO_MODE_DEF_INPUT);
gpio_set_direction(scl->number, GPIO_MODE_DEF_INPUT);
gpio_pullup_dis(sda->number);
gpio_pullup_dis(scl->number);
gpio_pulldown_en(sda->number);
gpio_pulldown_en(scl->number);
common_hal_mcu_delay_us(10);
gpio_pulldown_dis(sda->number);
gpio_pulldown_dis(scl->number);
#if CIRCUITPY_I2C_ALLOW_INTERNAL_PULL_UP
gpio_pullup_en(sda->number);
gpio_pullup_en(scl->number);
#endif
// We must pull up within 3us to achieve 400khz.
common_hal_mcu_delay_us((1200000 + frequency - 1) / frequency);
if (gpio_get_level(sda->number) == 0 || gpio_get_level(scl->number) == 0) {
reset_pin_number(sda->number);
reset_pin_number(scl->number);
mp_raise_RuntimeError(translate("No pull up found on SDA or SCL; check your wiring"));
}
#endif
with just:
gpio_pullup_en(sda->number);
gpio_pullup_en(scl->number);
as a test
jaunty juniper
right, I'll try looking into common hal busio something see what happens if I force the pull ups there, but that's for another day, I'll finish the PRs for the ones that work
like, these don't seem to be actually used
.sda_pullup_en = GPIO_PULLUP_ENABLE,
.scl_pullup_en = GPIO_PULLUP_ENABLE,
we have avoided providing for internal pullups for I2C in general, because the internal pullups are usually out of spec for the required pullup current. For instance, on RP2040 the internal pullups are 60k-80k ohms, which is way out of spec. on nRF they are about 15k, which is still out of spec.
tulip sleet
looks like the ESP32ish pullups are like 30k-80k?
in the datsheet they are even called "WPU" (weak pull up)
brazen hatch
It's finally done
Deletes tables autogically when empty. Applies formatting and can fetch, put, delete.
It's basically an aio for toml.
I have optimised it as much as possible.
I went nuclear with it as I want it to be called very often.
I perhaps shouldn't be talking about it in this channel..
ornate breach
I perhaps shouldn't be talking about it in this channel..
Whatβs wrong with this channel?
Seems relevant?
brazen hatch
It's for coredev? I should prolly have used #show-and-tell
ornate breach
Ah, maybe π€·ββοΈ
brazen hatch
But I basically live here, so I uncontiously sent it here.
idle owl
No worries. This is a perfect place for it. π
ornate breach
Hai Kattni π
idle owl
Hello π
ornate breach
Glad youβre back π
idle owl
Thanks!
jaunty juniper
@slender iron can I do a request on pid.codes for using Circuitpython for a RP2040 board that is not open source ? (I asked on the maker's discord, if they would request one from RPF)
slender iron
ya, closed source is ok as long as its not your closed source design
jaunty juniper
ok, thanks π
idle owl
@slender iron What milestone would you file a documentation issue under? It's not an error, it's that a code snippet should be added showing how to use the feature in your code.py file.
Trying to get the labels and milestones on it while filing it to make less work for you folks.
Long-term I'm guessing, but it's a core dev designed system, so I figured it was worth asking.
tulip sleet
<@252717193496756235> What milestone would you file a documentation issue under?...
If you think it's worth updating the doc soon, put it under 8.x.x. I've been putting most new issues under 8.x.x. We can triage them later.
idle owl
Ah ok.
I'll do that. I do think it's worth updating. Good first issue label is on it too.
jaunty juniper
Neradoc. let me know if I can contact you by DM. Thanks π
yes, I might be long to answer, depending on the circumstances
slender iron
welp, my amazing perf numbers were a lie
blissful pollen
that's frustrating
slender iron
yup
I learned a lot....
perf benchmarks that time themselves aren't a great idea π
blissful pollen
Was that with that spall tool?
tame grove
cool! I've read up on machine code in mpy last night and I think that's probably the by far easier first step compared to going full native internal. Did poke around the build system and was equal parts intimiated + impressed, heh
slender iron
Was that with that spall tool?
no, the mp perf bench times itself
and I had messed with what time we return π€¦
jaunty juniper
say, there are some commits in 8.0.x that have not been backported to main yet
(like the web workflow stability fix)
manic glacierBOT
manic glacierBOT
lone sandalBOT
manic glacierBOT
thorny jay
Would you have the matching BIN file? I don't know how to boot the LILIN S2 Pico into UF2 mode.
thorny jay
Would you have the matching BIN file? I don't know how to boot the LILIN S2 Pico...
Double click does not mork.
rugged spindle
Click reset button
Wait until LED is lit
Hold BOOT button
I spent ages playing around and then found the instruction on the CP download page. Doh!
That works on the S2 Mini so might work on the S2 Pico
jaunty juniper
Would you have the matching BIN file? I don't know how to boot the LILIN S2 Pico...
oh I'll get you the bin too https://www.neradoc.me/builds/lolin_s2_pico-8.1.0-beta.0-23-gbca7d6e99-add-display-lolin-pico-dirty.bin
thorny jay
oh I'll get you the bin too <https://www.neradoc.me/builds/lolin_s2_pico-8.1.0-b...
Thanks.
The good news is that it boot and does not complain about I2C. But the bottom of the screen is random and the top is bizarre.
thorny jay
Thanks.
The good news is that it boot and does not complain about I2C. But the b...
I would say there is one line out of two, that is why you can guess Wi-Fi but do not see the "-" that is eaten.
Do you have a branch with the code for this, I could have a look and try to compile CP myself.
Let me at least share some python code that work with stock CP.
import busio
import time
import displayio
import adafruit_displayio_ssd1306
displayio.release_displays()
i2c = board.I2C() # uses board.SCL and board.SDA
oled_reset=board.LCD_RST
display_bus = displayio.I2CDisplay(i2c, device_address=0x3C, reset=oled_reset)
display = adafruit_displayio_ssd1306.SSD1306(display_bus, width=128, height=32)```
I hope no one is hurted that I keep the label on the screen... LadyAda style!
thorny jay
Click reset button
Wait until LED is lit
Hold BOOT button
Thanks, but so far I had zero success.
It might be some timing issue, but I tried so many time that it should have at least worked once.
I know this info might not belong to circuitpython.org, but the download page is maybe the only place where tips and ticks about non-Adafruit board can be shared. And the update procedure is the one I struggle most of the time.
@jaunty juniper here is your code but without using board.DISPLAY and it does 8x2 "square" pattern.
manic glacierBOT
jaunty juniper
<@199811382793863168> here is your code but without using board.DISPLAY and it d...
ok, had to decipher how to change the init sequence based on that https://github.com/adafruit/Adafruit_CircuitPython_DisplayIO_SSD1306/blob/11a9b9e4a28e60bc23c61ab3451f723c132184db/adafruit_displayio_ssd1306.py#L79-L89
maybe this build will work
https://github.com/Neradoc/circuitpython/tree/add-display-lolin-pico
thorny jay
ok, had to decipher how to change the init sequence based on that <https://githu...
