numbers.do_weird = False I think
#circuitpython-dev
devout jolt
idle owl
`numbers.do_weird = False` I think
🤦🏻♀️ I can't believe I forgot.
Also, I don't see a refresh time in the ACeP7In lib, but I also don't know how to read... bytearrays I think they are?
@slender iron I see you moved epaperdisplay to its own thing 😄
In shared-module
Does that change anything in the board files? Or can I incorporate that as-is?
idle owl
@slender iron I went back to main, dumped the board files in from the PR, and was met with a boatload of build errors. I managed to fix all of them and build successfully! 🤯 Now to see if it actually works. 😄
It's using the new module now though, I think.
lone axle
In the docs for terminalio.Terminal it makes reference to a cursor and moving that cursor's position.
Is there any more information about that anywhere? Is that meant to be a cursor that can be visible on the display as a reverse colored character or something similar to indicate a position or highlighted text?
idle owl
Woof. Why does the REPL tell me I have CP 6.2.0-beta.1-somestuff when building from a modified version of main?
slender iron
In the docs for `terminalio.Terminal` it makes reference to a cursor and moving ...
It is internal state. There currently isn't a way to see where it is in displayio.
We could add it though
slender iron
Woof. Why does the REPL tell me I have CP `6.2.0-beta.1-somestuff` when building...
you likely need to fetch newer tags
goes for a short walk
idle owl
Ah thanks.
Blergh. Updating what appears to be the rotation value is changing nothing. It started as 180, which was upside down, so I changed it to 0, which did nothing, so I thought maybe it was something else that changed with the use of the new module, so I put it back to 180. Still same. Hmm.
idle owl
This is like waiting for an image to download in the early '90s. 😂
Ok, verified, updating rotation doesn't change anything. I tried 90 even. Not sure how to go about resolving this.
onyx hinge
@lone axle for editor.py in particular, it always tracks where it thinks the cursor is, so you could try relying on that (def move on class Screen).
lone axle
We could add it though
Do you have thoughts on what might be the path of least resistence to adding the ability to see the cursor?
I had a look in terminalio.Terminal core code and can see that it has cursor_x and cursor_y variables internally. If those were added to shared-bindings to be available from python then with them and some pixel multiplication it could place something over the top as a rudimentary way to indicate a location.
Is it possible to access the terminalio.Terminal instance that is visible by default that mimics the serial console? I know there is displayio.CIRCUITPYHON_TERMINAL which is a Group that contains the TileGrid(s) that the Terminal is using, but I don't know of a way to access the Terminal instance itself in order to access any properties it may have.
slender iron
Do you have thoughts on what might be the path of least resistence to adding the...
It'd probably be better to do it internally so it works when the VM isn't running
lone axle
<@382939733107408897> for editor.py in particular, it always tracks where it thi...
Thank you! I was just tracing through that code. ( I never imagined how tricky print debugging could be in an app whose main output is the same stream where prints go 😅)
I think I could use the locations tracked in there to translate to a location on the display in pixels.
I'm not very experienced with the sys.stdin / stdout concepts that this seems to be using (though hope to learn from studying this code more).
Is there somewhere in here that would make it easyish to access which character is currently at the position of the Cursor? Or to access as a String what is currently visible on the display?
ah, I think I just found the answer to that last question. buffer is the list of strings for the current "screen" it looks like.
onyx hinge
yes I think that's true. Though, I was thinking you'd just access the terminal's TileGrid to find it out, that may work better
I think buffer is only (part of) the edited file, so for instance the status line wouldn't be there
i could be wrong, I haven't looked at this file in months
lone axle
Thank you. I'll keep poking at it. I missed this when you created it, very neat project! I never knew the default serial console in circuitpython was capable of this sort of nano-like control
onyx hinge
you'll need to get ahold of the terminal's tilegrid anyway to find the true width/height of the display
slender iron
@onyx hinge remember that idea to separate keys from values in dict tables?
onyx hinge
@slender iron yeah?
slender iron
I wonder if that'd help us for samd21 instead of error messages
onyx hinge
it's no easy undertaking
slender iron
board on arduino zero alone is 264 bytes
and the savings would be 1/4 of the bytes
so 66
builtins is 784 with a savings of 196
manic glacierBOT
midnight ember
Does this mean a physical sd directory in boards now or will it be a placeholder for a virtual /sd?
onyx hinge
@midnight ember 9 now requires the "mount point" (technical term) of an SD card [or other more exotic thing] to exist within CIRCUITPY as a directory. To make it easier on folks, the idea is to create a directory named "/sd" when the filesystem is created. If there's enough flash space, additional code to create an explanatory "/sd/placeholder.txt" file is included too.
midnight ember
Will the directory be writeable while the filesystem is in read-only? I would love that. If it's for virtual mounts only then the use of virtual /sd not creating files in the physical /sd directory might be more confusing for beginners.
candid sun
@onyx hinge assuming example code will need to be updated similar to how we went thru and updated display.show() to display.root_group=
onyx hinge
I think it's more a matter of instructions -- because CIRCUITPY will be read only to Python code [without using boot.py or other methods] the user will have to create CIRCUITPY/sd manually from their host computer.
midnight ember
Since I'm not really using 9.0 in my main projects yet will just wait for the roll out and cross that bridge when the time comes.
All the new changes and amazing projects every week is brain candy. It's a wonderful thing.
manic glacierBOT
empty salmon
I’ve been thinking about how CircuitPython might support dual-core cpus.
Support for the end user has lots of complications.
[spitballing] Might there be a solution where the board developer can specify what might run on each core? It would logically be things which take a lot of cycles, run continuously, and/or use buffering so things can be “loaded up” and left to run.
stuck elbow
would there be any reason to not run on the second core something that could run there?
things like wifi, sound, display, etc.
empty salmon
That is my thinking. For boards with WiFi and two cores, it’s a logical choice. Sound is another great subsystem to consider.
