#circuitpython-dev
manic glacierBOT
slender iron
thanks for the testing @wraith crow
manic glacierBOT
onyx hinge
It turns out that a fisheye camera is NOT good for trying to scan a QR code with, qrio doesn't correct for optical distortion at all. that's part of why it was being so moody when I tried today
stuck elbow
I thought qr codes were designed for that?
onyx hinge
there are probably better libraries out there? but this one is good for a microcontroller
stuck elbow
I dream of one day porting the apriltags code from openmv
onyx hinge
Would love to see you do it!
stuck elbow
it's way over my head currently, but who knows
onyx hinge
whaaaaat? how did Intel manage to create a graphics driver that couldn't show QR codes? https://www.intel.com/content/www/us/en/support/articles/000059259/graphics.html
Images and QR codes viewed with Intel® graphics driver 27.20.100.9466 or earlier are distorted, causing them to be not readable by phone camera.
stuck elbow
well, it's Intel, they have decades of experience with making broken things
manic glacierBOT
tidal kiln
is there a good pin out or other ref somewhere that shows what pins on the RPI pico are PWM channel A vs B?
Adafruit CircuitPython 7.3.2 on 2022-07-20; Raspberry Pi Pico with rp2040
>>> import board
>>> import countio
>>> counter = countio.Counter(board.GP8, edge=countio.Edge.RISE)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
RuntimeError: Pin must be on PWM Channel B
>>>
idle owl
is there a good pin out or other ref somewhere that shows what pins on the RPI p...
Um.... The datasheet would have it. That's the first place I go with RP2040, tbh. Unless there's PrettyPins or our own Pinouts pages.
blissful pollen
is there a good pin out or other ref somewhere that shows what pins on the RPI p...
Rp2040 datasheet has it (page 523) just happen to have it open.
Basically A vs B alternate even is A odd is B. And pins 0-15 pair with 16-30
tidal kiln
thanks. looking for something less low level, to point forum user to.
idle owl
thanks. looking for something less low level, to point forum user to.
I'm not sure there is. FWIW, it's one of the clearest datasheets I've ever worked with. But I get it.
tidal kiln
no prettypins for pico 😦 its reusing the pi's graphic i think.
idle owl
no prettypins for pico 😦 its reusing the pi's graphic i think.
Yeah, it's not our board.
It would involve updating PP to support it.
tulip sleet
PP for this board would be great
jaunty juniper
this has it in advanced mode https://pico.pinout.xyz/
tulip sleet
yeah, the graphic in guide is their graphic but I moved the legend around a bit to make it narrower
jaunty juniper
tidal kiln
^^ neat!
idle owl
Nice find!
tulip sleet
or at the very least put a link in
tidal kiln
@jaunty juniper thanks. i can just give that url and say "click advanced".
jaunty juniper
pinout.xyz is a nice reference for the raspberry pi, but it has other boards like the pico, it doesn't show license information though. No mention here it seems https://github.com/gadgetoid/pico.pinout.xyz
the main repo has a Non Comercial Creative Common attached to the repo https://github.com/gadgetoid/Pinout.xyz
tidal kiln
@blissful pollen also thanks. that's a nice summary table.
manic glacierBOT
manic glacierBOT
kind meadow
Hey...question. I just flashed my MagTag with the CircuitPython bin file on Mac OS and it won't show up as a drive afterwards. Even after pressing reset. Does this not work on Mac or something?
jaunty juniper
hi, what bin did you flash and how ?
kind meadow
I installed the bin located here: https://circuitpython.org/board/adafruit_magtag_2.9_grayscale/
I used esptool to install it.
michaelmain@MacBook-Pro Downloads % ~/Library/Python/3.8/bin/esptool.py --port /dev/cu.usbmodem01 --after=no_reset write_flash 0x0 adafruit-circuitpython-adafruit_magtag_2.9_grayscale-en_US-7.3.2.bin
esptool.py v4.2.1
Serial port /dev/cu.usbmodem01
Connecting...
Detecting chip type... Unsupported detection protocol, switching and trying again...
Detecting chip type... ESP32-S2
Chip is ESP32-S2
Features: WiFi, No Embedded Flash, No Embedded PSRAM, ADC and temperature sensor calibration in BLK2 of efuse V1
Crystal is 40MHz
MAC: 7c:df:a1:1a:e3:12
Uploading stub...
Running stub...
Stub running...
Configuring flash size...
Flash will be erased from 0x00000000 to 0x00167fff...
Compressed 1471216 bytes to 957917...
Wrote 1471216 bytes (957917 compressed) at 0x00000000 in 16.9 seconds (effective 698.4 kbit/s)...
Hash of data verified.
Leaving...
Staying in bootloader.
ok, I'm almost sure it's not supposed to go at address 0, but I use the UF2 (or I flash the UF2 bootloader)
kind meadow
My MagTag sadly doesn't have UF2 on it. How do I go about installing the UF2 bootloader?
Or do we know what address we are supposed to start at?
jaunty juniper
there's a "factory reset" part in the guide, or you can grab the tinyuf2 from the bottom on the page, and flash the combined.bin from the zip (at address 0 this one)
kind meadow
Okay. Got uf2 bootloader flashed. I think we are in business now.
jaunty juniper
👍
kind meadow
Thank you very much! I was pulling my hair out XD.
manic glacierBOT
kind meadow
Question. Is there a way once you have your code the way you like it in circuit Python to make it permanent so to speak? Like basically make it so that it can’t mount the circuit Python storage device anymore. Like essentially could I compile and link my code and Circuit Python into a uf2 or something?
Also, is there a way to store data on the a RP2040 or an ESP32 without it being mounted on a PC’s file system?
manic glacierBOT
spiral elk
Question. Is there a way once you have your code the way you like it in circuit...
You can disable mounting of mass storage with code in boot.conf, however that's not security, as you can force the device to boot into safe mode.
https://learn.adafruit.com/customizing-usb-devices-in-circuitpython
proven garnet
https://docs.circuitpython.org/en/latest/shared-bindings/nvm/index.html
Check out @lone axle's excellent helper library for seamlessly working with nvm: https://github.com/FoamyGuy/Foamyguy_CircuitPython_nvm_helper
kind meadow
You can disable mounting of mass storage with code in boot.conf, however that's ...
Cool. I’m not doing it so much for security, just so when I build something for my little sister that she won’t accidentally mess it up. Granted I’m thinking of just leaving it open to make making adjustments easier.
manic glacierBOT
devout jolt
Is there anything that precludes me from making a C-based library that emits USB- and/or UART- MIDI at regular rates? Something similar in concept to WAV playing, but instead of samples, it's MIDI messages?
blissful pollen
Is there anything that precludes me from making a C-based library that emits USB...
Is there a reason to do it in C Vs python?
devout jolt
afaik, no access to reliable timing in Pythonland
(at least in my several years of trying to do it in CircuitPython)
slender iron
@devout jolt I think you could do it from C
blissful pollen
Figured may be that. I don’t see why it wouldn’t work in C but you would have to find a free timer. If it was interrupt driven should be reliable and avoid the GC delaying things
jaunty juniper
yeah, GC and all, though to be fair the USB poll rate is 8ms, so CP might be able to hold that
spiral elk
Is the inaccuracy so much that it would make an audible difference in the delivery of midi messages?
devout jolt
perhaps my question is: is there a timer subsystem I can hook into that is common_hal
devout jolt
Is the inaccuracy so much that it would make an audible difference in the delive...
oh yes, it's very noticeable. can be like 30ms on something as slow as 16th notes at 100bpm (150ms)
spiral elk
not being difficult, I just know very little about midi
devout jolt
MIDI is basically robot fingers for playing other synths, so think in terms of how fast your fingers can move. But humans are very sensitive to timing changes
spiral elk
would pio on rp2040 also be a potential option?
devout jolt
would pio on rp2040 also be a potential option?