Don't tell me about that piece of code... I was the last one to change it to support the 64x32 tiny oled I got from LILIGO. I am still unsure if it can be written better, and if it should be tested for "rotation" because I did not...
It works as expected.
That is perfect now. π π₯³
jaunty juniper
ok I'll PR it then
thorny jay
Maybe we can discuss how the text "wrap" on the prompt line, or how the prompt line does not blank the text that was there before.
But this is I believe a general issue, just exacerbated by the tiny screen.
On the serial I have ```Code done running.
Press any key to enter the REPL. Use CTRL-D to reload.
]0;Wi-Fi: off | REPL | 8.1.0-beta.0-24-gf532318d6-dirty
Adafruit CircuitPython 8.1.0-beta.0-24-gf532318d6-dirty on 2023-03-19; S2Pico with ESP32S2-S2FN4R2
The line on screen display >>> S2-S2FN4R2 with E and that is SP32S2-S2FN4R2 with E so the text with ESP32S2-S2FN4R2 wrapped where the 4 first character are overwritten by the REPL >>> .
If I that point I type spaces I can delete the offending S2-S2FN4R2 with E and get a clear prompt.
crimson ferry
@thorny jay do you want to use https://docs.circuitpython.org/en/latest/shared-bindings/supervisor/index.html#supervisor.StatusBar
(also, I print with line endings at the start of lines on small displays so there is no blank line at the bottom wasting space)
thorny jay
<@681528999520108564> do you want to use <https://docs.circuitpython.org/en/late...
Does that actually works? I have a white line with the Blinka logo, so the saved space is not used at all.
crimson ferry
there is some weird wrapping happening there, isn't there
thorny jay
We only have >>> on top of H*ll.
I forced to have spaces in my print to clear the space if anything is there.
jaunty juniper
yeah I did notice that when it's a small screen it wraps on itself
crimson ferry
would be nice if CIRCUITPY_REPL_LOGO was a boot.py thing
jaunty juniper
but I don't know what you would show in a REPL that small, the status bar shows the IP when you have one
thorny jay
Anyway, with such a size, seeing the REPL is not really the key function.
jaunty juniper
the API for the status bar and REPL is still rather wonky
tulip sleet
@jaunty juniper @thorny jay I will approve the PR if you think it is good enough.
thorny jay
<@199811382793863168> <@681528999520108564> I will approve the PR if you think i...
It is.
lavish saffron
How do I adding typing for a microcontroller.Pin?
try:
from typing import Union
import adafruit_pca9685 as pca9685
import pwmio
import microcontroller
except ImportError:
pass
...
def __init__(
self,
red_pin: Union[microcontroller.Pin, pwmio.PWMOut, pca9685.PWMChannel],
green_pin: Union[microcontroller.Pin, pwmio.PWMOut, pca9685.PWMChannel],
blue_pin: Union[microcontroller.Pin, pwmio.PWMOut, pca9685.PWMChannel],
invert_pwm: bool = False,
) -> None:
...
Works on my laptop, but fails in the Github CI tests:
937 NameError: name 'microcontroller' is not defined
938
939 Error: Process completed with exit code 2.
I'm perplexed!
still zephyr
How do I adding typing for a `microcontroller.Pin`?
```
try:
from typing i...
π digitalio.DigitalInOut would do
lavish saffron
π `digitalio.DigitalInOut` would do
No, I don't think it would. The __init__() takes a microcontroller.Pin and not a digitalio.DigitalInOut instance, e.g. board.D5 or board.GP20 which are instances of the class that are also passed to digitalio.DigitalInOut
onyx hinge
@lavish saffron Exactly what is failing? It is a step of CI? What command is running? Can you link to the full code & log?
adafruit_magtag/magtag.py is an example of a CP library that uses microcontroller.Pin in its type annotations
lavish saffron
adafruit_magtag/magtag.py is an example of a CP library that uses microcontrolle...
It's failing on the CI step that builds the Sphinx Docs after I opened a pull request on GitHub (the docs build on my local system):
917 Warning, treated as error:
918 autodoc: failed to import module 'adafruit_rgbled'; the following exception was raised:
919 Traceback (most recent call last):
920 File "/opt/hostedtoolcache/Python/3.11.2/x64/lib/python3.11/site-packages/sphinx/ext/autodoc/importer.py", line 60, in import_module
921 return importlib.import_module(modname)
922 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
923 File "/opt/hostedtoolcache/Python/3.11.2/x64/lib/python3.11/importlib/__init__.py", line 126, in import_module
924 return _bootstrap._gcd_import(name[level:], package, level)
925 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
926 File "<frozen importlib._bootstrap>", line 1206, in _gcd_import
927 File "<frozen importlib._bootstrap>", line 1178, in _find_and_load
928 File "<frozen importlib._bootstrap>", line 1149, in _find_and_load_unlocked
929 File "<frozen importlib._bootstrap>", line 690, in _load_unlocked
930 File "<frozen importlib._bootstrap_external>", line 940, in exec_module
931 File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
932 File "/home/runner/work/Adafruit_CircuitPython_RGBLED/Adafruit_CircuitPython_RGBLED/adafruit_rgbled.py", line 35, in <module>
933 class RGBLED:
934 File "/home/runner/work/Adafruit_CircuitPython_RGBLED/Adafruit_CircuitPython_RGBLED/adafruit_rgbled.py", line 90, in RGBLED
935 red_pin: Union[microcontroller.Pin, pwmio.PWMOut, pca9685.PWMChannel],
936 ^^^^^^^^^^^^^^^
937 NameError: name 'microcontroller' is not defined
938
939 Error: Process completed with exit code 2.```the link I was looking for: https://github.com/adafruit/Adafruit_CircuitPython_RGBLED/pull/23
@lavish saffron The true error is
$ python3 -c 'import pwmio'
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/tmp/_env/lib/python3.9/site-packages/pwmio.py", line 50, in <module>
raise NotImplementedError("pwmio not supported for this board.")
NotImplementedError: pwmio not supported for this board.
this causes the line import microcontroller to not be run
or the true error may be: ModuleNotFoundError: No module named 'adafruit_pca9685'
you can choose to:
from __future__ import annotationsbefore anything else, even before the docstring. this means that the type annotations will not be evaluated as Python expressions at import time- change specific annotations to be strings (e.g.,
red_pin: "Union[microcontroller.Pin...]") - there may be other possibilities
- you can try setting microcontroller, pwmio, and/or adafruit_pca9685 as mocked in sphinx
- you can add adafruit_pca9685 to requirements.txt
pwmio is already mocked in sphinx config so maybe it's adafruit_pca9685 that is the problem during doc build
lavish saffron
Thanks. I'll experiment and see which works best. I'd never have figured this out by myself. I don't think adding adafruit_pca9685 to requirements.txt is a good idea, because it's only needed for typing.
onyx hinge
-autodoc_mock_imports = ["pwmio"]
+autodoc_mock_imports = ["adafruit_pca9685", "pwmio"]
Yeah try this
If your types are not merely annotations then it is a requirement
lavish saffron
you can choose to:
* `from __future__ import annotations` before anything else,...