It would require a buffering solution to handle data from CircuitPython in one core to/from the subsystem running on the other core. But an “under the covers” solution could be easier to implement and little/no change for the CircuitPython user.
stuck elbow
we already have sounds playing in the background using buffers
the main problem I see is that it would all still need to work the same way on boards with a single core, so the implementation might be tricky
a good architecture would help
empty salmon
Background execution on “the other core” is the goal. The most popular microcontrollers (by far) now are dual core.
mortal kernel
Background execution on “the other core” is the goal. The most popular microcont...
For more capable multi-core SOCs it may make sense to add an RTOS kernel above CircuitPython. This would have the advantage of exposing a mature interface for multi-core support and would future-proof for coming 4 and up multi-core SOCs.
solar whale
for the OV5640 camera with auto focus https://www.adafruit.com/product/5840 the product page says Note that to use the auto-focusing capability, you need to load a new firmware binary over I2C Does anyone know where to get the new firmware and how to load it? I have not been able to find any instructions.
spare jacinth
after some digging, seems to be on adafruit_pycamera
and (trying?) using the bin is the default option in the initializer (it ends up causing autofocus_init() a couple of nested functions later)
crimson ferry
@mortal kernel I'm still getting AP could not be started with 9.0.0-beta.0-5-ged345f4ea2 on 2024-01-30
(adafruit-circuitpython-raspberry_pi_pico_w-en_US-20240130-main-PR8861-ed345f4.uf2)
solar whale
and (trying?) using the bin is the default option in the [initializer](<https://...
Ah, thanks, so it is already implemented in the library. That’s nice!
slender iron
That is my thinking. For boards with WiFi and two cores, it’s a logical choice. ...
We already do this. On the S3 CircuitPython runs on one core and the network stack on the other. On RP2040 CP runs on one core and usb host or picodvi will run on the other.
mortal kernel
<@849815081251766292> I'm still getting `AP could not be started` with `9.0.0-b...
That's a bummer. I've got an exploded water heater here, so I won't be able to get back to it until later today or early tomorrow.
empty salmon
@slender iron - nice! Where can a development learn how to do this?
Specificly, on an RP2040,which is not implementing USB host, how can the developer move “something” to the second core?
slender iron
<@252717193496756235> - nice! Where can a development learn how to do this?
Spe...
It's manual now. Moving CP to an RTOS would make this easier (and a reason to do the switch.)
it is easier with esp-idf stuff because it uses freertos
but all of CP runs in one task currently
slender iron
what if we used python stubs to generate the locals dict tables?
that way we don't have the requirement to list key next to value
mortal kernel
<@849815081251766292> I'm still getting `AP could not be started` with `9.0.0-b...
I think I've got the submodule updated correctly. The branch parameter in .gitmodules was ignored. I had to take the additional step of manually checking out the branch.
slender iron
I think I've got the submodule updated correctly. The `branch` parameter in `.gi...
ya, you'll likely need to checkout the commit you want to use in the submodule too
crimson ferry
@mortal kernel testing now, looks good 
mortal kernel
ya, you'll likely need to checkout the commit you want to use in the submodule t...
Yes, that's what worked for me. And there I thought I was being so clever using git submodule set-branch -b circuitpython9 ports/raspberrypi/lib/cyw43-driver.
slender iron
that's good to set too so that it's clear where it is coming from
mortal kernel
Phew, back to the water heater.
keen eagle
I have a PR that adds a couple of methods to wifi.Radio for discovering connected stations in AP mode. What's the best way for y'all to figure out the specifics of the calls that would feel right to you? Currently it returns a list of dicts for the stations but I can see the wisdom in defining a wifi.Station class and returning those instead. Would you prefer discussing it in the PR or here or something else?
slender iron
I have a [PR](https://github.com/adafruit/circuitpython/pull/8820) that adds a c...
PR is probably best. chat is pretty ephemeral
empty salmon
how to get latest ?
Today I sync’d my fork of CircuitPython with that from Adafruit. I then checked out main, did a pull, checked out my board branch. Did a merge. And sync’d to my fork.
Everything seemed right.
When I build and install, boot_out.txt still shows 9.0.0-alpha.5-50
Is there instructions I can review to do this correctly so I have the latest (currently beta.0)?
stuck elbow
git pull --tags
I usually just do git remote update or how it was
by default git doesn't pull tag information, because Linus doesn't like it
empty salmon
Where is the version info within the repository so I can verify I have it without going thru the build and deploy every time?
idle owl
Hello, all. I'm working on trying to get the Inky Frame 7.3" working with CircuitPython. It's definitely on its way, with a few hiccups. One of them is the display is rotated 180 degrees. I changed the rotation value in board.c but it has no effect. I tried 90 even, simply to verify. I have no idea where to go with this.
The other issue is that the buttons are on a shift register, and they're reading fine with the keypad module, but they show up as number 3-7. There are 5 buttons total. I also don't know where to look into that.
slender iron
Where is the version info within the repository so I can verify I have it withou...
git describe --tags
empty salmon
@slender iron - OK. That still shows “alpha” so I am missing a step somewhere.
When I look at my fork on GitHub it said the main branch is up-to-date
I pull from my fork to my build machine. It still shows “alpha”
slender iron
<@252717193496756235> - OK. That still shows “alpha” so I am missing a step some...
I usually do git fetch adafruit and it gets the tags
tags aren't technically part of a branch. so it may only be updating the branch
empty salmon
Sorry to be a pain. … from my local machine, what is the full command to get tags from aidsfruit?
slender iron
what remotes do you have? (git remote -v)
onyx hinge
@empty salmon in the top directory, make fetch-tags
empty salmon
I get an error “does not appear to be a git repository” if I use “upstream”
onyx hinge
(which should work regardless of your git remote settings)
empty salmon
Thanks! Worked.
onyx hinge
good
iron fractal
Is there an up to date guide on building C extensions for CircuitPython? Or just a good example somewhere on GitHub? The Adafruit guide dates back a few years.
slender iron
Is there an up to date guide on building C extensions for CircuitPython? Or just...
No updated guide but this is a good place to get help with it.
This PR adds a new module with only functions at the top level: https://github.com/adafruit/circuitpython/pull/8786
iron fractal
Is it still the case that a new C extension requires recompiling all of CircuitPython, there's no dynamic linking?
stuck elbow
yes
empty salmon
Did something change?
- my board has an integrated ePaper display.