I don't think so, unless there's a provision for triggering high-level objects like usb_midi and busio.UART which I think there is not
blissful pollen
Pulseio I think has code that basically sets a timer if you wanted one every 1/64bpm or something. Adafruit_bus_device shows calling passed python objects from C
devout jolt
Pulseio I think has code that basically sets a timer if you wanted one every 1/6...
ahh! this could be my ticket in for understanding that. thank you!
blissful pollen
(Pulseio may tie to a pin though trying to think of something that doesn’t but just on my phone right now).
manic glacierBOT
devout jolt
(Pulseio may tie to a pin though trying to think of something that doesn’t but j...
maybe I require a "bpm pin" ahaha
thank you all, this is good food for thought
idle owl
maybe I require a "bpm pin" ahaha
The rockin'est pin of all.
blissful pollen
When I’m home I’ll try to glance at the rp2040 data sheet (have it open from something else). It doesn’t sound too hard to build
proven garnet
New MAX1704x driver is ready for the bundle if anyone gets a chance to review the PR!
slender iron
@tulip sleet this should be an easy review: https://github.com/adafruit/circuitpython/pull/6744
tulip sleet
<@329766224093249548> this should be an easy review: <https://github.com/adafrui...
I was waiting for RetiredWizard to do the real testing but if you're satisifed I'll approve, np
slender iron
@tulip sleet it was something I saw too. I think I caused it when I added the timeout-ability
manic glacierBOT
idle owl
So.... I'm trying to copy a UF2 to a BOOT drive, and it's consistently copying the UF2 file to the BOOT drive.
blissful pollen
ahh! this could be my ticket in for understanding that. thank you!
RP2040 seems to have 4 alarms of which only 1 is being used for ticks (supervisor.port.c shows how it is used). Seems pretty simple to get going
idle owl
It's technically not the right file for this board, but the boards are nearly identical, so it should work. But either way, it's a UF2, so it should load even if it's not going to work, right?
devout jolt
It's technically not the right file for this board, but the boards are nearly id...
Which board? I think most UF2 bootloaders read the header and ignore the file if it’s the wrong platform. The OS may still think it copied the file tho
idle owl
Which board? I think most UF2 bootloaders read the header and ignore the file if...
S3 TFT Feather. I'm trying to load the S2 TFT Feather factory reset UF2.
Oh hmm ok.
jaunty juniper
yeah if the file appears "in the drive" that means it refused it
idle owl
Bugger.
jaunty juniper
oh S3 and S2 are not considered the same chip
idle owl
I think I remember being told to recompile it and then load it and pull off the UF2.
That makes sense.
But I don't think I have the code. So.....
idle owl
oh S3 and S2 are not considered the same chip
I thought I'd try it since the boards are so similar. I know the chips are different.
devout jolt
RP2040 seems to have 4 alarms of which only 1 is being used for ticks (`supervis...
Ah thanks!
idle owl
Ooh wait.
jaunty juniper
I thought I'd try it since the boards are so similar. I know the chips are diffe...
the key is that that tinyuf2 checks the chip variant
jaunty juniper
and I don't know all the differences, but I'm pretty sure that there is no way a binary for S2 runs on S3
idle owl
After doing the pinouts page, I'm inclined to agree.
jaunty juniper
pinout or not, they might share an Extensa core (although not even the same number), but I'm assuming that even basic setup for ram and clocks, etc. is probably different
idle owl
There is a section on that page that talks about the chip itself.
And it was definitely different than S2
(brb, need to check on food in the oven)
idle owl
Actually bbiab. Food is ready. :)
jaunty juniper
enjoy your meal!
idle owl
Thanks. It was meant to be a quick one.
hidden rain
I managed to create an extension of analogio by adding an interface AnalogFastin. It creates an objects which setup the ADC and DMA with a user supplied pin, buffer and sample_rate. I have been able to get ADC conversions in CPy at 2 microseconds.
It needs some polish and a good review.
idle owl
Hmm. No idea how to generate the factory-reset-and-bootloader.bin file.
Thanks for the help with sorting out the UF2!
manic glacierBOT
crimson ferry
A sampling of absolute latest from 8/11 and 8/10 don't mount CIRCUITPY after installing (QT Py ESP32-S2). The last one from 8/9 works fine. 7.3.2 works fine. Is this a known thing?
also noticed that if there is a pre-web-workflow mDNS circuitpython.local device on the network, it may answer first and preclude web workflow on the desired device
wraith crow
I just flashed adafruit-circuitpython-adafruit_qtpy_esp32s2-en_US-20220811-f9b9f55.uf2 and it mounted CIRCUITPY for me
@crimson ferry I don't know how mDNS works but I assume the old curictpython.local entry will fall off after some time? Either way you should be able to access the web workflow by IP address. Determining the address can be tricky if you don't have a display and can't connect to the REPL via USB serial. If you can the REPL should display the IP address in terminal header.
I'm not on a mDNS network so I'm stuck using IP address. I've taken to leaving a QT Py C3 online next to my router. I've got it's IP address bookmarked and when I bring it up, it lists any other CP boards I've plugged in.
crimson ferry
I just flashed adafruit-circuitpython-adafruit_qtpy_esp32s2-en_US-20220811-f9b9f...
Thanks, not sure what was happening, but it's happy on UF2 from a few days ago.
Yeah, I guess it would be expected that anything answering to circuitpython.local could grab the browser's attention. Just a note to be aware of, it becomes obvious what happened pretty quickly. This was a case where there were pre-web-workflow and post-web-workflow CP devices on the network simultaneously - the good news is that the web workflow GUI shows all of them. Luckily, in my local set-up, all CP devices have DHCP reservations so the IP addresses are predictable.
thorny jay
I remember that @idle owl wanted the yearly CircuitPython review of what users wanted for the future to be more frequent. So maybe the CircuitPython Day could be an opportunity to ask again to the community for input. Of course that is a late through. But if you don't do it on CircuitPython Day, then the next opportunity will be #CircuitPython2023 in December/January.
random junco
I remember that <@330227457296957440> wanted the yearly CircuitPython review of ...
That's a good idea. I know for the panel discussion, I reviewed those posts to create questions. You should email circuitpythonday(at)adafruit(dot)com.
thorny jay
That's a good idea. I know for the panel discussion, I reviewed those posts to ...
Then you missed my contribution to #CircuitPython2022 because I was lazy and forgot to do one, but I have a 2020 and a 2021 one. 🙂 I don't remember if a "decision" was taken, but that idea from Kattni was at least discussed "in the Weed", she would know the status or if a decision was taken. Since I guess she is working on the planning of CircuitPythonDay too, the above message would be sufficient. The problem I see is doing one in August... then that make the #CircuitPython2023 too close. Also how would you name that thing??? The #CircuitPythonDay2022WishList ? And it take a lot of time to read and analyze and compile all the contribution (if there are many).
random junco
Good questions, I liked the idea of a mid-year checkin
hidden rain
Good Morning! I have a new interface for the Pico, AnalogFastIn, which allows for a DMA read of the ADC. It might be useful for other platforms too! What is the normal process to request a little feedback. Is a formal Pull Request required first?
blissful pollen
Good Morning! I have a new interface for the Pico, AnalogFastIn, which allows fo...
You can do a PR in draft mode. For something with a bunch of code that can be a good place to get feedback, and at least I find it easier to review code in that format.
An issue can work too but depending on how much you want to show off can be a little more cumbersome.
hidden rain
ok... PR Draft..Thanks
solar whale
HAs anyone tried the current main on a FunHouse -- I am just getting into a boot loop. 7.3.0 is OK
tulip sleet
An issue would be welcome so we can track it, thanks. anecdata also has S2 issues, but then tried again and it worked ?!
solar whale
OK -- I'll open an issue
manic glacierBOT
proven garnet
Nothing cleans your computer files better than having a partition resize fail and having to reformat...
Eternal thanks to all the guide authors for the intro to dev guides, I got reinstalled and running in no time at all
manic glacierBOT
proven garnet
Maybe there was another way, but hey, I've got twice the space and it's all shiny clean
jaunty juniper
HAs anyone tried the current main on a FunHouse -- I am just getting into a boot...
I run the latest on FunHouse in my tests of the web workflow, though i don't necessarily pull every day, let me build the tip of main and see
manic glacierBOT
jaunty juniper
(or like download from S3, but I'm too lazy for that when I have the terminal opened to build)
solar whale
I run the latest on FunHouse in my tests of the web workflow, though i don't nec...