They're merely annotations.
onyx hinge
try:
import doesnt_exist_ever
except ImportError: pass
x: doesnt_exist_ever.whatever
this won't work, because doesnt_exist_ever.whatever has to be a Python expression that is evaluated at import time. On standard Python.
if you have x: "doesnt_exist_ever.whatever" then it's not evaluated as an expression at import time
or if you have from __future__ import annotations
still zephyr
Just try the mocking of the pca9685 module and that solve the issue, and its building the docs locally
onyx hinge
I'm talking about how its treated on desktop Python which is important since pylint and sphinx work on modules with desktop python, even if it's a module like magtag that's only ever "used" on circuitpython
lavish saffron
I'm talking about how its treated on desktop Python which is important since pyl...
I see, so from __future__ import annotations looks like the best solution.
lavish saffron
I'm talking about how its treated on desktop Python which is important since pyl...
Would it be appropriate to add the modules that are only used for type checking to optional_requirements.txt instead of requirements.txt?
onyx hinge
Would it be appropriate to add the modules that are only used for type checking ...
I don't know what optional_requirements is used for
lavish saffron
I don't know what optional_requirements is used for
I usually stick pytest and other testing modules in there if I'm writing unit tests. No idea apart from that. I read in PEP 563 https://peps.python.org/pep-0563/#runtime-annotation-resolution-and-type-checking that imports that are only used for type checking can be ignored at runtime so I think this may be appropriate:
try:
import typing
from typing import Union
if typing.TYPE_CHECKING:
import adafruit_pca9685 as pca9685
import pwmio
import microcontroller
except ImportError:
pass
So that the modules wouldn't need to be installed on systems that are only running the code.
orchid basinBOT
onyx hinge
I noticed this on ReadTheDocs .. does anyone know what we need to do to make 8.0.x appear here and be treated as the newest?
(old docs say " You are not reading the most recent version of this documentation. 7.3.x is the latest version available." and the version chooser in the readthedocs sidebar doesn't have an 8.0.x choice)
<@&356864093652516868> we are about 2 hours from the weekly Discord meeting! It's a great time to add your notes to the document: https://docs.google.com/document/d/1y7kZuukcZXC1eiStMaMsI_YlglAAoW_fMbaQysK_CAY/edit?usp=sharing
I look forward to catching up with y'all in the meeting over in the CircuitPython voice channel.
proven garnet
I don't know what optional_requirements is used for
It's meant for requirements that are functional but not required for primary use of the library. The best example are libraries that allow for use of PIL for some functionality or examples and are "opt in".
So you could do pip install library_name[optional] (I think that's how I have the syntax set)
proven garnet
I noticed this on ReadTheDocs .. does anyone know what we need to do to make 8.0...
I'm under the weather from a recent vaccine booster but let me check on this today
onyx hinge
that's .. interesting. A square wave is decoded by this logic analyzer as an apparently legal CAN frame
@proven garnet feel better soon .. and no need to push yourself over this, probably
proven garnet
Would it be appropriate to add the modules that are only used for type checking ...
So given the above, they would go in requirements.txt because they are fully required for primary use. A litmus test I use is if they are needed outside of typing. If yes, it's worth putting them in requirements.txt because Blinka will still need them to interop with the code. If it's opt in or only used in examples, its best to put it in optional_requirements.txt and use strings as jepler mentioned.
slender iron
(old docs say " You are not reading the most recent version of this documentatio...
I think this is a setting on rtd. lemme look
yup. I activated it under the versions tab
onyx hinge
@slender iron thank you
lone axle
I am getting these errors in the console trying to view the docs atm:
onyx hinge
@proven garnet @lavish saffron The other thing about rgbled is that maybe it's circuitpython_typing's pwmio.PWMOut that you want, because what's actually important is that there's a duty_cycle property on the object. Then you don't need to refer to that adafruit_pca9685 module at all. However, it checks the object for having a frequency property which isn't in the circuitpython_typing version of the type (not sure why it doesn't check for duty_cycle)
@lone axle and as a practical consequence the module support matrix filter doesn't work https://docs.circuitpython.org/en/latest/shared-bindings/support_matrix.html
I think they removed jquery at some deep level, we'll need to switch away from it or add it back? https://github.com/sphinx-doc/sphinx/issues/10070
lone axle
Yep, unable to filter. Clicking the version pop-up thing doesn't seem to work on my end either. Poking a bit further: I do not have the same problem if I build the docs locally. Looking at where it's loading loading from I'm seeing a different file. The live page uses this URL: https://docs.circuitpython.org/_/static/vendor/jquery.js and in a local build the file is in _build/html/_static/jquery.js
onyx hinge
https://github.com/nextcloud/documentation/commit/1ac225a0a6c1e51058f69ce1a20e2436ff2a8091 some folks have pinned a specific (and non-release!) version of sphinx-rtd-theme
Or maybe depend on sphinxcontrib-jquery? https://github.com/sphinx-contrib/jquery
tulip sleet
I noticed this on ReadTheDocs .. does anyone know what we need to do to make 8.0...
I will do this -- I forgot when I released 8.0.0. It's various settings in RTD.
tulip sleet
I noticed this on ReadTheDocs .. does anyone know what we need to do to make 8.0...
There appears to be Some Other Problem: the v:latest button is not sensitive as it should be.
lone axle
There appears to be Some Other Problem: the v:latest button is not sensitive as ...
I think it's related to the missing/changed jQuery dependency with some info mentioned and linked above here in a the chat a little.
digital shoreBOT
modern wing
Happily lurking today π
turbid radish
π
slender iron
π
gilded cradle
I'm available
idle owl
All good.
turbid radish
modern wing
π
idle owl
π Thanks!
turbid radish
I missed you too!
idle owl
Thanks!
lone axle
Had never heard of distox before and was curious:
DistoX Bluetooth Protocol - mimic the DistoX protocol for communicating with paperless cave surveying tools
Thanks everyone!
gilded cradle
Thanks
modern wing
Thanks everyone for an excellent meeting!
errant grail
Thanks!
ember iris
Thanks everyone, have a great week!
lone axle
Thanks for hosting Jeff, have a great day everyone.
slender iron
thanks all!
onyx hinge
Here is the notes document for next Mondayβs 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 youβll be attending the meeting - itβs super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and weβll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1l8zP4t9RGURIoeS_foo5jcmmiRLUWkAJ8NpuB6i30cs/edit?usp=sharing
manic glacierBOT
midnight ember
Missed the meeting today. Was really nice to hear @idle owl again. π Welcome back.
idle owl
Missed the meeting today. Was really nice to hear <@330227457296957440> again. ...