- 2 or 3 months ago, if there was an error that dumped to REPL, the error would show on the display. (I didn’t like it).
- Now, it seems errors don’t show on the ePaper display.
I’m happy but confused.
slender iron
Did something change?
- my board has an integrated ePaper display.
- 2 or 3 mon...
I recently fixed an issue with refresh sequences breaking. It should still show the error though
empty salmon
@slender iron - hmmmm …. I will do more testing. I tried a few things and could never get an error on the ePaper display.
Addendum: I want to reiterate, I like that errors don’t appear on the ePaper display. I just know we rode this merry go round a couple months ago and at that time the consensus was that errors should appear on ePaper. 😉
lone axle
It is very cobbled together but I've managed to show a visible cursor on top of the Terminal with some modifications to Jeff's editor project 🎉
onyx hinge
It is very cobbled together but I've managed to show a visible cursor on top of ...
well done!
onyx hinge
oh hey there's a way to see your open pull requests across all repositories on github: https://github.com/search?q=type%3Apr+author%3A%40me+is%3Aopen&type=pullrequests
huh there is no way to close your own PR in an archived repo, so it'll hang out in your list of open PRs forever. e.g., https://github.com/tenko/gltools/pull/3
aha, you can use archived: false https://github.com/search?q=type%3Apr+author%3A%40me+is%3Aopen+archived%3Afalse&type=pullrequests&ref=advsearch&s=updated&o=asc
onyx hinge
@slender iron did you want the whole "uvc" module renamed "usb_uvc", or just the #define ? I asked on the PR last night but asking again so I can get working on it now.
@random junco hey, what's a good permalink for circuitpython show podcasts? awesome-circuitpython has an outdated domain that doesn't work (hxxps://circuitpythonshow.com)
random junco
<@273162248615100417> hey, what's a good permalink for circuitpython show podcas...
Thanks for asking, please use https://www.circuitpythonshow.com/@circuitpythonshow
onyx hinge
@tulip sleet I'm talking to Kattni about the Inky Frame. One of the problems is that the buttons share a shift register with "other stuff" including the e-ink busy signal .. and those bits get shifted out first. What do you think about adding a "skip # bits at start" for ShiftRegisterKeys to accomodate this 'feature'?
D7 gets shifted out first so "key 0" is "inky busy", and physical buttons E...A are mapped onto keys 3..7 in that order, if I understand right what's happening
otherwise we can probably work through some code to discard events that don't matter
onyx hinge
[belatedly remembers dan's out this week, sorry for the ping]
slender iron
<@252717193496756235> did you want the whole "uvc" module renamed "usb_uvc", or ...
I like usb_uvc more
onyx hinge
I like `usb_uvc` more
OK
tulip sleet
<@329766224093249548> I'm talking to Kattni about the Inky Frame. One of the pro...
I think just or ignoring unwanted events is easiest. The unwanted signals may be in any order, and could be unused buttons, etc. Do I wouldn’t special case anything. If the transitions on the unwanted pins are really swamping the event queue, then I think a bit mask would be the thing to do.
slender iron
@limber tinsel I'm not sure whether we've tried building silabs on mac
limber tinsel
<@1202326084712923136> I'm not sure whether we've tried building silabs on mac
Ah...
slender iron
You may need to change this line: https://github.com/adafruit/circuitpython/blob/main/ports/silabs/Makefile#L39
seems like you got almost to the end of the build
limber tinsel
OK, time for me to learn how those make files work 😆
slender iron
you can do make V="steps commands" to see what it is running
limber tinsel
Thanks for the help, I'll see if I can edit the file to get to the correct place.
slender iron
👍 I'm in pacific time so I'll be around for five or so more hours
I did manage to build it on linux
limber tinsel
ok cool, thanks, I am in PT as well.
atomic summit
Any improvements in the latest macOS beta with the FAT issues?
limber tinsel
ok cool, thanks, I am in PT as well.
slender iron
Any improvements in the latest macOS beta with the FAT issues?
I don't think so. I think there was an update on the issue that went by
atomic summit
I don't think so. I think there was an update on the issue that went by
Oh, ok, I'll take a look. Thanks Scott.
I'm still running Ventura on my main work Mac as I was hoping this would get sorted before I upgraded it... but I'm going to have to move soon.
Oh, lots of updates in that issue - I've not been getting email notifications, weird.
slender iron
@atomic summit can all of your boards do quad spi for flash and ram at 80mhz?
atomic summit
<@195765176035835904> can all of your boards do quad spi for flash and ram at 80...
Technically they should all be able to. I thought we had already set all of that up, at least for 9.0.0?
slender iron
I think your boards I did. I'm searching and replacing for the rest...
atomic summit
Ahh, ok.
slender iron
it won't change octal ones. just dio
idle owl
The Inky Frame is working with the latest CircuitPython changes. There are two sections towards the beginning of pins.c that handle making user-facing names for the buttons (because they are otherwise handled by a shift register, and return bit numbers). However, they also include the RTC alarm, ext trigger, and the display busy pin. This means that if use keypad.ShiftRegisterKeys and print the key_number, every time the busy pin is triggered, it shows that key_number 0 pressed. I wonder whether there's a way to separate out the three extra items, and still make them available in board, but not have them being picked up by keypad.
Anyway, that's my latest update. I'm heading out.
orchid basinBOT
willow totem
@slender iron apologies for the ping, got a last minute #circuitpython-2024 wish... I would love to be able to use long range wifi on the esp, it gets about 500m... no LoRa license fees or extra hardware required (but benefits from external aerial of course)... <#help-with-wippersnapper-and-adafruitio message>
I should have waited five minutes, they just gave the email address for it 🤦
midnight ember
That would be nice. Might negate the need for RFM with my mailbox project.
main furnace
from adafruit_featherwing.tft_featherwing_35 import TFTFeatherWing35
wing = TFTFeatherWing35()
Adafruit CircuitPython 9.0.0-beta.0 on 2024-01-27; Adafruit Feather RP2040 USB Host with rp2040
Traceback (most recent call last):
File "code.py", line 2, in <module>
File "adafruit_featherwing/tft_featherwing_35.py", line 49, in __init__
File "adafruit_featherwing/tft_featherwing.py", line 81, in __init__
RuntimeError: Mount point directory missing
Fix: Create /sd on CIRCUITPY.
slender iron
It is very cobbled together but I've managed to show a visible cursor on top of ...