I have used a local build and latest from S3 -- I can't get to a REPL to load a .env file for web workflow
tulip sleet
f9b9f55 also works for me on a funhouse. I'm using the UF2
solar whale
hmm -- I even tried erasing the flash ...
tulip sleet
is anything connected to the funhouse?
solar whale
PIR
tulip sleet
take that off and see what happens; there was some pin initialization stuff that changed
solar whale
same without it
tulip sleet
which combined.bin or factory reset did you use, the guide one?
solar whale
none -- I used the UF2 or flashed from the build
tulip sleet
but you said you erased the flash
solar whale
yes, then just did a make BOARD=adafruit_funhouse flash /dev/tty.usbmode01
do I need to do something else?
tulip sleet
ok, now I'm getting a boot loop, after erasing the flash
solar whale
ah? sorry about that
tulip sleet
ah? sorry about that
np, reproducible is the best!
solar whale
I installed the uf2 boot loader by loading an arduino sketch (blink)
tulip sleet
Note that in the issue, to show it's multiple ways to get the bootloader
solar whale
OK
tulip sleet
it's weird that @crimson ferry had this problem, and then it mysteriously went away
manic glacierBOT
solar whale
it's weird that <@360465312518504448> had this problem, and then it mysteriously...
That is odd -- Thanks for reproducing it -- I originally thought it was because I had WipperSnapper installed on it -- that is why I erased the flash. Just to get a clean start. You "should "be able to load the .bin directly to the board if it is booted with the boot button, no?
That does not work for me either -- still loops.
solar whale
I'll see If I can find the culprit...may not be until later today...
Dog needs a walk now...
tulip sleet
you can leave it for us 🙂
solar whale
Thanks -- you know better what you are looking for. It's not a crisis for me -- I was just doing some board updating and thought I 'd try web workflow to a FunHouse. I have several other things to keep me busy...
manic glacierBOT
idle owl
I remember that <@330227457296957440> wanted the yearly CircuitPython review of ...
I think if Paul is up for it, we can work it into the panel discussion to put it out there that folks can do up a post/whatever-form-of-thing. I think it won't be as big of a deal as the one at the beginning of the year, but it would give us some more info to work with, and also we can aggregate everything on the blog like we did this year.
idle owl
@tulip sleet When you're around, let me know please. I noticed weirdness with the S3 AREF, and asked Limor about it, and she said they're all weird, but that you can help me with the S3. It's for the Analog In template. I think mostly what I need is if you could do an analog read and tell me the max value on an S3 on your end, that would be greatly appreciated. I need to find the average maximum for the S3 to fill in this template. I'm doing the guide for the S3 TFT, if you have one of those, checking would be appreciated.
No rush on this, I have plenty of other guide to do.
OK, redirect. Anyone with a potentiometer and an Adafruit ESP32-S3 board, ideally the TFT Feather, please let me know if you can test some analog read code for me. Thanks!
Really, you can do it with a wire, I'm mostly concerned about max raw analog value. Not so much with the rest of the values.
Ping or reply if you can help, thank you!
Woops. This is why you test your code kids.... Turns out CPU temp is not implemented on this board. Don't need to do the template in the guide then. But, already did the bulk of it. 🙄
crimson ferry
ESP32-S3 theoretical max appears to be 3.1V https://docs.espressif.com/projects/esp-idf/en/v4.4.2/esp32s3/api-reference/peripherals/adc.html as opposed to the ESP32-S2 of 2.5V
idle owl
ESP32-S3 theoretical max appears to be 3.1V <https://docs.espressif.com/projects...
Ooooooh nice find.
Thanks for digging into that! That's about what I found with mine.
crimson ferry
nice when the real world matches the specs 😉
idle owl
Right?!
jaunty juniper
Woops. This is why you test your code kids.... Turns out CPU temp is not impleme...
it will be some day 😉
idle owl
it will be some day 😉
Guide'll have to be updated. Can't go around added not implemented things to guides, now, can we. 😉
hidden rain
🤓 Hope it can be used across many platforms. Really need some feedback from Scott and Dan, Cheers!
idle owl
🤓 Hope it can be used across many platforms. Really need some feedback from Sc...
FYI, they're both out until Monday, so please be patient with them. Thank you!
hidden rain
NP. If there is anything I can or should do while waiting, please instruct. I figured it may take a while; good thigs do!
things
idle owl
NP. If there is anything I can or should do while waiting, please instruct. I fi...
Nope, you're all good! Nothing to do until you receive some feedback. 🙂
blissful pollen
NP. If there is anything I can or should do while waiting, please instruct. I fi...
I'll try to take a look at it this weekend to see if anything jumps out at me. I took a quick glance and the general format looks good. Maybe cause its draft but I noticed all the docstrings aren't there yet and (I miss this all the time) making sure errors are checking if calls like claim_dma fail. I'll do a proper review later though! Nice work
hidden rain
Thank You! I appreciate the review1
manic glacierBOT
manic glacierBOT
manic glacierBOT
warm stump
...might have to save this one for a future revision of circuitpython...and maybe a future revision of desktop and mobile python...all I wanted to do was (-12345.678)[-8:8]...
jaunty juniper
what is that supposed to do ?
warm stump
"get me bits 1 to 4 inclusive" from e.g. 255
but for more cases where the range is negative and decimal
jaunty juniper
bits of what ? the mantissa ?
there is not bit array in CP, I think python has libraries to read ints as bit arrays
warm stump
in general, both at the same time or something that can be used on both parts separately
then there was the version for pico-dvi which does [0:7] but on an int
jaunty juniper
so, a float is a mantissa and an exponent, what it means to read a "bit" might not be what you think it is
it's not a part before the coma and after the coma
warm stump
that's in 32 bit float right? this is something slightly different where bit1 is bit0, bit0 is the sign bit and bit-1 is the first decimal place
jaunty juniper
as far as I know, bits don't match decimal places unless it's BCD
warm stump
which this does by unfortunately having to convert it to a string and then reversing it with non-builtin rev, treating it as a normal int and reversing it back again
on hardware this is all done ideally in one step, my thing is preliminary and might have some shifting errors somewhere while trying to stay on the same line
jaunty juniper
like 10.0 and 100.0 should have the same mantissa, but 10 and 100 don't have the same bits
warm stump
Ah I see, from the python side I mean the bits from the perspective from a mathematical binary representation of the number,
there is a different one for getting a raw memory view of the bits stored in IEEE format. maybe the mathematical one will also help decode the IEEE one
the other slightly confusing thing is that the number line goes backwards. the furthest left is positive and the furthest right is negative...on arm this is the right way round maybe 😄
jaunty juniper
I'm not aware of the formula to convert a number after the comma into binary
is there even a single way to do that or multiple representations ?
warm stump
0b1.1, or in french 0b1,1 - removing the 0b has complicated things slightly
jaunty juniper
no but how to convert
warm stump
idk that's why I wrote that
jaunty juniper
something like this:
def get_bits(num, pos):
x = s - int(s)
bits = []
for i in range(pos):
# while x > 0:
x = x * 2
bits.append(int(x))
print(x, int(x))
x = x - int(x)
return bits
for example 0.2 converts to the infinite development: 0b0.0011001100110011...
and the limit of floating point numbers shows up in C python with the debug print:
0.4 0
0.8 0
1.6 1
1.2000000000000002 1
0.40000000000000036 0
0.8000000000000007 0
1.6000000000000014 1
tame creek
there's a hexadecimal format for binary floating point numbers. it might be supported in newer versions of Python (and is (mostly?) compatible with the one in C99)
warm stump
I saw the hex property while subclassing it just now to add subscript, could also add the def get_bits(num, pos): to it along with getting/setting a bit and masking a binary decimal
used all the time in assembly using very weird convoluted methods for e.g. setting GPIO pin direction and drive strength, or idk, pretty much anything that uses the class
some hardware acceleration for this would also be good
tame creek
@warm stump rev might be more easily implemented as string[::-1]
jaunty juniper
reversed() is a thing
tame creek
@warm stump you might also want to look at math.frexp if you want to portably and directly inspect the mantissa and exponent as stored by the machine
jaunty juniper
note that negative steps are not implemented in CP:
>>> "123456"[::-1]
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
NotImplementedError: only slices with step=1 (aka None) are supported
that seems unfortunate. what's the rationale for not implementing it? too much code?
jaunty juniper
don't ask me
but usually yeah limits are caused by things taking space and having alternatives
for the language aspects of CP, you gotta go back to MP issues and: https://github.com/micropython/micropython/wiki/Differences
it is implemented for lists, I don't see a reference to string slices steps in the issues, but it's just a quick search
jaunty juniper
ok now I get a hard crash when I ctrl-C/ctrl-D on the FunHouse
solar whale
control-c/d works in web workflow serial terminal
also OK for me via screen on my MAC
running Adafruit CircuitPython 8.0.0-alpha.1-135-gfa3e265bf
jaunty juniper
hmmm I'l look into that another time, might be linked to code.py, see if older versions do it and try a flash erase...