Thanks!
still zephyr
in the design guide we have stated that when printing we should use the " ".format() option, as some boards are not capable of using f-strings, have this changed recently?
stuck elbow
m0 boards still can't fit f-strings
still zephyr
Thanks (y)
onyx hinge
(though counter-intuitively an mpy-compiled module that uses f-strings can be used on even those boards)
stuck elbow
oh, right, because it's just a parser hack
stuck elbow
if we did, I missed that
jaunty juniper
yeah since 7.3.0
tulip sleet
we tried really hard to get f-strings on for everything, because otherwise we have to write examples that don't use f-strings
onyx hinge
oh I forgot about that. yays.
still zephyr
Got it, thanks.
slender iron
dan or jeff want to review my performance PR?
vale spire
hi! we're looking into some fixes for os.dupterm, which I know CircuitPython doesn't use, but you must have solved a similar problem for the various BLE/WiFi workflows?
(i'm spelunking right now, but still figuring out where to look exactly... in particular I'm curious how you handle things like Ctrl-C and whether this works for user streams or is specifically handled for the various workflows)
(actually i'm also not sure whether it does work like dupterm -- if you're using the wifi workflow, does stdout still get mirrored to the UART/VCP?)
speaking of... I guess this also applies to just regular UART/VCP... do you mirror stdout and support input from both between a default UART and the VCP on the tinyusb-capable ports?
tulip sleet
hi! we're looking into some fixes for os.dupterm, which I know CircuitPython doe...
hi - maybe you are looking for this?? https://github.com/adafruit/circuitpython/blob/main/supervisor/shared/serial.c
we don't usually do both UART and USB CDC, but I think it does work. Also we mirror the REPLoutput onto an on-board display if present
manic glacierBOT
vale spire
hi - maybe you are looking for this?? <https://github.com/adafruit/circuitpython...
yes! thank you!
and to answer my own question Ctrl-C appears to be handled in each of the sources individually (much like we do)
so, just like micropython, writes to stdout are blocking across all destinations, and poll-for-writing on stdout is not implemented. So https://github.com/micropython/micropython/issues/11026 might be of interest to you.
lavish saffron
<@135415127879974912> <@369681896034009101> The other thing about rgbled is tha...
Thanks to you and @proven garnet for all your help and advice. I've gone with putting the types in string quotes and that passes the CI. I'll have a look at duty_cycle v frequency when I get home to a Raspberry Pi, as I plan to try to solve the Raspberry Pi unrecognised pins issue.
bronze kernel
So, there is a circuitpython port for stm32f411 controllers, specifically the black pill board. There is another variant of this board sold with the f401 chip. Is porting circuitpython to this chip possible or for what reasons it isnβt?
stuck elbow
So, there is a circuitpython port for stm32f411 controllers, specifically the bl...
I don't know, but I would start by comparing the flash and ram available on those chips.
bronze kernel
From what I can see, the f401 has half as much ram as the f411. In my case (f401cc) it is 64 kb
stuck elbow
From what I can see, the f401 has half as much ram as the f411. In my case (f401...
that answers it
bronze kernel
What would be the minimum ram requirement for circuitpython? I canβt seem to find it
bronze kernel
that answers it
I believe there is a micropython port for it
stuck elbow
it varies depending on the port
yeah, micropython has looser requirements, it can have many things disabled, so it has more ports
bronze kernel
Gemma M0 has 32MB of ram, and it is still supported
stuck elbow
the minimum is around 64kB, but the closer you get to that minimum, the more work it takes
jaunty juniper
my understanding too is that STM chips are tricky, they have different configurations of how the RAM is split between different blocks or whatever, and other odities
stuck elbow
you can try porting it, but it will be tricky
bronze kernel
What a nice opportunity to get familiar with stm chips!
I had no prior experience working with stm before lol
gotta read that datasheet, I guess
small mantle
how can I clone a branch?
For research I cloned the adafruit repo and want to check something between 7.3.3 and main
stuck elbow
if you have cloned the repository then all the branches are in there, you can checkout them as you want
also, there is #help-with-git
small mantle
oh yes sorry
tulip sleet
I had no prior experience working with stm before lol
The Meowbit is an STM32F401 and we have a port for it
bronze kernel
The Meowbit is an STM32F401 and we have a port for it
Oh, thanks! Then, it just goes to making a proper board definition
slender iron
yes! thank you!
and to answer my own question Ctrl-C appears to be handled in e...
Also note that we have a separate CDC for the user to use for cases like this. That way we don't need to worry about them ctrl-cing their code. π
@tulip sleet or @onyx hinge want to review my imx improvements?
tulip sleet
@vale spire second CDC channel is described here: https://learn.adafruit.com/customizing-usb-devices-in-circuitpython/circuitpy-midi-serial#usb-serial-console-repl-and-data-3096590
onyx hinge
@slender iron I'll dive in right now
slender iron
thank you!
slender iron
I'm gonna rework how we integrate the sdk this week
hal should be the same though
gonna spend a little time chasing a crash on samd21
(that perf bench triggers)
onyx hinge
wow pull request review really needs some keyboard shortcuts, like "mark as viewed", "next/prev file"
slender iron
have you tried vimium?
0x20007f2c
heh, what's it say about me that my clipboard contents is a memory address?
slender iron
@onyx hinge was your think ink rp2040 def for the feather form factor board or something else?
onyx hinge
@slender iron I would have said I thought it was, but based on the pins.c file it doesn't look feather-ish at all
yeah it's not, there's a different non-feather-format board that's probably been teased but is certainly not released
(not that the feather is released either)
onyx hinge
hooray! I finally have audio coming out of this little speaker!
@slender iron π’ ```
build-metro_m7_1011/firmware.elf section .itcm' will not fit in region ITCM'
region `ITCM' overflowed by 912 bytes
I'd suggest looking at the map file to see what has been added to that section
do you have debug on?
onyx hinge
no.
actually it's before enabling mp3, hmm, I just rebased my i2s work onto main and assumed without checking that it still worked
did we switch compilers for main?