This looks great! We should add it natively
lone axle
This looks great! We should add it natively
I can give it a try to add in the core, but It's probably beyond my means to get it working in time to use for the editor project for next week.
The current implementation of this relies on code in Jeffs adafruit_editor library which holds strings representing what is currently visible within it's terminal window.
I only had a quick look, but I didn't see anything inside of terminalio.Terminal that keeps that data as a string, or provides a way to read it back out of the Terminal after it's been written.
Right now the cursor is a 1 character Label that gets updated with the character currently at the X,Y location of the cursor each time it updates. So being able to get the string representation is necessary at least how it works now.
If there were a way to invert, or manipulate the palette used by a single tile within a TileGrid that might offer an even easier way to achieve a very similar visual.
idle owl
Anyone have any idea how to install an older version of uncrustify on MacOS? Their own GitHub repo only has the two latest versions, which do not appear to be compatible with CircuitPython pre-commit.
slender iron
Right now the cursor is a 1 character Label that gets updated with the character...
The terminal doesn't know the string but it does know the tile index. It can copy that out. No worries if we need to add it later. It looks good either way!
slender iron
If there were a way to invert, or manipulate the palette used by a single tile w...
I thought about that but it might mean more memory use to store which of the tiles are inverted
lone axle
do you know if there is a way to get from a tile index to the character that it represents within a font? get_glyph_index() goes the other way from character to tile index. If we could reverse that then we could read characters from the TileGrid.
slender iron
I think you'd need to build a reverse table lookup
the tilegrid only knows what is on screen though
lone axle
Maybe it could copy the bitmap data for 1 tile into a new bitmap to put on top. Then it won't really need to know the character, just be able to copy from the source region and flip the colors.
slender iron
Maybe it could copy the bitmap data for 1 tile into a new bitmap to put on top. ...
it doesn't need to know the character. only the index into the bitmap
mortal kernel
Anyone have any idea how to install an older version of uncrustify on MacOS? The...
I'm running this version: Uncrustify-0.78.1_f from Homebrew on MacOS Sonoma. pre-commit works as expected with this config.
idle owl
I'm running this version: `Uncrustify-0.78.1_f` from Homebrew on MacOS Sonoma. `...
Really? I got a ton of errors seemingly related to the circuitpython/tools/uncrustify.cfg file.
I reinstalled uncrustify, and now it works. 🤷🏻♀️
Thanks @mortal kernel!
Inconsistent behavior. My favorite. 🙄
mortal kernel
Inconsistent behavior. My favorite. 🙄
Part of the fun as I see it 🥸 . Seriously though, it's little problems like this that keep me on Linux for most of my dev work.
Quick question: What is the best way to measure elapsed wall clock time inside CP core code? Looking for something with millisecond accuracy suitable for use in a polling loop.
I'm thinking that supervisor_ticks_msxx() is the way to go.
slender iron
I'm thinking that `supervisor_ticks_msxx()` is the way to go.
I think so too
amber sundial
@gilded cradle I was wondering, on your journey to make the display ports for qualia to circuit python, did you first make them for the ESP IDF?
They seem to have drivers for some chips here .. https://github.com/espressif/esp-iot-solution/tree/master/components/display/lcd
I was wondering if you have something for the nv3052 .. there's something for the nv3022 here
gilded cradle
<@478616919721771019> I was wondering, on your journey to make the display ports...
@onyx hinge (who is off for the next couple weeks or so) originally wrote them and wrote up some technical information in a guide. I took that and reformatted a lot of it into a regular guide, did a lot of testing, figured out init settings if they were missing or not quite right, and wrote the qualia library as well as adding some Arduino stuff. I don't believe they were written specifically for the ESP IDF, but I could be wrong.
amber sundial
<@426200547024961539> (who is off for the next couple weeks or so) originally wr...
OK! I bought an awesome looking nv3052-based display on alibaba and have a qualia coming in tomorrow .. I hope I can get it to work. If so I will share the results!
gilded cradle
OK! I bought an awesome looking nv3052-based display on alibaba and have a quali...
I believe the round 4" 720x720 display is actually NV3052, so I would start with that. There are pages in the learn guide about figuring out the settings.
amber sundial
I believe the round 4" 720x720 display is actually NV3052, so I would start with...
perfect I had a read through already and that's what I'll try first!
lone axle
Is it possible to call something equivalent to usb_cdc.data.write() from within the core? I found this object: https://github.com/adafruit/circuitpython/blob/8e3e95307b56f66ef3833cf547e74ff5e1368a03/shared-module/usb_cdc/__init__.c#L162-L166 which seems possibly related, but I'm not certain if it actually is and thus far I've been unable to call write() on it if so.
stuck elbow
Is it possible to call something equivalent to `usb_cdc.data.write()` from with...
looks like you want to call https://github.com/adafruit/circuitpython/blob/8e3e95307b56f66ef3833cf547e74ff5e1368a03/shared-bindings/usb_cdc/Serial.c#L109 with it?
lone axle
is something like this getting close?
common_hal_usb_cdc_serial_write(usb_cdc_data_obj, "hi", 2, 0);
I'd guess so, but I'm not really good with C
not sure if the size should include the terminating \0 for example...
lone axle
are you trying to print debug messages out to the console?
yes, in the circuitpython code I started doing that so that debugging prints don't interfere with the Terminal. Being able to do the same from the core will help me poke around and try implementing the cursor in there.
lone axle
I'm not sure what that entails or how to do it.
slender iron
makes a console_uart_printf() that only goes out the UART
you could make something similar for usb_data
lone axle
Interesting, I could use UART as output as well if that one is hooked up already.