solar whale
Good luck! Try standing on one foot 😉
I just have the default "hello world" as code.py
manic glacierBOT
lone axle
https://github.com/FoamyGuy/Adafruit_CircuitPython_SI1145/runs/7820800078?check_suite_focus=true#step:5:9
It's a baby step, but exciting as a proof of concept 🎉
This actions task was executed on a "self hosted" runner that lives on my Raspberry Pi. The printed output is the size of RAM consumed by importing the library that was built in the existing build.yml tasks, measured by subtracting the values returned by gc.mem_alloc()
So we were able to run a test against a physical hardware device automatically as part of an actions workflow. Right now it's just measuring memory size, but eventually maybe it could execute other test cases and pass / fail based on assertions on the device itself.
warm stump
<@539881446756843560> you might also want to look at `math.frexp` if you want to...
nice, these also work. I did mine a certain way to be sure it would deep copy instead of using it like a pointer, but these solutions are also good for their contexts
manic glacierBOT
manic glacierBOT
manic glacierBOT
CircuitPython version
Adafruit CircuitPython 7.3.2-dirty on 2022-07-30; Waveshare ESP32-S2-Pico with ESP32S2
Code/REPL
import wifi
from ipaddress import ip_address
wifi.radio.connect(...) # insert your newclear launch codes here
while True:
wifi.radio.ping(ip_address("1.1.1.1"))
Behavior
Description
I was implementing th...
jaunty juniper
uh oh that sucks, any solution ?
❯ pip install -e .
ERROR: File "setup.py" or "setup.cfg" not found. Directory cannot be installed in editable mode: /Users/*****/Desktop/temp/Adafruit_CircuitPython_NeoPixel
(A "pyproject.toml" file was found, but editable mode currently requires a setuptools-based build.)
the answer is: update pip, but...
ValueError: invalid pyproject.toml config: `project.version`.
configuration error: `project.version` must be pep440
[end of output]
just symlink it to the site-packages?
jaunty juniper
we could switch from using 0.0.0-auto.0 to something pep440 ? Like 0.0.0+auto.0 as it turns out
(it's the placeholder for updating the version by the bundle tools)
or maybe we don't need the version in the file ? some libraries seem not to have it, this is the neopixel library
ah wait, they do, I wasn't looking at the right version
hidden rain
I have new files in shared-bindings/analogio ... AnalogFastIn[.h and .c]. However, its only implemented for the RaspberryPi Pico. It fails on SAMD21, etc. How do I exclude in shared-baindings?
blissful pollen
I have new files in shared-bindings/analogio ... AnalogFastIn[.h and .c]. Howeve...
Just a quick thought at the moment, but have you tried setting CIRCUITPYTHON_ANALOGFASTIN = 0 in py/circuitpython_mpconfig. If you look at something like CIRCUITPY_CAMERA in that file, you will want something simliar
hidden rain
@blissful pollen It seems the shared-bindings portion is failing because there is no filter for just Raspberry Pi Pico. Perhaps I don't understand the model here. Shared-Binding seems to imply across platforms while the module defined in shared-bindings appear to define the CPy interface and ports/<platform>/common-hal/<feature> the implementation algorithm. Is that NOT the model?
blissful pollen
<@228322543507865601> It seems the shared-bindings portion is failing because th...
I'll take a look at it to see I'm probably missing pointing you to some item as well.
Generally if a module exists in shared-bindngs, you are correct the port implementation is in ports/<platform>/common-hal/. But if there is no implemenation for a port both bindings and module have to be excluded from the build
blissful pollen
<@228322543507865601> It seems the shared-bindings portion is failing because th...
I added a comment to what I spotted right now that is causing the build to fail. The init.c file in shared-bindings is referring to AnalogFastIn so you will need #ifdefs around the lines there.
manic glacierBOT
lone axle
we could switch from using `0.0.0-auto.0` to something pep440 ? Like `0.0.0+auto...
+1 vote for switching to use a format that will follow pep440 if that will allow us to use pip install . Thats definitely a very helpful utility for testing changes to libraries, especially for use with Blinka on Linux.
jaunty juniper
I opened an issue in the build tools since it's relevant, and a good place to reach the people concerned (I think ?) https://github.com/adafruit/circuitpython-build-tools/issues/91
hidden rain
Can I use nested #ifdef to exclude code, even another #include?
blissful pollen
Can I use nested #ifdef to exclude code, even another #include?
Yeah I believe so. Like:
#inclue "a.h"
#ifdef B
#include "b.h"
#endif
#endif
Please look at the latest shared-bindings/AnalogFastIn.h. I have the nested ifdefs, but it still failed
blissful pollen
Will do. I'll try to pull it to my machine too and see if I can get it going. Just haven't had a chance. I think it must be close!
hidden rain
Thanks Dan. I guess I was rather self-centered on the purpose (for me). You are correct, that 'Bulk' makes more sense than 'Fast'. I am connected to some that may want to help continue the interface for other platforms, so yes the DMA is merely an implementation detail as is the sampling rate. The user should only be aware that it is 'bulk' at some rate. I will change the name and omit reference in example to DMA.
manic glacierBOT
lone axle
@proven garnet do you know if changing py-modules to packages inside of pyproject.toml is the only thing needed when a library would be changing from a single file to a folder? I'm familiar with that step in the setup.py way, and think I found the same thing in pyproject.toml but wanted to deck if there is anything else needed.
onyx hinge
Missing the meeting today, no notes.
hidden rain
Is there a way to listen to the meeting live?
tame creek
Is there a way to listen to the meeting live?
i think it's usually on one of the voice channels in this Discord?
hidden rain
at 2:00pm ET right?
random junco
Yup, join the CircuitPython voice channel at 2pm ET
proven garnet
<@135415127879974912> do you know if changing `py-modules` to `packages` inside ...
Yup! It's equivalent to the setup.py step (setuptools is still the backend so it's actually doing exactly the same thing behind the scenes)
lone axle
Thank you
idle owl
<@&356864093652516868> CircuitPython Weekly is under an hour away! Please add your notes to the notes doc if you plan to participate. Looking forward to seeing everyone there! https://docs.google.com/document/d/1AQqEMgemmn7GiPWfkrGbyr3ViaGPSAtYIwdclGa3XPE/edit?usp=sharing
solar whale
Lurking today 👋
proven garnet
@idle owl if you have time after the meeting (or honestly whenever, not critical), I want to check/set the secrets in the adabot repo which might help my PR pass CI.
idle owl
<@330227457296957440> if you have time after the meeting (or honestly whenever, ...
If today, it'll have to be later. I have another meeting after this one.
proven garnet
Sounds good!
blissful pollen
Oops thanks, had my mic unmuted from earlier this week and normally never have it like that
gilded cradle
Same. It's usually the first thing I check.
lone axle
😺 👋
turbid radish
Hi kitty!
ember iris
😺
lone axle
orchid basinBOT
lone axle
orchid basinBOT
lone axle
random junco
I'll definitely reach out, thank you!
slender iron
hee - mel, I've said it wrong for a long time
solar whale
@mental nexus -- Do you have a link to the FreeRTOS video?
mental nexus
hee - mel, I've said it wrong for a long time
Actually went to check a video but he never said his name!
slender iron
😄
candid sun
thanks for doing that @tulip sleet
lone axle
does the web workflow / code.circuitpython.org stuff live inside of the main circuitpython.org repo? Or is there a separate one for that?
lone axle
Thank you
silver tapir
proven garnet
Can we have those workflows be by approval only? Saves sending jobs that don't affect memory size (e.g., documentation fixes)
Adafruit IO does that, I saw
thorny jay
One other question is if you want member of the community to offer a runner by "dedicating" a "Pi" to this, or have only Adafruit to host such an infrastructure? If you want "contribution", they you have to deal with onboarding, saying what is connected to the runner, ...
(My idea was also to do that on container, on Pi4. While foamyboy was testing "native" on a Pi3.)
silver tapir
And yeah, it will work on arm, riscv, whatever...
ember iris
What about a different approach. Pull request workflow works as is right now, but then once a week a job runs over everything and raises an issue if a recent pull request expands the memory beyond a threshold. This would alleviate the need to have a job always ready to run hosted on a third party machine. But it's an entirely different idea so maybe this is more a detraction of an idea than it is helpful
silver tapir
But you'd have to wait a week to merge the pr's.
thorny jay
It might be "too late" to stop a pull request, but at least it gets detected at one point.
blissful pollen
I think it's important to at least keep in mind, future use beyond just memory usage. In a dream world a full regression, on hardware, test would be ideal on a PR run.