I'm still on gcc-arm-none-eabi-10-2020-q4-major
slender iron
I don't think so
slender iron
onyx hinge
@slender iron I had -UNDEBUG which turns back on assert messages, I think; and it was probably causing the problem
still ITCM is 96.90% full for me, shivers
slender iron
a lot of that is optional and we can move it out
I've been thinking about the ideal setup for it
theres error handling stuff that should always be in there
onyx hinge
I am guessing that mp3 decoding & audio mixing would benefit from having some key part in ITCM but it looks like that's not likely to happen
slender iron
and then everything else should really be performance guided
ITCM should really only be used for things that are always used
there is an icache for other code
onyx hinge
OK mp3 fits now without affecting ITCM use
slender iron
π
onyx hinge
sorry for the false alarm! and thanks for the quick guidance
slender iron
np, I know its new
llvm/clang has a monthly meeting about making it work for embedded
I think PGO is the way to go for fast memory placement (versus detailed linker scripts)
onyx hinge
oops there went my CIRCUITPY drive
onyx hinge
Adafruit CircuitPython 8.1.0-beta.0-49-g06fbd521c5-dirty on 2023-03-21; Metro MIMXRT1011 with IMXRT1011DAE5A
>>> import audiomp3
>>> import audiomp3
>>> r = audiomp3.MP3Decoder(open("/alot-128.mp3", "rb"))
Traceback (most recent call last):
[tio 12:29:28] Disconnected
```next thing to debug, creating an MP3 decoder object fails. boo.well, the 64kbit/s encoded version works, so yay?
onyx hinge
@slender iron I added new interrupts, do those have to be in ITCM?
namely this is called from interrupt context ```static void i2s_transfer_callback(I2S_Type *base, sai_handle_t *handle, status_t status, void *self_in) {
i2s_t *self = self_in;
if (status == kStatus_SAI_TxIdle) {
// a block has been finished
background_callback_add(&self->callback, i2s_callback_fun, self_in);
}
}
Only if you want them to run during flash writes
onyx hinge
OK, so that's effectively "yes"
slender iron
Iβm not sure how often user code will write flash and playback audio at the same time
onyx hinge
@slender iron rewriting CIRCUITPY via USB mass storage is something most folks will do
idle owl
Are we intentionally still including 7.x on circuitpython.org/libraries? At this point, I don't recall what our thoughts were on this, whether we always keep the current stable and most recent stable, or what the plan was there. I'm currently doing a bit of work on that page, so if that needs to be updated, I can do so. But I also don't want to remove it if it's intended to remain.
slender iron
<@252717193496756235> rewriting CIRCUITPY via USB mass storage is something most...
right, but why do you need audio running at the same time?
onyx hinge
@slender iron what's the alternative? make CIRCUITPY read-only to the host when audio's running?
slender iron
a write to CIRCUITPY will cause a reload anyway
so why not just stop the audio?
or have the interrupt in flash
onyx hinge
right now that's not what it does, though nobody's happy about the skipping audio during CIRCUITPY writes
slender iron
much of the time you'll be loading more audio from flash anyway
so it doesn't make sense to have the interrupt run while flash is unavailable
onyx hinge
there's 2 parts. there's the interrupt, which is small and only registers work to be done from background callback
do you mean just masking the interrupt temporarily, or reaching into running audio objects and stopping them?
hm, SAI_TransferTxHandleIRQ / I2S0_DriverIRQHandler are just in text so maybe this is either a red herring or an incomplete fix. What I'm chasing is, the board just crashing with the new audio code. and no I haven't attached a debug probe to it yet.
but only sometimes, and usually when I have tried to write CIRCUITPY
slender iron
I don't think the interrupt needs to be in ITCM because it'll be masked during flash erase and write
definitely attach the debugger
onyx hinge
yeah tomorrow
it's too bad I believed the ITCM change made it not crash, I must have changed something else in my testing procedure too
slender iron
you could put it back into the ITCM to see if it fixes it
tulip sleet
Are we intentionally still including 7.x on circuitpython.org/libraries? At this...
I think there are still cases where we might recommend going backwards in versions if there is a new problem. I think there are some regressions that haven't been fixed yet. I thought we'd leave it for a couple more months, maybe at least until 8.1.0 final is out.
still zephyr
When I am building circuitpython with the latest commit https://github.com/adafruit/circuitpython/commit/461d833c1c9ead38542dbbfe665af5c3446035bd for the Pyportal Titano, Circuitpython is crashing and taking me to safe mode. is there any change that we have included since yesterday that could cause this. I build CP in https://github.com/adafruit/circuitpython/commit/08c9eb9f00a4c88d0a81f9b6d5e4298d021ca74c and it works fine without crashing.. I did make clean, and V=2 if anyone is interested in see the output.
tulip sleet
When I am building circuitpython with the latest commit https://github.com/adafr...
I will test this.
tulip sleet
When I am building circuitpython with the latest commit https://github.com/adafr...
Did you do make fetch-submodules?
still zephyr
No I did not, should I test with that?
tulip sleet
Yes, I am updating my fork and will try building from that. I built Metro M4 and tested that before I submitted the PR, but I will test PyPortal and PyPortal TItano
still zephyr
ok, will do
tulip sleet
ok, will do
Yes, I am seeing it crash into safe mode. I'll debug
still zephyr
Yes, I am seeing it crash into safe mode. I'll debug
Ok Thank you very much, I do not have a powerful machine so I did not bisect π
tulip sleet
I have to bisect tinyusb, I think
the other changes should not be making it crash -- they are just git things
still zephyr
got it. thanks again
tulip sleet
got it. thanks again
The problem is further back than the latest PR, so I am bisecting. Thanks for giving a good commit.
still zephyr
No worries, thank you
wraith crow
I'm seeing a problem with web workflow on the Pico W and the latest S2 build that also wasn't happening on the 8.0.4 S2 release. As soon as I try and connect via web workflow (serial terminal or file browser) the board crashes. If I'm connected to the REPL an OSError: [Errno 11] EAGAIN is displayed. Do you think it's related to the PyPortal issue you're looking at or should I open an issue?
tulip sleet
I'm seeing a problem with web workflow on the Pico W and the latest S2 build tha...
still bisecting, not sure yet. But I saw that EAGAIN error, and then stopped seeing it after some fix.
I think it was in 8.0.3, fixed in 8.0.4, but not sure I remember right
wraith crow
Do I need to update something to get the new make fetch-submodule function to work? I build on Debian and git doesn't seem to like git submodule update --init --filter=blob:none
jaunty juniper
what do you mean ?
wraith crow
I just cloned a new copy of circuitpython and when I attempt to run the make fetch-submodules, it now runs git submodule update --init --filter=blob:none which the git on my install rejects with a usage error. If I go back to the makefile in 8.0.4 the fetch-submodules function consisted of two different commands so I'm assuming the one that' s failing for me is an updated procedure
I am running the old version now but I assume the new version is the preferred way to go
tulip sleet
I just cloned a new copy of circuitpython and when I attempt to run the make fet...
what version of git are you using?
I tried to find out when --filter was added to git submodule update` but it was not immediately obvious.