I tried a build with this code:
char myMessage[] = "hi";
common_hal_usb_cdc_serial_write(usb_cdc_data_obj, (uint8_t *)myMessage, 2, 0);
but it gets this error
../../shared-module/terminalio/Terminal.c:67:37: error: incompatible type for argument 1 of 'common_hal_usb_cdc_serial_write'
67 | common_hal_usb_cdc_serial_write(usb_cdc_data_obj, (uint8_t *)myMessage, 2, 0);
| ^~~~~~~~~~~~~~~~
| |
| usb_cdc_serial_obj_t
In file included from ../../shared-module/usb_cdc/__init__.c:33,
from ../../shared-module/terminalio/Terminal.c:33:
../../shared-bindings/usb_cdc/Serial.h:35:69: note: expected 'usb_cdc_serial_obj_t *' but argument is of type 'usb_cdc_serial_obj_t'
35 | extern size_t common_hal_usb_cdc_serial_write(usb_cdc_serial_obj_t *self, const uint8_t *data, size_t len, int *errcode);
| ~~~~~~~~~~~~~~~~~~~~~~^~~~
make: *** [../../py/mkrules.mk:85: build-pyportal_titano/shared-module/terminalio/Terminal.o] Error 1
My guess was that I have pointer vs. referenc eof the cdc_data_obj wrong, but I'm not sure how to do it properly (I really should try to take a step back and learn more C fundamentals).
I did also try:
common_hal_usb_cdc_serial_write(&usb_cdc_data_obj, (uint8_t *)myMessage, 2, 0);
and it seems to get further in the build process but does still result in other errors.
slender iron
Interesting, I could use UART as output as well if that one is hooked up already...
the & addition is right. what are the other errors?
lone axle
the & addition is right. what are the other errors?
GEN build-pyportal_titano/frozen_content.c
/home/timc/repos/circuitpython/gcc/gcc-arm-none-eabi-10-2020-q4-major/bin/../lib/gcc/arm-none-eabi/10.2.1/../../../../arm-none-eabi/bin/ld: build-pyportal_titano/shared-module/terminalio/Terminal.o (symbol from plugin): in function `usb_cdc_console_enabled':
(.text+0x0): multiple definition of `usb_cdc_console_enabled'; build-pyportal_titano/shared-module/usb_cdc/__init__.o (symbol from plugin):(.text+0x0): first defined here
/home/timc/repos/circuitpython/gcc/gcc-arm-none-eabi-10-2020-q4-major/bin/../lib/gcc/arm-none-eabi/10.2.1/../../../../arm-none-eabi/bin/ld: build-pyportal_titano/shared-module/terminalio/Terminal.o (symbol from plugin): in function `usb_cdc_console_enabled':
(.text+0x0): multiple definition of `usb_cdc_data_enabled'; build-pyportal_titano/shared-module/usb_cdc/__init__.o (symbol from plugin):(.text+0x0): first defined here
/home/timc/repos/circuitpython/gcc/gcc-arm-none-eabi-10-2020-q4-major/bin/../lib/gcc/arm-none-eabi/10.2.1/../../../../arm-none-eabi/bin/ld: build-pyportal_titano/shared-module/terminalio/Terminal.o (symbol from plugin): in function `usb_cdc_console_enabled':
(.text+0x0): multiple definition of `usb_cdc_descriptor_length'; build-pyportal_titano/shared-module/usb_cdc/__init__.o (symbol from plugin):(.text+0x0): first defined here
/home/timc/repos/circuitpython/gcc/gcc-arm-none-eabi-10-2020-q4-major/bin/../lib/gcc/arm-none-eabi/10.2.1/../../../../arm-none-eabi/bin/ld: build-pyportal_titano/shared-module/terminalio/Terminal.o (symbol from plugin): in function `usb_cdc_console_enabled':
(.text+0x0): multiple definition of `usb_cdc_add_descriptor'; build-pyportal_titano/shared-module/usb_cdc/__init__.o (symbol from plugin):(.text+0x0): first defined here
/home/timc/repos/circuitpython/gcc/gcc-arm-none-eabi-10-2020-q4-major/bin/../lib/gcc/arm-none-eabi/10.2.1/../../../../arm-none-eabi/bin/ld: build-pyportal_titano/shared-module/terminalio/Terminal.o (symbol from plugin): in function `usb_cdc_console_enabled':
(.text+0x0): multiple definition of `common_hal_usb_cdc_enable'; build-pyportal_titano/shared-module/usb_cdc/__init__.o (symbol from plugin):(.text+0x0): first defined here
/home/timc/repos/circuitpython/gcc/gcc-arm-none-eabi-10-2020-q4-major/bin/../lib/gcc/arm-none-eabi/10.2.1/../../../../arm-none-eabi/bin/ld: build-pyportal_titano/shared-module/terminalio/Terminal.o (symbol from plugin): in function `usb_cdc_console_enabled':
(.text+0x0): multiple definition of `usb_cdc_set_defaults'; build-pyportal_titano/shared-module/usb_cdc/__init__.o (symbol from plugin):(.text+0x0): first defined here
/home/timc/repos/circuitpython/gcc/gcc-arm-none-eabi-10-2020-q4-major/bin/../lib/gcc/arm-none-eabi/10.2.1/../../../../arm-none-eabi/bin/ld: build-pyportal_titano/shared-module/terminalio/Terminal.o (symbol from plugin): in function `usb_cdc_console_enabled':
(.text+0x0): multiple definition of `common_hal_usb_cdc_disable'; build-pyportal_titano/shared-module/usb_cdc/__init__.o (symbol from plugin):(.text+0x0): first defined here
Memory region Used Size Region Size %age Used
FLASH_BOOTLOADER: 0 GB 16 KB 0.00%
FLASH_FIRMWARE: 578204 B 1000 KB 56.47%
FLASH_FILESYSTEM: 0 GB 0 GB
FLASH_CONFIG: 0 GB 0 GB
FLASH_NVM: 0 GB 8 KB 0.00%
RAM: 10936 B 256 KB 4.17%
collect2: error: ld returned 1 exit status
make: *** [Makefile:382: build-pyportal_titano/firmware.elf] Error 1
perhaps the way I did the include is incorrect?