(but that's a big "dream" state)
thorny jay
The on device size vs mpy size difference is not very clear...
You think that "Static" analysis on the MPY file is possible and sufficient?
Also interesting is that you want to know and remember the previous size and the new size... so you have to compare the new artifact to "existing" file. Not sure how to do that.
lone axle
Also interesting is that you want to know and remember the previous size and the...
worst case the actions could clone / checkout the main branch and build the mpy file from it. Or possibly it could grab them from the bundle without needing to do the build "manually" inside the action.
ember iris
I personally am on board with looking into it, the notion of receiving a score that you work to improve rather than a pass/fail label seems like a good way to slowly, gently way to be mindful and build good practices
silver tapir
This two bossac tools for linux are binary artifacts.
https://github.com/adafruit/circuitpython/tree/main/ports/atmel-samd/tools
proven garnet
Thanks!
mental nexus
Thanks all!
gilded cradle
Thanks everyone
ember iris
Thanks all!
slender iron
@gilded cradle I'm getting hterm errors with http://code.circuitpython.org/#backend=web&host=cpy-951032.local
I don't see any network traffic to the cp device
gilded cradle
<@478616919721771019> I'm getting hterm errors with <http://code.circuitpython.o...
Ok thanks. I hadn't tested it once it went live. I'll take a look and see if I can reproduce the errors and create a fix.
slender iron
why?
gilded cradle
Maybe it's just web bluetooth that wasn't supported. I can add firefox support to the phase 3 part of the project.
I mostly hadn't been tested
slender iron
I've done the local html and js from ff
gilded cradle
Ok, it's likely something simple. For now can you test with chrome and I'll try and see if I can get FF working before oyu're out?
slender iron
yup, totally
the websocket fails on chrome so I'm taking a look from the device side now
will also add a link to code.cp.org to the welcome html
gilded cradle
Thanks
slender iron
np, to make it easy for myself 🙂
gilded cradle
It looks like it's forwarding from http to https and failing
Can we disable that? I don't think it's the Javascript doing the forwarding.
gilded cradle
If I click on the http link you have above, it forwards it to https://code.circuitpython.org/#backend=web&host=cpy-951032.local. You told me it had to be http because the device didn't have any SSL certificates.
slender iron
I was able to open it in http in chrome
and in firefox
try copy and pasting the link
gilded cradle
Hmm, that isn't working for me. Then it fails an error due to the mixed content.
slender iron
in what browser?
gilded cradle
chrome
slender iron
do you have an extension that forces https?
I have it open and chrome is warning it isn't secure
gilded cradle
Let me check
slender iron
gilded cradle
I turned off all extensions, restarted, and it's still doing it.
slender iron
I'm on linux
gilded cradle
I'm on mac and running chrome 104.0.5112.79.
slender iron
ah, it just did it for me
gilded cradle
Here' the redirect.
Could be a browser thing
Looks like it's seeing "strict-origin-when-cross-origin" and that prompts it to redirect.
slender iron
ok, I cleared my cache and it didn't redirect
manic glacierBOT
gilded cradle
Ok, I'll try the same.
slender iron
may be because we used to set it for https
gilded cradle
Ok, that worked for me too.
Probably adding a popup with some fix ideas might be worth it.
Automatic redirects could result in an annoying loop
slender iron
or disable web under https
gilded cradle
Ah, if we could do that just for code.circuitpython.org, that would be ideal.
And then leave regular circuitpython.org as https
slender iron
ya, I think that's what we're going
I'm not sure cp.org applies to code.circuitpython.org
gilded cradle
Probably not, but possibly.
gilded cradle
We may need to do that and then disable https on code
Or I'm thinking maybe there was some issue and we turned it off for a reason
slender iron
ah, my websocket is failing because I don't have the auth credentials
I turned https off for code.
gilded cradle
Oh, ok. I may need to attempt to listDir first
gilded cradle
Yeah. it's strange because I'm pretty sure I got the websocket going before I even tried to tackle the auth stuff.
slender iron
its possible your browser cached it
gilded cradle
Yeah
I'll give it a work through using code tomorrow morning and see if I can get it functioning.
gilded cradle
are you done for the day?
No, I was in the middle of preparing for my livestream and lately I tend to work better in the morning.
gilded cradle
Ok, we can focus on this more tomorrow.
idle owl
@tulip sleet This I2C terminology change... so much so that we should update the I2C template, and send out a new blog post (this is the old one: https://blog.adafruit.com/2021/06/24/controller-and-peripheral/) ? I was about to start updating the template, but realised I should ask first.
tulip sleet
<@329766224093249548> This I2C terminology change... so much so that we should u...
Yes and yes
but you can point out it's CircuitPython 8
idle owl
Yes and yes
Well, this isn't about using it in code, this is about terminology in text and Learn guides.
So, this would simply be info, not so much tied to CP specifically.
tulip sleet
if guides don't mention I2CPeripheral, then yes,
idle owl
My template certainly doesn't. I'm not sure I was even aware of it.
tulip sleet
https://www.nxp.com/docs/en/user-guide/UM10204.pdf is the revised spec
idle owl
Ok. Wanted to make sure it extended to general terminology and wasn't specific to CircuitPython code
Perfect thank you.
tulip sleet
second page of that pdf describes the change
proven garnet
@tulip sleet, @slender iron, @jaunty juniper I used sed to fix the version strings, I can push all of those to main if you think 0.0.0+auto.0 is good (and PEP 440 compliant, just double checking)
proven garnet
Yeah, I had sed change the workflow files too. The workflow files themselves use sed to do this at release time, so it should have gotten everything
jaunty juniper
and build tools and cookiecutter ?
proven garnet
Ah right, it was already 0.0.0-auto.0 when I did the pyproject.toml stuff, so I forgot I'd need to touch up a few other places
I can PR those
pure swallow
I think there's a bug in dotenv. When you have a # in a quoted value (KEY='a#bc') it stops working after that line.
manic glacierBOT
proven garnet
Alright, submitted the fix for the build-tools, and wrapped the changes for the cookiecutter into the existing PR I had out for it.
I had stepped away from the meeting today, thanks for the hug report on the ghost user. It was a fun little "Easter Egg" (ish) to find: https://github.com/ghost
manic glacierBOT
manic glacierBOT
kind meadow
Does anyone know if there is a way to make the Magtag stay connected to Wifi? I am using my phone’s hot spot for Wifi and it disconnects from the Wifi between fetch calls.
random junco
Does anyone know if there is a way to make the Magtag stay connected to Wifi? I...
You may have better luck getting an answer in #help-with-circuitpython - this channel is related to development. Your question reminded me of this answer though, try using compatiblity mode: #help-with-wippersnapper-and-adafruitio message
manic glacierBOT
Adds support for the Waveshare ESP32-S2-Pico Board with LCD ST7735R
USB PID from Espressif: [Espressif USB customer-allocated PID repository](https://github.com/espressif/usb-pids/blob/b39d288d4821bedfb589b7ec3cc8a3e2779e6237/alloca...
shut musk
People, is there a special design pattern here? I'm looking at usb_hid module. It consists of two parts: shared-module/usb_hid and shared-bindings/usb_hid. shared-module/usb_hid/Device.c defines some functions with names starting with "common_hal_usb_hid_device_" which are used in shared-bindings/usb_hid/Device.c. However, shared-module/usb_hid/Device.h does not contain those functions. Instead, they are declared in shared-bindings/usb_hid/Device.h. Huh?
tulip sleet
People, is there a special design pattern here? I'm looking at usb_hid module. I...
There are potentially multiple implementations of the common_hal_ routines. They are declared in shared-bindings/usb_hid/Device.h because the API is the same across all implementations (all impls must provide the same API).
In most cases they would be in port/<portname>/common-hal. However, when the API can implemented in a non-port-specific way, which is true for USB stuff (because we use TinyUSB), then the common_hal_ implementations are in shared_module.
shut musk
There are potentially multiple implementations of the `common_hal_` routines. Th...