You can use the old make fetch-submodules, it is just a little bit slower, and it creates shallow clones, which can be a bit of a nuisance if you want to change commits in the submodule.
wraith crow
git version 2.30.2
wraith crow
Linux 5.10.0-21-amd64 #1 SMP Debian 5.10.162-1 (2023-01-21) x86_64 GNU/Linux
I'm not sure why my git isn't updated to the latest, but I assume it will make it to my distro eventually. If the shallow clones are causing the issue I think they are then I'm happy the new method is in place, now I just have to dig in and try and figure out how to get 2.40 π
tulip sleet
I looked at the .0 release notes for the releases between 2.30.0 and 2.40.0, and I can't see which one added it. But there are many more bugfix releases in between, and the descriptions are such that something obscure might have added this feature.
what is /etc/issue on your system?
suggests using the -backports package for your release
wraith crow
pi@Rohan:~$ cat /etc/issue
Debian GNU/Linux 11 \n \l
tulip sleet
i can make the target be more robust
do apt show package-name -a to find out about backports
apt show git -a
wraith crow
Well your link looks promising, I'm a little slow in Linux as the only thing I do on it is build CP. I suspect you've dug up all I need π Thanks!!!
pi@Rohan:~$ apt show package-name -a to find out about backports
N: Unable to locate package package-name
N: Unable to locate package to
N: Unable to locate package find
N: Unable to locate package out
N: Unable to locate package about
N: Unable to locate package backports
E: No packages found
tulip sleet
let me know what the backports version is and if it works; that will give us some indication package-name is a metavariable, like "foo". Do apt show git -a
wraith crow
tulip sleet
@wraith crow ok, you have to add the backports repository first. Start at the top of that page: https://wiki.debian.org/Backports
add the repo either via synaptic or via the command line
after it is added, then you can do apt show package-name -a
manic glacierBOT
wraith crow
Okay the apt show git -a now indicates that 2.39.2 is available, I'll try installing that
It looks like 2.39.2 has the feature the new make fetch-submodules is working now.
tulip sleet
I should probably make it do it the old way if the command fails.
wraith crow
Probably, but I'm glad it failed for me first so I could get the new version. If it's possible to throw a message up after the old way completes hinting that there's a better way that might be worth while, unless you think all distros will get the 2.39 or later version relativly soon.
Another nice feature of the new command is that I often have to run the fetch-submodules command more than once because of failed submodule clones, with the old version it was hard to tell when everything was complete, but this version just comes back with nothing if there's nothing to do
manic glacierBOT
jaunty juniper
[edited: issues with fetch-submodule, but I'll see if it's a local problem fixed with recloning]
manic glacierBOT
onyx hinge
anybody run into an error like this? ```jepler@eric:~$ pyocd list
Probe/Board Unique ID Target
0 Segger J-Link 50128794 n/a
jepler@eric:~$ pyocd gdbserver -t MIMXRT1010
0000798 I Target type is mimxrt1010 [board]
0000935 C No emulator with serial number 50128794 found. [main]
Traceback (most recent call last):
File "/home/jepler/.local/pipx/venvs/pyocd/lib/python3.9/site-packages/pyocd/probe/jlink_probe.py", line 168, in open
self._link.open(self._serial_number_int)
File "/home/jepler/.local/pipx/venvs/pyocd/lib/python3.9/site-packages/pylink/jlink.py", line 723, in open
raise errors.JLinkException('No emulator with serial number %s found.' % serial_no)
pylink.errors.JLinkException: No emulator with serial number 50128794 found.
"sudo" doesn't help
slender iron
@onyx hinge I haven't seen that. I just use JLink's software though
JLinkGDBServer -if SWD -device MIMXRT1011xxx5A
onyx hinge
yeah I did eventually get it working that way
and then couldn't reproduce my crashes π
slender iron
π
I think I'm going to write a script to generate the pin mappings
so I can fill in the i2s stuff for you
onyx hinge
so I can fill in the i2s stuff for you
thanks, I think I finished the lists for the 3 chips we have now but it was manual so there are errors
slender iron
I'm planning on adding the rest of the chips so a python script will make it simpler
my sdk update makes it all easier
tulip sleet
i haven't yet tried pyocd, sorry
slender iron
I'm not sure pyocd does flashing with external flashes that well
slender iron
@onyx hinge want me to wait to merge the sdk changes? it'll conflict with your i2s PR
slender iron
π
onyx hinge
if it's not, well,
slender iron
yup, just changing the location of the added source file
slender iron
don't forget to git submodule sync to pick up the submodule source switch
tulip sleet
I should probably add git submodule sync to make fetch-submodules
tulip sleet
maybe you saw last night it requires a pretty new git (somewhere past 2.30). bullseye doesn't have it by default, but it's in backports. Not sure if I should add a fallback or just document it. I need to find the minimum version: the release notes are not that helpful
onyx hinge
I already have the backports one on my debian bullseye (but I also just directly enter git commands rather than using the makefile for submodules; I should reeducate myself on that)
tulip sleet
it SHOULD be an option in .gitmodules, I would say, but this is better than nothing
bronze kernel
ok
I am trying to build circuitpython from source
But make says that it has no rule for building peripherals/samd/samd21/adc.c
(i am trying to build it for a circuit playground express as a test)
tulip sleet
you need to do make fetch-submodules at the top level to get all the git submodules brought in
tulip sleet
what did it print when you did that? It newly requires a recent version of git, and perhaps it just printed an error.
what OS are you doing it on?
bronze kernel
endavourOS
tulip sleet
what does git --version say?
bronze kernel
git version 2.40.0
tulip sleet
that should be fine
yes, try it again, and see whether ports/atmel-samd/peripherals exists and has files in it
tulip sleet
that isn't good: what does it print when you do make fetch-submodules
nothing? It should take several minutes
is this your fork? Did you update main in it, if it is?
bronze kernel
git submodule update --init --filter=blob:none
yes, it is my own fork
and i am on my branch
bronze kernel
still nothing
tulip sleet
testing it myself...
try a fresh clone of your repo under another name, e.g. git clone https://github.com/yourname/circuitpython 1circuitpython; cd 1circuitpython; make fetch-submodules
it is working for me on a fresh clone
slender iron
(I'm on arch so its unlikely an OS issue)
tulip sleet
is endeavour arch?
slender iron
arch based
bronze kernel
is endeavour arch?
arch with gui installer and dracut
seems to have worked!
thanks!
tulip sleet
not sure what is up with your other clone, but great!
idle owl
I think there are still cases where we might recommend going backwards in versio...
Ok. Sounds good.
jaunty juniper
@onyx hinge sould it do n_args+1, i+1 ? (because the format is an additional argument)
(void)mp_arg_validate_length(n_args, i, MP_QSTR_args);
>>> struct.pack("3H", 1, 4)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: args length must be 3
or is it clear enough ?
it's not the same as function takes x positional arguments but y were given
I guess I answered my question !
onyx hinge
<@426200547024961539> sould it do `n_args+1, i+1` ? (because the format is an a...