#include "shared-module/usb_cdc/__init__.c"
with either of these:
#include "shared-bindings/usb_cdc/__init__.h"
// or
#include "shared-module/usb_cdc/__init__.h"
it gets this same undeclared error:
../../shared-module/terminalio/Terminal.c: In function 'common_hal_terminalio_terminal_construct':
../../shared-module/terminalio/Terminal.c:67:5: error: implicit declaration of function 'common_hal_usb_cdc_serial_write' [-Werror=implicit-function-declaration]
67 | common_hal_usb_cdc_serial_write(&usb_cdc_data_obj, (uint8_t *)myMessage, 2, 0);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../../shared-module/terminalio/Terminal.c:67:5: error: nested extern declaration of 'common_hal_usb_cdc_serial_write' [-Werror=nested-externs]
../../shared-module/terminalio/Terminal.c:67:38: error: 'usb_cdc_data_obj' undeclared (first use in this function); did you mean 'usb_cdc_data_enabled'?
67 | common_hal_usb_cdc_serial_write(&usb_cdc_data_obj, (uint8_t *)myMessage, 2, 0);
| ^~~~~~~~~~~~~~~~
| usb_cdc_data_enabled
../../shared-module/terminalio/Terminal.c:67:38: note: each undeclared identifier is reported only once for each function it appears in
cc1: all warnings being treated as errors
make: *** [../../py/mkrules.mk:85: build-pyportal_titano/shared-module/terminalio/Terminal.o] Error 1
where in the file are you including it?
try shared-bindings/usb_cdc/Serial.h
rg common_hal_usb_cdc_serial_write ../..
../../shared-module/usb_cdc/Serial.c
81:size_t common_hal_usb_cdc_serial_write(usb_cdc_serial_obj_t *self, const uint8_t *data, size_t len, int *errcode) {
../../shared-bindings/usb_cdc/Serial.c
109: return common_hal_usb_cdc_serial_write(self, buf, size, errcode);
../../shared-bindings/usb_cdc/Serial.h
35:extern size_t common_hal_usb_cdc_serial_write(usb_cdc_serial_obj_t *self, const uint8_t *data, size_t len, int *errcode);
The include statement itself is at the top of the file with the other pre-existing includes. I've put it inside of shared-module/terminalio/Terminal.c and the line trying to actually use it is down inside of common_hal_terminalio_terminal_construct() where I just stuck it as a test to see if I could get it building.
with #include "shared-bindings/usb_cdc/Serial.h" it gets undeclared error:
../../shared-module/terminalio/Terminal.c: In function 'common_hal_terminalio_terminal_construct':
../../shared-module/terminalio/Terminal.c:68:38: error: 'usb_cdc_data_obj' undeclared (first use in this function); did you mean 'usb_cdc_serial_obj_t'?
68 | common_hal_usb_cdc_serial_write(&usb_cdc_data_obj, (uint8_t *)myMessage, 2, 0);
| ^~~~~~~~~~~~~~~~
| usb_cdc_serial_obj_t
../../shared-module/terminalio/Terminal.c:68:38: note: each undeclared identifier is reported only once for each function it appears in
make: *** [../../py/mkrules.mk:85: build-pyportal_titano/shared-module/terminalio/Terminal.o] Error 1
../../shared-module/usb_cdc/__init__.c
162:static usb_cdc_serial_obj_t usb_cdc_data_obj = {
258: usb_cdc_set_data(data ? MP_OBJ_FROM_PTR(&usb_cdc_data_obj) : mp_const_none);
260: usb_cdc_data_obj.idx = idx;
You'll need to remove the static and put an extern in the corresponding header
keen eagle
It's stored as data in usb_cdc_module's .globals, would it make sense to fetch it from there rather than exposing the underlying C struct's name as a global? Not that I know how to get a value from a module's dict in C.
manic glacierBOT
solar whale
I just tried building from a clean install on MAC OS sonoma -- I am trying an espressif build for adafruit_esp32s3_camera but it fails with ```The following Python requirements are not satisfied:
Requirement 'pyelftools<=0.29' was not met. Installed version: 0.30
To install the missing packages, please run "install.sh"
Diagnostic information:
IDF_PYTHON_ENV_PATH: /Users/jerryneedell/.espressif/python_env/idf5.1_py3.11_env
Python interpreter used: /Users/jerryneedell/.espressif/python_env/idf5.1_py3.11_env/bin/python
Constraint file: /Users/jerryneedell/.espressif/espidf.constraints.v5.1.txt
Requirement files:
- /Volumes/CircuitPythonBuild/circuitpython/ports/espressif/esp-idf/tools/requirements/requirements.core.txt
Python being checked: /Users/jerryneedell/.espressif/python_env/idf5.1_py3.11_env/bin/python
CMake Error at esp-idf/tools/cmake/build.cmake:366 (message):
Failed to run Python dependency check. Python: python, Error: 255
Call Stack (most recent call first):
esp-idf/tools/cmake/build.cmake:498 (__build_check_python)
esp-idf/tools/cmake/project.cmake:547 (idf_build_process)
CMakeLists.txt:12 (project)
-- Configuring incomplete, errors occurred!
gmake: *** [Makefile:420: build-adafruit_esp32s3_camera/esp-idf/config/sdkconfig.h] Error 1
jerryneedell@Mac-mini espressif %
When runnig the install.sh I see ```Installing collected packages: pyelftools, cryptography
Attempting uninstall: pyelftools
Found existing installation: pyelftools 0.30
Uninstalling pyelftools-0.30:
Successfully uninstalled pyelftools-0.30
Attempting uninstall: cryptography
Found existing installation: cryptography 36.0.2
Uninstalling cryptography-36.0.2:
Successfully uninstalled cryptography-36.0.2
Successfully installed cryptography-39.0.2 pyelftools-0.29
All done! You can now run:
. esp-idf/export.sh
I think that is where it is accessed from in python land, but I'm not sure if it's possible or how to access it from C in the core.
lone axle
When runnig the install.sh I see ```Installing collected packages: pyelftools, c...
Thats what it looks like to me in the first error it mentioned wanting <=0.29 but you had 0.30 originally.
solar whale
Is the install.sh not getting the right versions?
lone axle
the output in your second copy/pasted chunk looks like it did successfully uninstall 0.30 and then install 0.29, so from that output it seems like it suceeded in getting the version it wants to have as far as I can tell.
solar whale
hmmm -- okay -- it is happier now -- confusing, but happier...