Ok, I see. Still a bit confusing, because usually headers define functions that are declared in accompanying .c file. Would it make sense to add "extern" modifier to those declarations? It does not change its functionality, but will hint that those functions are "external" to that module
tulip sleet
it doesn't hurt to add extern, I believe. We had to add extern for non-function declarations at some point. The idea is that there is a single place where the declarations are written, so that the identical list of declarations does not need to be duplicated in multiple places.
yes, it's unusual, and I was confused the first term I saw it, but now I see the reasoning.
extern for functions vs non-functions is a bit different, I think.
shut musk
`extern` for functions vs non-functions is a bit different, I think.
Yes, forward declarations for functions have extern implied by default
tulip sleet
right, and for non-functions that used to work (I think), but at some point it didn't
there are a few shared-bindings .h file where they are declared extern
we haven't gone back to make them all consistent in that way
shut musk
oh, right. I see functions declared extern in some header files in shared-bindings/
tame creek
objects (not functions) declared at file scope without initializers or extern can end up in a "common block", which has some annoying implementation-defined behaviors
shut musk
objects (not functions) declared at file scope without initializers or `extern` ...
Right.. which is not the case here
tame creek
i think using extern as a convention to flag function declarations that might be in a differently-named .c file isn't necessarily obvious, and the practice should be documented if you choose to use it as such
shut musk
i think using `extern` as a convention to flag function declarations that might ...
What is the best place for such documentation?
tame creek
What is the best place for such documentation?
i think it depends on the project. repeating the documentation of the convention in each header file is one option, but in a large project, there might be a style guide to centralize such conventions
tulip sleet
it could be an (Identical) comment in each shared-binding .h, or it could be in https://docs.circuitpython.org/en/latest/docs/common_hal.html
kind meadow
You may have better luck getting an answer in <#537365702651150357> - this chann...
Thank you! My bad for posting in the wrong place.
gilded cradle
@slender iron when you're online, where did you get the hterm_all.js file? This is a file I've not touched at all since you had made the original code editor. It doesn't seem to be in the Code Mirror docs and I'm wondering if there's an updated way of loading the terminal that doesn't require such a huge file.
slender iron
@gilded cradle this is the original source: https://chromium.googlesource.com/apps/libapps/+/master/hterm/
if you want to switch to a different terminal that's fine
there is also https://github.com/xtermjs/xterm.js
gilded cradle
<@478616919721771019> this is the original source: https://chromium.googlesource...
Thanks
I'll see if I can get hterm working first since I have the colors already customized, but if not I may look at switching.
slender iron
👍
@gilded cradle ok, I'm at my desk. are you working on getting code.cp.org triggering the basic auth before doing the websocket?
gilded cradle
<@478616919721771019> ok, I'm at my desk. are you working on getting code.cp.org...
I worked on that a bit (just on my system), but the hterm errors were getting in the way, so I turned back to that.
slender iron
kk
gilded cradle
@slender iron I think switching over to xterm would be a good move so we can use firefox too. However, since that may take some time, I'm going to just use chrome for the moment and concentrate on the web socket issue. I need to head over to my office though, so I'll probably be offline for like 20 minutes or so.
slender iron
ok, sounds good
chrome-only is fine
better to ensure it works with the device for now
gilded cradle
yeah
proven garnet
@slender iron I think I understood your review, resubmitted
slender iron
👍 just approved
proven garnet
Thanks!
slender iron
good to decouple changes like that
proven garnet
good to decouple changes like that
Agree, good idea!
@idle owl I want to run sed on the bundle libraries to change 0.0.0-auto.0 to 0.0.0+auto.0 and push everything to main
Just wanted to run that by you
It's probably more reliable than an adabot patch
idle owl
<@330227457296957440> I want to run `sed` on the bundle libraries to change `0.0...
Sure. Cookiecutter as well?
proven garnet
Yup!
I'll run it tonight when I get back to my dev laptop
The cookiecutter update got wrapped into my outstanding pr
idle owl
Ok, perfect.
@proven garnet Will this require a new release on everything?
Or is it internal only
proven garnet
I think internal only. Editable installs only make sense if you clone the repo, so a release isn't needed to enact it.
idle owl
I think internal only. Editable installs only make sense if you clone the repo,...
Alright, think on that. So I know whether to tell Eva to do another sweep.
idle owl
@candid sun Hey, do you happen to have a RPi Pico (original), a MAX98357A amp, and a little speaker? And a few minutes to test something for me?
candid sun
@idle owl let me do some part digging. i allegedly have the amp
idle owl
<@330227457296957440> let me do some part digging. i allegedly have the amp
Ooh ok. Thank you!
candid sun
Ooh ok. Thank you!
okay yes
idle owl
okay yes
Sweet. Can you wire it up as shown on this page, and then test the different examples? All files needed should be in the bundle downloads.
I had to update the template to use a non-discontinued part, and I need to make sure it actually works. :/
candid sun
gotcha! yeah i will get that tested, just got to solder some headers
idle owl
Frankly, as long as one example works, they all should. But the bundle makes it easy to test multiple things at least.
idle owl
gotcha! yeah i will get that tested, just got to solder some headers
Thank you so much!
gilded cradle
@slender iron I think the auth issue is fixed. I just pushed the change, so it should be updated now.
Though it's back to its behavior of forwarding me to https again
slender iron
ok, lemme try it
Access to fetch at 'http://cpy-f57ce8.local/fs/' from origin 'http://code.circuitpython.org' has been blocked by CORS policy: The request client is not a secure context and the resource is in more-private address space
private.
candid sun
Thank you so much!
so...i'm not getting anything out of my amp/speaker combo. tried cp 7and 8 with an original pico and then an rp2040 feather. not sure where the issue is
idle owl
so...i'm not getting anything out of my amp/speaker combo. tried cp 7and 8 with ...
Hmmmmm
gilded cradle
> Access to fetch at 'http://cpy-f57ce8.local/fs/' from origin 'http://code.circ...
sounds like an https trying to retrieve http error
slender iron
the tool bar says it isn't secure
idle owl
it might be the speaker i wired up. i only have the oval one
Oh hmm. Maybe? Ok. I don't want you to spend all day on this. I'll test it on Feather RP2040 (good call! I hadn't thought of that) and see what I get.
Thank you for trying! I really appreciate it.
gilded cradle
the tool bar says it isn't secure
I'm going to do some research on this.
candid sun
Thank you for trying! I really appreciate it.
no problem! i can see if i have another speaker around too if you also experience issues
gilded cradle
@slender iron can you check the flag chrome://flags/#block-insecure-private-network-requests
idle owl
no problem! i can see if i have another speaker around too if you also experienc...
I'll keep you posted!
slender iron
<@252717193496756235> can you check the flag chrome://flags/#block-insecure-priv...
set to default
gilded cradle
Ok, I think that's blocking it. It's one way to get around the problem, but I'll see if I can find a better way.
gilded cradle
oof
slender iron
I think reverse embedding is the way to go
gilded cradle
One possible solution recommended there is WebTransport which I hadn't heard of before.
slender iron
webtransport still does a bunch of security stuff I think
jaunty juniper
ugh that's why we can't have nice things !
slender iron
which will take lots of code on the device side
MP just had to do the reverse embedding technique
gilded cradle
Ok, basically the way we're trying to do it now is a dead end, so we have to figure out the next best thing
slender iron
ya, I think so
jaunty juniper
I have been testing running my web UI (and a USB workflow) in an electron app
gilded cradle
So does the reverse embedding technique pull in code externally?
slender iron
unless a cors preflight can get us around it
ya, it does
but that code is pulled over https
gilded cradle
Hmm
gilded cradle
unless a cors preflight can get us around it
I'll have to look into that as well. I think doing a little research at this point is in order to explore potential options.
slender iron
I think the preflight thing requires an https fetch
gilded cradle
Hmm, ok
slender iron
MP just did the host in reverse thing
gilded cradle
Yeah, it's an option, but would still require quite a rewrite.
slender iron
yup
gilded cradle
Assuming we can get a self signed SSL cert somehow for the device, how much would need to be added to support https on the device?
slender iron
I have no idea
gilded cradle
Ok, no worries.
slender iron
espruino loads the https site in an iframe and then calls out somehow
gilded cradle
Hmm, ok.
I'll think about that. Even a hidden iframe to mask the calls might work.