I'm open to that being better but also keep in mind (I didn't think about it before now) it could be pack_into which had more arguments than pack...
jaunty juniper
no, my bad, it makes sense that it's the size of the vararg
onyx hinge
or maybe there's another QSTR that could be used, such as MP_QSTR_values
standard documentation uses that word: "Return a bytes object containing the values v1, v2, ... packed" as does our docstring def pack(fmt: str, *values: Any) -> bytes:
jaunty juniper
it is called values in the docs
onyx hinge
Thanks @jaunty juniper that makes it better
idle owl
Uff, who wants to help me with an error I'm encountering trying to update my local copy of CircuitPython?
git pull is happy, says up to date. make fetch-submodules on the other hand, got through part of the update, then threw this error. Rerunning it changed nothing except that it doesn't go through any updates, it simply throws the error now. ```
$ make fetch-submodules
git submodule update --init --filter=blob:none
fatal: 'adafruit' does not appear to be a git repository
fatal: Could not read from remote repository.
Please make sure you have the correct access rights
and the repository exists.
fatal: Fetched in submodule path 'ports/broadcom/peripherals', but it did not contain d3a6b50a21e7dd49ba4bfa0374da3407594caa50. Direct fetching of that commit failed.
make: *** [Makefile:332: fetch-submodules] Error 128```
It ran earlier...
tulip sleet
`git pull` is happy, says up to date. `make fetch-submodules` on the other hand,...
is your clone in a case-sensitive filesystem?
idle owl
That's the unusual one right? I created a separate DMG for it to do exactly that.
So yes I believe so
tulip sleet
ok, good
try doing git submodule sync, and then repeat make fetch-submodules. If that doesn't work, can you just do a fresh clone?
idle owl
Probably on the fresh clone. Was pretty proud of not having needed to do that in a long time though. π
I'll try that first.
sync did a thing, and then make throws the same error.
I guess it's fresh clone time. Bleh.
Unless anyone else has run into this before and has a suggestion to try.
tulip sleet
You could try git clean -ffdd
idle owl
Ooh, what does that do
tulip sleet
I think it might clean out the submodules ... no it didn't
tulip sleet
@jaunty juniper did you have any suggestions?
idle owl
I have one untracked directory that I need to dig out of there, a new board def I think I was working on.
Otherwise, I don't think there's anything else in progress.
jaunty juniper
no I have been having issues with fetch-submodules but sync solved the last one
it was that:
β― make fetch-submodules
git submodule update --init --filter=blob:none
fatal: remote error: upload-pack: not our ref 2b9354539e6e4f722749e87b0bdc22966dc080d9
fatal: Fetched in submodule path 'ports/mimxrt10xx/sdk', but it did not contain 2b9354539e6e4f722749e87b0bdc22966dc080d9. Direct fetching of that commit failed.
make: *** [fetch-submodules] Error 128
ok, got to go cook but I will do a little more research. I have seen like three unique problems which I haven't duplicated
idle owl
Hmm.
I'll rename the current clone directory, and keep it around for testing or trying to fix.
I'm kind of blocked on this for something else, so it's fresh clone time.
jaunty juniper
I had another one in a different working directory, but that one might have been due to weird things on my side, so i recloned, then had weird problems again when switching branches that solved themselves by a checkout
(some submodules had been left in a modified state)
tulip sleet
i would like to find a way to "uninit" and remove the submodule fetches
timid bolt
Hi everyone! I've been out sick for the past 1+ weeks, but I'm starting to come back now. I'm a little sad that the PRs I sent 2 weeks ago haven't attracted any attention. π₯
idle owl
Hi everyone! I've been out sick for the past 1+ weeks, but I'm starting to come ...
Glad to hear you're feeling better and making your way back!
tulip sleet
Hi everyone! I've been out sick for the past 1+ weeks, but I'm starting to come ...
i'm sorry I haven' been able to follow up. I have been thoroughly busy getting releases out. There will be another beta 8.1.0 soon. There's a lot of short-term stuff to fix, and since asyncio does work now, to rework it does not get as much attention, even though in the long run it is important
@jaunty juniper @idle owl Ooh neat git submodule deinit --all (or you can deinit a particular one). That might help.
tulip sleet
you could try it on the one you moved aside. adding -f will deinit them even if there are changes inside
idle owl
Ooh ok
make fetch-submodules ran through a bunch of things, and then threw the same error.
idle owl
you could try it on the one you moved aside. adding `-f` will deinit them even i...
I ran it with -f.
It did a lot of things.
I'll keep this clone around for now though.
tulip sleet
@blissful pollen21 you just did something which added a huge number of changes. Can you back that out? just pull your branch and you will pick up my single commit change. You can do a git reset --hard 4d09afa to just go back to your doc addition and then do a git push -f
blissful pollen
<@228322543507865601>21 you just did something which added a huge number of chan...
Okay I will in a few. Feel free to stop it. I had tried to rebase to pick up some doc changes before I made others
tulip sleet
Okay I will in a few. Feel free to stop it. I had tried to rebase to pick up som...
we could merge the current PR and you could submit a separate doc PR
blissful pollen
Okay makes sense I'll revert it back. Sorry about that
tulip sleet
i am just worried about reversions to main from such a large PR
blissful pollen
With the reset will it set it back before I did the rebase then? I think I can find the original commit for my branch still
ugh sorry I missed that commits went to 132 π¦ now I see clearly why that was bad
tulip sleet
With the reset will it set it back before I did the rebase then? I think I can f...
I added 4ff586e, and then 4d09afa was the doc change. You could hard reset back to that
blissful pollen
okay thanks I'll try that. I know I should just avoid anything crazy in git by now!
blissful pollen
Fingers crossed I think I fixed it (not perfectly maybe but removed all the commits)
tulip sleet
Fingers crossed I think I fixed it (not perfectly maybe but removed all the comm...
it looks ok, I will do a local diff to make sure, and wait for the builds
blissful pollen
it looks ok, I will do a local diff to make sure, and wait for the builds
Sorry again about that. I hate making more work for you. I just need to remember avoid rebase and realize merges aren't that bad
tulip sleet
Sorry again about that. I hate making more work for you. I just need to remember...