It seems like it took a few tries to get it right, but it is working now...
manic glacierBOT
CircuitPython version
Adafruit CircuitPython 9.0.0-alpha.6 on 2023-12-12; Waveshare RP2040-Zero with rp2040
Code/REPL
from collections import deque
sa = deque((), 4)
list(sa)
Behavior
It's pretty self-explanatory:
Description
No response
Additional information
No response
lost wing
Good morning! I was trying to create a USB MIDI host using a Lolin S2 Mini (equipped with an ESP32-S2, so supposedly should have support for USB host). But... I found out that usb.core is not supported on that board. Why is that? Is there any workaround for this?
stuck elbow
a far as I know usb host is still very much work in progress
random junco
I thought USB host was iMX and rp2040 based on the release notes
lost wing
I understand it's released for several boards, isn't it?
stuck elbow
it usually requires special hardware to be added to the board
manic glacierBOT
lost wing
@stuck elbow I understand. But ESP32-S2 supposedly has built in host capabilities. I've been trying to make it work in Arduino framework to no success. I was hoping I was more lucky with Circuitpython...
stuck elbow
<@351136009922019331> I understand. But ESP32-S2 supposedly has built in host ca...
the fact that the chip has some capabilities doesn't mean that circuitpython has support for them, or that the particular board has the hardware for them
lost wing
the fact that the chip has some capabilities doesn't mean that circuitpython has...
By the manufacturer, it has USB OTG built in
stuck elbow
By the manufacturer, it has USB OTG built in
do they also have documentation on how to use it?
lost wing
Kind of... They have a library for ESP-IDF framework. It should work in Arduino too . This example should work https://github.com/touchgadget/esp32-usb-host-demos (but in my case, it didn't)
digital shoreBOT
slender iron
It's stored as `data` in `usb_cdc_module`'s `.globals`, would it make sense to f...
Ya, or it looks like there is a usb_cdc_set_data(). We could add usb_cdc_get_data() if it doesn't exist.
slender iron
Kind of... They have a library for ESP-IDF framework. It should work in Arduino ...
USB OTG isn't supported by TinyUSB yet. We only support USB host in CircuitPython where we can have a second USB instance dedicated for host.
(So usb_host isn't implemented on ESP)
we also want to support an external host chip from circuitpython but we haven't added that yet
lost wing
USB OTG isn't supported by TinyUSB yet. We only support USB host in CircuitPytho...
I understand... things are much more complicated with 1 USB only. But there are some ESP boards with 2 USBs, like these, no? https://docs.espressif.com/projects/esp-idf/en/latest/esp32s3/hw-reference/esp32s3/user-guide-devkitc-1.html
slender iron
I understand... things are much more complicated with 1 USB only. But there are ...
One is just an external USB to serial adapter. So from the ESP perspective it is only one USB still.
we do want to support OTG eventually
but the external host chip is more of a priority
keen eagle
I think that is where it is accessed from in python land, but I'm not sure if it...
I was thinking there's probably a way to lookup something in the dict but I'm already at the limit of my knowledge here
slender iron
I was thinking there's probably a way to lookup something in the dict but I'm al...
Ya, there definitely is.
looks
mp_map_lookup
keen eagle
if you don't know it off the top of your head I feel better about failing to figure it out 😉
main furnace
Unpopular work-around for Sonoma: Disable CIRCUITPY, use ampy to write files to the device.
mortal kernel
My knowledge of CP memory management wouldn't fill a teacup, so before I resort to the code I'm wondering if there's a write-up somewhere?
(I'm making some assumptions about what gets GC'ed that might be bad)
stuck elbow
My knowledge of CP memory management wouldn't fill a teacup, so before I resort ...
it's a mark and sweep gc
it starts with the root pointers, scans the memory they point to, and follows anything that looks like a pointer in there, marking all that as used, then it frees everything that's left
manic glacierBOT
mortal kernel
it starts with the root pointers, scans the memory they point to, and follows an...
So far, so good. I understand mark and sweep. If I use an m_ prefixed allocator like m_alloc I'm guaranteed that the allocation is GC'able?
stuck elbow
So far, so good. I understand mark and sweep. If I use an `m_` prefixed allocato...
sorry, no idea
the trick is that it doesn't know for sure if it's a pointer or just a part of some string or something, so sometimes it will keep pieces of memory that are not actually being pointed to
this is why it can't move things around - it can't go back and update the pointer, because it's not sure that it's a pointer and not just some random data
mortal kernel
this is why it can't move things around - it can't go back and update the pointe...
Looking into the code, I think I have it. m_new() is a macro that invokes m_alloc() which invokes malloc that is overridden by a macro in py/malloc.c that invokes gc_alloc(). So, the allocation will be GC'ed.
slender iron
So far, so good. I understand mark and sweep. If I use an `m_` prefixed allocato...
Yes
One thing we could do to save code size is remove the context manager of the core classes. I don't think I've ever seen someone use that
BusDevice uses it for transactions but not the core classes
mortal kernel
Yes
I would love to take a deep dive into CP memory management. Maybe after I've got _bleio buttoned up.
slender iron
note that you can only use m_malloc when the vm is running
(and ble runs outside the vm sometimes)
mortal kernel
(and ble runs outside the vm sometimes)
Whoa, where is that happening? With BTstack events are delivered only when polled, I'd assumed that was true for other BLE stacks as well.
slender iron
the ble workflow runs ble outside the vm
mortal kernel
the ble workflow runs ble outside the vm
I see. So memory allocations that happen in BLE workflow need to be static or ?
slender iron
they used to be static only
but you can use port_malloc in 9.0 to allocate outside the vm
no GC though
mortal kernel
Thank you! All good to know.
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
warm stump
Had another look at MKRVidor4000 as a FeatherM0Basic
mainly because I found a Dallas DS18B20 that I got ages ago and realised it might have some EEPROM inside it
...which originated from a search to find 1 floating gate memory cell as a through hole component...
CircuitPython is able to drive pins ~3, ~4, ~5 as D0, MOSI, SCK on a board it has no idea it's running on!...