Perhaps some kind of proxy might be another idea, but also might be a big security hole.
gilded cradle
I'm not sure how it will work with websockets, but it may be enough to get around the issue.
slender iron
ya, that's a bummer
gilded cradle
Well, I'll try a bit of experimentation. As a base I'll set localhost to https and load the local circuitpython web server in a hidden iframe. That should be close enough to live to simulate the issues.
gilded cradle
I'll think about that too, but that approach would likely have other issues we aren't seeing.
slender iron
ya, definitely aren't great options
gilded cradle
I think I might at least be able to get the file ops working with the iframe. I'm not as sure with the web sockets.
Unless maybe I can proxy through the device's serial terminal app.
slender iron
post message looks pretty straight forward
idle owl
@candid sun Uff. I'm not getting anything either...........
Speaker clicks as though it's getting a signal.
But no tone or wav.
gilded cradle
post message looks pretty straight forward
yeah and looking over the mp code, it may not be that hard to set up secure websockets in CP. That's something I may look at doing while you're gone if I can't find a quick solution.
manic glacierBOT
slender iron
secure websockets requires ssl afaik
gilded cradle
secure websockets requires ssl afaik
Yeah, I'm not sure at this point.
candid sun
<@347203168121126912> Uff. I'm not getting anything either...........
oh noooo. how about with an m4?
slender iron
idle owl
So maybe those pins on the Pico aren't supported anymore?
idle owl
oh noooo. how about with an m4?
I did A0, A1, A2 on the Feather and it immediately worked (once I sorted out what needed to go where). I need to find a Pico is what needs to happen here.
Time to dig!
gilded cradle
I think our two options are: 1. have a skeleton served by CP that loads js from ...
I like the second option better as it's closer to what we were going for.
slender iron
kk
candid sun
I did A0, A1, A2 on the Feather and it immediately worked (once I sorted out wha...
ah yeah i did use D5-D9 so that explains it that it needs to be analog
idle owl
I didn't think it did though...
@candid sun Pico works for me. As shown in the guide. So it must be your speaker situation. 🙂
Super crunchy sound, but I guess to be expected with small speakers.
lol I think this is the Pico I tested this with last time. Because it asked to replace the WAV and the MP3 files. 😄
candid sun
<@347203168121126912> Pico works for me. As shown in the guide. So it must be yo...
ah okay, glad it worked!
idle owl
For sure! I was about to be baffled. Still baffled why the same pins didn't work on Feather, but you know what? Not my problem right now. This is a future-me problem.
proven garnet
Alright, think on that. So I know whether to tell Eva to do another sweep.
Alright, let's not do a release, after thinking about it
idle owl
Alright, let's not do a release, after thinking about it
Sounds good! Let me know when you've done your thing.
proven garnet
@idle owl @jaunty juniper Alright, everything is changed over to 0.0.0+auto.0
idle owl
@proven garnet Please send me an email to deal with the release tomorrow. Thank you!
proven garnet
<@135415127879974912> Please send me an email to deal with the release tomorrow....
Oh, I said I don't think we need one, but let me know if you still want to do one anyway, I might have missed that.
If we do I can email you
manic glacierBOT
slender iron
oops
proven garnet
If anyone is up for a quick PR review, this is for the version strings, but also for fixing the neopixel library (and anything that depends on it) that should have gotten the followup patch during the pyproject.toml migration: https://github.com/adafruit/Adafruit_CircuitPython_NeoPixel/pull/135
It's keeping a handful of libraries from passing CI (and from downloading correctly, in general)
Might be keeping itself from installing correctly, hence the concern 😅
Ah, I saw the issue you responded to @jaunty juniper
jaunty juniper
If anyone is up for a quick PR review, this is for the version strings, but also...
hey I looked at a few other libraries recently modified and they don't have the release.yml change from - to +
proven garnet
Can you tag me in them?
proven garnet
ticks looks like it has it, unless you mean changes like those
proven garnet
huh, I'll double check
I think I missed some on the sed command, thanks for catch! I'll push the workflow file changes.
I think I forgot to add *.yml
jaunty juniper
that would do it !
proven garnet
Pushing now, thank you so much for catching!
manic glacierBOT
The Folders "Choose Files" button is active even when CIRCUITPY is mounted. The upload doesn't work but the interface runs through the upload process just the same.
The progress bar doesn't do anything on a folder with a single file in it. That's not a problem, just an observation.
In the Developer Console I'm getting a number of timeout errors
It appears that if a folde...
small cypress
Hey, does anyone know what the status of possible ULP support for the ESP32-Sx chips is? I've been able to find some messages from a couple of years ago about it, but nothing more recent.
jaunty juniper
Hey, does anyone know what the status of possible ULP support for the ESP32-Sx c...
I thought it was used for deep sleep
small cypress
I thought the purpose of ULP was to actually let you run code during deep sleep?
I guess maybe CP uses ULP to do its deep sleep, though Espressif has sample code showing how you can have it read an external sensor while in it, which obviously would be very useful, especially if it could do I2C: https://github.com/espressif/esp-idf/tree/5b11895700/examples/system/ulp_riscv/ds18b20_onewire
blissful pollen
Just taking a quick look at it and while I can see some great uses for it, I'm not sure how you could have python libraries to read sensors via i2c without the sensor driver being written in C
manic glacierBOT
I turned off auto-reload and performed a folder upload. While the upload was underway I ran some python scripts via the USB serial REPL and the upload failed with an ERR_CONNECTION_RESET.
manic glacierBOT
manic glacierBOT
small cypress
Just taking a quick look at it and while I can see some great uses for it, I'm n...
Yeah, though I guess I'm wondering if you could have most of your project be in CP, and just this little bit in C
blissful pollen
Yeah, though I guess I'm wondering if you could have most of your project be in ...
The problem I think is the C code for ULP has to be linked to the main code. Which is possible with a custom build, or maybe there is a way to not have to link it. and then CP would need a new or maybe addition to the sleep module to set this on/access the RTC memory.
small cypress
Indeed, I wasn't imagining it'd be super easy to implement, especially as I'd imagine it'd require changes in terms of the usual CP workflow to compile + put the code on there, though it really would be super useful/cool, even if you couldn't pass sensor data etc. between the ULP and CP world.
spiral elk
Well, I built an image for the Ai-Thinker ESP32-CAM but have to wait until I get home to actually flash it to hardware
spiral elk
Have a working image on the ESP32-CAM, lot of pin issues to work out though but serial REPL and web workflow is up and running
tidal kiln
is there a current Learn page that shows installing CP on ESP32? like with esptool etc?
tulip sleet
@slender iron @onyx hinge fyi ja is now the largest build by a lot (not ru), in case you are testing whether something fits
tulip sleet
is there a current Learn page that shows installing CP on ESP32? like with espto...
not yet, we are not even listing the builds yet on circuitpython.org, to avoid excessive support load (but really mainly because we haven't done an alpha.2 or beta.0 yet)
but lack of instructions adds a load to you in other ways
the instructions are really the same for loading a .bin on ESP32-S2
etc
tidal kiln
is there page for that? this is for a new guide. would rather link to something than re-write. looking around....
tulip sleet
is there page for that? this is for a new guide. would rather link to something ...
only the .bin is different
tidal kiln
thanks. yah, that's the best one i could find too.
slender iron
@tulip sleet what changed to make it more?
slender iron
I thought the purpose of ULP was to actually let you run code during deep sleep?
It isn't used afaik. We do use the power domain for some sleep things.
tulip sleet
<@329766224093249548> what changed to make it more?
it didn't become more, ru became less
slender iron
@gilded cradle I'm at my desk. How's it going?
gilded cradle
<@478616919721771019> I'm at my desk. How's it going?