I am not actually sure why the rebase didn't work. I have done that before. But I have also had rebases go haywire. I once went to a talk where someone was extolling the virtues of rebase, using a very simple example, and managed to totally mess it up. So it made me quite wary of rebase. Don't worry about the work -- it's part of the review process.
blissful pollen
Dan just so you know somehow I lost my first commit but I know what was in it / can find it so I'll recommit it hopefully later tonight.
tulip sleet
Dan just so you know somehow I lost my first commit but I know what was in it / ...
got it - I fixed one thing, a missing curly brace, which was confusing, but I'm guessing there was more. I have your previous branch in another clone, if you can't find it.
blissful pollen
I'm not even sure how I lost that brace. It was just the definition of deinit in OnDiskGif so simple enough to put back
tulip sleet
I'm not even sure how I lost that brace. It was just the definition of deinit in...
what I have
blissful pollen
Thanks I have the commit id on my own machine too if i need. Just will be a few till I can get to it. Not sure when you're aiming for the next release if that matters
tulip sleet
not for a couple of days at least, I think
blissful pollen
Okay I'll get to it before then for sure
manic glacierBOT
onyx hinge
on metro m7 I've been seeing that it's easy to get it into a state where ctrl-c does't work to send KeyboardInterrupt. this tends to happen if I accidentally type something in on the repl but my program is never going to read it. I think maybe @slender iron made the USB CDC buffer really small to reduce RAM footprint? but the newer "serial break" feature, if you can get your terminal program to do it, does succeed in causing a KeyboardInterrupt.
manic glacierBOT
ESP32-S3 PSRAM (non-TFT) working with 2.9 EPD - UC8151D using Adafruit_UC8151D nightly build Adafruit CircuitPython 8.1.0-beta.0-58-gdb76fbd55 on 2023-03-23; Adafruit Feather ESP32S3 4MB Flash 2MB PSRAM with ESP32S3.
Haven't tested the TFT with 8.1.0-beta.0-58-gdb76fbd55 with the UC8151D
Have not tested the ESP32-S3 PSRAM (non-TFT...
still zephyr
@idle owl Hello, do you think it would be a good Idea to use the picture that they used yesterday in ask an engineer, celebrating the 100th libraries in the community bundle, in the readme of the repo?. I could put it there if the answer is yes π
random junco
I like that idea. π
idle owl
<@330227457296957440> Hello, do you think it would be a good Idea to use the pic...
Sure! I can get you a higher resolution copy of it. Give me a moment.
idle owl
<@330227457296957440> Hello, do you think it would be a good Idea to use the pic...
Use whichever one you like better!
still zephyr
Thank you !!!\
ornate breach
Look how cute blinka looks in that banner π
idle owl
I came up with the design, Bruce made it a beautiful reality. π
still zephyr
I like it very much!
slender iron
on metro m7 I've been seeing that it's easy to get it into a state where ctrl-c ...
I thought it was set as 512 bytes still. could be something with the HS path in tinyusb
onyx hinge
or something from the upgrade to tinyusb?
onyx hinge
I have to actually hook up some speaker or headphones but maybe "mqs" (PQMAudioOut) is starting to work too. top are L/R channels as digital, bottom is as analog with the saleae sampling at just 15kHz. The carrier frequency is about 192kHz so it "should be" un-hearable.
some musical notes
tulip sleet
@slender iron ok, we are down to 22 PR's after I closed some stalled ones.
and we finished off some recent ones
manic glacierBOT
slender iron
thanks @tulip sleet !
idle owl
@tulip sleet The CircuitPython install template still has "entering safe mode in 6.x" separate from 7.x, which I assume is the same as 8.x. Should we remove the 6.x at this point?
Either way, I need to explain that 7.x is the same if it is.
There's no mention of 8.
Only the RP2040 install template has it.
tulip sleet
<@329766224093249548> The CircuitPython install template still has "entering saf...
I think that's fine. you could say 7.x and later. You could add the 6.x instructions to the FAQ or Troubleshooting, or some Older Versions page if that makes sense. But I don't see any reason to mirror 6.x insturctions everywhere.
idle owl
Ok.
tulip sleet
I saw this in the Lyre guide, seemed a little excessive
idle owl
Fixed.
tulip sleet
@onyx hinge @slender iron I am starting on release notes for 8.1.0-beta.1, can wait for #7785 above, etc., not in a big rush
68 PR's already
tulip sleet
really helps to give perspective on how much work is going on; I don't mind the review
onyx hinge
is it new to have the logos next to the roles?
random junco
added this morning
onyx hinge
aha
digital shoreBOT
adafruit
tulip sleet
@onyx hinge you saw my comment about I2SOut in use after ctrl-D, is that right?
manic glacierBOT
brazen hatch
Spamming exec still crashes with a large enough pystack.
I am inclined to believe heap is at fault somehow.
With stock pystack though it all works fine.
I still have no idea how to decode the cpu crash dumps.
manic glacierBOT
slender iron
I still have no idea how to decode the cpu crash dumps.
there is a python script in ports/espressif/tools
you give it the elf file and then copy and paste in the Backtrace: lines
tulip sleet
This is better in #help-with-circuitpython . Let's move there.
hollow crater
This is better in <#537365702651150357> . Let's move there.
ok
idle owl
The Windbond chip is the QSPI flash, right? I mean it's the only thing on this board that's the proper size, but I wanted to verify anyway.
tulip sleet
The Windbond chip is the QSPI flash, right? I mean it's the only thing on this b...
yes
idle owl
Thanks!
lone axle
Has anything changed with regards to make fetch-submodules I'm getting this error trying to run that on a fresh cloned repo:
β― make fetch-submodules
git submodule update --init --filter=blob:none
usage: git submodule [--quiet] [--cached]
or: git submodule [--quiet] add [-b <branch>] [-f|--force] [--name <name>] [--reference <repository>] [--] <repository> [<path>]
or: git submodule [--quiet] status [--cached] [--recursive] [--] [<path>...]
or: git submodule [--quiet] init [--] [<path>...]
or: git submodule [--quiet] deinit [-f|--force] (--all| [--] <path>...)
or: git submodule [--quiet] update [--init] [--remote] [-N|--no-fetch] [-f|--force] [--checkout|--merge|--rebase] [--[no-]recommend-shallow] [--reference <repository>] [--recursive] [--[no-]single-branch] [--] [<path>...]
or: git submodule [--quiet] set-branch (--default|--branch <branch>) [--] <path>
or: git submodule [--quiet] set-url [--] <path> <newurl>
or: git submodule [--quiet] summary [--cached|--files] [--summary-limit <n>] [commit] [--] [<path>...]
or: git submodule [--quiet] foreach [--recursive] <command>
or: git submodule [--quiet] sync [--recursive] [--] [<path>...]
or: git submodule [--quiet] absorbgitdirs [--] [<path>...]
make: *** [Makefile:332: fetch-submodules] Error 1
yes you need an upd-to-date version of git
lone axle
This page in the docs mentions Ubuntu 22.04 and shows sudo apt install git which is what have / have done as far as I am aware π«
jaunty juniper
yeah that change is causing a lot of problems, I don't think we fully know what it does even
lone axle
Is it completely impossible to do it with the an older version? I feel like I did it very recently. I'm willing to try some workarounds if some are available.
jaunty juniper
you could use the old fetch-submodules ?
git submodule update --init -N --depth 1 || true
git submodule foreach 'git fetch --tags --depth 1 origin $$sha1 && git checkout -q $$sha1'