Good. I was just going through github notifications and then going to test the iframe thing some more.
tulip sleet
<@329766224093249548> what changed to make it more?
or more exactly, ru was the largest on SAMD21 builds, due to font. I turned off terminalio on ru builds on SAMD21, which made that no longer be largest build. On non-SAMD21 builds, I think ja has been the largest build for a while
tidal kiln
@tulip sleet yah, wipper just says follow webpage instructions
tulip sleet
which webpage?!
tidal kiln
AIO
you select board, then "follow instrucitons"
could login and run thru it i guess. see what it says.
tulip sleet
oh, wippersnapper is a UF2, ok, never mind
I looked at the io.adafruit.com pages
somehow I thought it was .bin
tulip sleet
what r you using, firefox?
tidal kiln
yah
slender iron
or more exactly, ru was the largest on SAMD21 builds, due to font. I turned off ...
why did you turn it off? don't we have decent font coverage for it?
tulip sleet
the cyrillic font is too big; this is only on SAMD21. Same reason we turned the font off for ja, ko, etc., on SAMD21
I just mentioned this now because I was looking at flash consumption on matrxiportal and wanted to test with the largest build
tulip sleet
I did that a month ago
yes, it is, I feel it was necessary
or maybe I didn't turn it off, lemme check, maybe I just reported it. I was trying to figure out why the ru builds were so big
slender iron
can we turn it back on with the -Os change?
tidal kiln
@slender iron what's your suggestion for getting/loading the esp32 firmware (for web workflow)? i just grabbed from s3 and used esptool.
slender iron
I don't have a good answer. I think the future is a chrome browser and web serial but I don't know the current state
I use make to do it
tidal kiln
thanks. i'll point to the web serial info. but i guess also quickly show s3/esptool.
tulip sleet
@slender iron https://github.com/adafruit/circuitpython/pull/6562#pullrequestreview-1044205087 Really, I don't think this is a big deal. It's only an omission for SAMD21 display support. I do see we might try turning onewireio back on, on SAMD21. Right now it is off for all non-Express builds
tidal kiln
@gilded cradle is this the best page to point people to for general coverage of using webserial for firmware loading?
https://learn.adafruit.com/adafruit-magtag/web-serial-esptool
tulip sleet
we have had other regressions before. the ru font is very large
I would turn onewireio or even pulseio back on first if we had the space
gilded cradle
<@478616919721771019> is this the best page to point people to for general cover...
Yeah, that's fine.
slender iron
I would turn onewireio or even pulseio back on first if we had the space
ya I agree with this
tulip sleet
SAMD21 display support is a niche, like gizmo
or even turn longint on, if possible, since that would fix some library longint blocks
stuck elbow
SAMD21 display support is a niche, like gizmo
and hallowing?
tulip sleet
text russian vs images
hallowing is fine, just russian won't have terminal on it
same is already true for ja
stuck elbow
ah, I see what you mean
tulip sleet
if we had space on samd21, what would you add back for your boards?
this is all theoretical, I'm not sure there's enough new space to add anything back
longint??
stuck elbow
well, for my boards I always have my own custom list anyways, depending on what the board is using
tulip sleet
but you do omit things that are just too big. onewireio is not important to you, but other things might be nice if you could have them
stuck elbow
I usually remove everything that is not strictly necessary on a given board
tulip sleet
limor i think is keen on pulseio or onewireio being back
stuck elbow
we can also leave that space empty
if you do that, it will be invisibly eaten away
tulip sleet
i have to go afk for a while, but interesting convo
slender iron
I didn't realize pulseio was disabled
tulip sleet
yeah, it's been a while, she would like to have it back
onewireio you can substitute i2c sensors, but pulseio has no substitute
slender iron
pulseio is off for non-full builds and some other select samd21 builds
tidal kiln
@slender iron looks like thonny does not handle the additional escape sequences in the banner text
slender iron
Next beta will
tidal kiln
ok. yah. i've managed to get it connected a few times. but only with multiple tries. and not sure how/why it finally was ok.
slender iron
@tulip sleet is https://github.com/adafruit/circuitpython/issues/6756 on your todo list?
tulip sleet
<@329766224093249548> is <https://github.com/adafruit/circuitpython/issues/6756>...
not specifically; I'll assign myself. I do want to do a bisect on the failing builds.
slender iron
k, I'm running out of time to pick anything else up
would you mind doing a beta.0 release tomorrow too?
tulip sleet
sure, for CPy day
so, the S3 sleep problems are ESP-IDF related?
@slender iron also did you see the report of a bug with a # inside a .env value? I don't see an issue for it.
it was in discord
also 6609 looks like it might be simple
slender iron
so, the S3 sleep problems are ESP-IDF related?
ya, I think we need to bug espressif about it
slender iron
<@252717193496756235> also did you see the report of a bug with a `#` inside a ....
jeff mentioned it here: https://github.com/adafruit/circuitpython/issues/6661#issuecomment-1216868316
I'm not sure what to do about 6609
because we'd need to implement insert by deleting characters and re-sending them
tulip sleet
if there were a better terminal impl in the browser, would it help?
tulip sleet
maybe arrow keys should just do nothing (but it would be frustrating)
slender iron
I'm not sure we can fully prevent moving the cursor
tulip sleet
suppose you didn't send the line until you got an EOL
so the type-in box would be browser-local only. kind of like Arduino serial window
then the arrows would work, as handled by the browser
slender iron
ya, that could work too
jaunty juniper
we might as well do that, unless it handles tab completion
slender iron
I don't feel like the best person to refine that html/js
tulip sleet
i can comment in the issue
small cypress
It isn't used afaik. We do use the power domain for some sleep things.
Yeah, it's that I found some stuff from 2020 on here and Github but I'm not sure what, if anything, came of it:
slender iron
the ULP is also tricky because it isn't xtensa
I was hoping to treat it more like PIO
tidal kiln
what's that one pin quirk about using pico uf2 on picow?
jaunty juniper
what's that one pin quirk about using pico uf2 on picow?
board.LED is actually connected to the wifi module
it's one of the communication lines with the module, the real LED being controlled by the module it can't be used in Circuitpython
tidal kiln
thanks. so basically. just ignore that pin?
jaunty juniper
yeah, don't use it, treat the board as having no LED
tidal kiln
@jaunty juniper thanks!
idle owl
How do I redo this using with? ```py
import audiocore
import board
import audiobusio
audio = audiobusio.I2SOut(board.A0, board.A1, board.A2)
wave_file = open("StreetChicken.wav", "rb")
wav = audiocore.WaveFile(wave_file)
print("Playing wav file!")
audio.play(wav)
while audio.playing:
pass
print("Done!")```
I've not done it myself without something to work from before. I thought there was usually more code involved than is here. Pylint is complaining.
About 6 files, but I think if I get one, I'll be able to do the others.
jaunty juniper
for the open ? We want to keep the file open forever, so that means putting the whole rest of the code in the with clause
idle owl
Pylint says, "Consider using 'with' for resource-allocating operation".
slender iron
we've been meaning to have WaveFile take in the filename
jaunty juniper
yeah like OnDiskBitmap
tidal kiln
# pylint: disable=consider-using-with
tell pylint you considered, and thank you very much 🙂
idle owl
I really dislike adding that to short code in templates. 😕
But yeah, agreed.
Short code means I should explain why I'm disabling.
Long code? There's so much going on, I can gloss.
stuck elbow
I find myself writing more pylint directives in my code than the actual code
idle owl
Um.... Ok this one has me baffled. ```py
decoder = audiomp3.MP3Decoder(open("slow.mp3", "rb"))
audio.play(decoder)
while audio.playing:
pass
print("Done playing!")```
Not sure how to do with here.
stuck elbow
with open("slow.mp3", "rb") as f:
decoder = audiomp3.MP3Decoder(f)
audio.play(decoder)
while audio.playing:
pass
print("Done playing!")
Is that what you call "linterate programming"?
slender iron
@tulip sleet looking at 6756 already? otherwise I'll take a quick peek. I could also start beta.0 release notes
tulip sleet
i need your specialized skills now 🙂
small cypress
the ULP is also tricky because it isn't xtensa
As in, it's a different kind of chip than the main chip so it's harder to program/compile to? From a bit of cursory Googling it just seemed to be C though no idea if the overall toolchain (?) is a horrifying pain.
slender iron
As in, it's a different kind of chip than the main chip so it's harder to progra...
ya, you can't just run code for the main cores on it. It has two modes, one custom isa and riscv iirc.
small cypress
ya, you can't just run code for the main cores on it. It has two modes, one cust...
Yeah, the custom one looked like a horrifying pain to deal with, but I was figuring the newer ULP might not be as awful
Guess that assumption was a bit puts on shades RISCy on my part (I am so sorry)
slender iron
ya, I'd target the risc-v part if anything
jaunty juniper
would that make it easier on the C3 ?
slender iron
@jaunty juniper does the C3 have a ULP?
small cypress
I figured they meant that the C3 is RISC-V itself?
small cypress
I checked, it doesn't have one: https://docs.espressif.com/projects/esp-idf/en/latest/esp32/hw-reference/chip-series-comparison.html
small cypress
though there is MicroPython for the C3 itself