The procedures outlined might be a bit experimental and latest versions should be used if possible
but I have included a quick port of the OWI/OWB driver from the webpage rewritten as def DS18B20_<Method>(Port, Data, etc):
manic glacierBOT
manic glacierBOT
Changes in 18 files is a lot. I copied the changes to my local build, compiled the image and uploaded the created firmware.bin to my T-Display S3 with esptool write_flash 0x0 firmware.bin. It appears as FTHRS3BOOT flash drive after reset. Now I copied the firmware.uf2 into this drive and the board is now running Circuitpython 9.0.0 with display:
The MADCTL value 0x60 seems...
manic glacierBOT
Pin description for STEMMA connector is also working correctly. Warning of missing pull up resistors if nothing is connected, and finding attached bus devices.
manic glacierBOT
short tendon
Interesting Note: on the ESP32S3 TFT, running 8.2.9 I got 3 open sockets before it raised an exception. On 9.0.0-beta.0 I got 4 (SSL to be specific)
celest marsh
Sending thanks to you all for having the majority of the conversations you all have be so open and available, I know that it must be slower to do it that way but it sure makes for a good community.
midnight ember
Is Liz hosting tomorrow? Unsure if Scott is hosting where to put my status update above or below in the doc.
Usually I'm the first one during the week to write in the doc so I don't have this dilemma. 😅
tame creek
so eventually i'll probably try to track down the RP2040 hard fault with a debugger, possibly with Pico SDK (because it's starting to look like the bug is there, or in the TinyUSB driver). still need an RP2040 board with accessible SWD lines. let me see if there's anything else i'm likely to want to order at the moment to piggyback on
manic glacierBOT
manic glacierBOT
manic glacierBOT
candid sun
<@&356864093652516868> We'll have our weekly meeting in about 90 minutes from now in this text channel and in the circuitpython voice channel. Please take the time to add your notes in advance to the document: https://docs.google.com/document/d/1TwKXc8VJyR-Zju5QWouz7psBYjrVBhrkBT4DtXMLVlQ/edit -- I look forward to everyone's updates!
slender iron
Is Liz hosting tomorrow? Unsure if Scott is hosting where to put my status upda...
I think Liz is. Everything but the host is usually alphabetical
ember iris
the use of the with statement does help a lot of traditional python programmers transitioning into circuit python. It has helped me explain a few concepts of what a micro can and can't do, but that's a different group
crimson ferry
probably worth an audit, hard to know how much extrnal code will break ; definitely need context managers on files, sockets, and other core-core stuff
ember iris
that sounds prudent 👍
short tendon
@slender iron if you let me know the branch for SSL changes, I can try to build and see if it's any different
solar whale
Things like uart would still allow it?
lone axle
Thanks for hosting Liz! Hope everyone has a great week 🎉
ember iris
Thanks all! Have a great week!
slender iron
Things like uart would still allow it?
potentially not. that's why I'm asking if people like having it
short tendon
<https://github.com/adafruit/circuitpython/pull/8868>
Is there a command to build a single board? I haven't done this yet
ember iris
I do really like the context manager because it's design pattern is something that's native to python, so I've found it helpful when explaining how to manage resources on a micro--if someone see's it in a with statement, it makes sense that the statement handles opening and closing of the resource. That said, the demographic I normally engage with is python programmers who have little to no electronics experience, and I can see that making up a fraction of a fraction of the userbase
lone axle
<@252717193496756235> if you let me know the branch for SSL changes, I can try t...
fwiw on PRs there are also "artifacts" that get built and uploaded to github. You have to navigate the actions section to find them as their a bit buried. But they can be downloaded and used from there without needing to build if you're specifically trying to build a PR branch. For that PR they're here: https://github.com/adafruit/circuitpython/actions/runs/7732060051 if you scroll down on the right half of the page to the artifacts section then find the device you want in the list.
short tendon
I think having it on large resourses (responses) and things you spicifically want to close after (sockets) makes the most sense if you want to limit it
lone axle
(note that the thing at the top of the right column prevents normal mouse scrolling because it's some kind of canvas drag n drop thing. But once you scroll past it by moving mouse outside then the page scrolls like normal.)
short tendon
Nice! I'll still try to build it, but will test with that one!
Still just got 4. Might try out a few other boards to do some comparisons
candid sun
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1vjpP9hQ4GBs0NdMWnkvaJG_9DMT7LGeDYivrRXcfbl4/edit
manic glacierBOT
short tendon
Question for the ConnectionManager.
I know you all have a ton to do, and for me when I have too many places to look it's hard to prioritize.
Would you taking ownership of the repo so it's in the Adafruit domain make it easier to see? I can easily fork and recreate the pull requests.
slender iron
Question for the ConnectionManager.
I know you all have a ton to do, and for me...
ConnectionManager is already under Adafruit right?
short tendon
Nope
short tendon
Hmmmm. I get You don’t have the permission to create public repositories on adafruit, is there a different way?
You could also just cookie-cutter it since that's all there is on main and then I can fork from there...
slender iron
Hmmmm. I get `You don’t have the permission to create public repositories on ada...
You can add me as an owner and then I can move it
short tendon
You can add me as an owner and then I can move it
Sent!
slender iron
got it
short tendon
Yay!
manic glacierBOT
slender iron
I’ll be off and on today
short tendon
@slender iron for whenever you are around: should I fork ConnectionManager and redo the PR (and close the other one)?
short tendon
@candid sun since you mentioned interest in my TimeHelper class, I threw together a quick GIST of my class I'm using. Would love any thoughts if you take time to look.
candid sun
<@347203168121126912> since you mentioned interest in my `TimeHelper` class, I t...
oh cool! yeah i'd love to take a look at it
manic glacierBOT
(Please ignore the value field for the ESP32-S3, a N16R8 was placed.)
Along with an additional 100uF on the 3v3 rail close to the ESP32 power in.
slender iron
<@252717193496756235> for whenever you are around: should I fork ConnectionManag...
nope, it looks like it moved just fine. I'll look now
you can fork for further changes
short tendon
so fork and make changes to the branch that has an open PR?
It will also only take 30 seconds to re-open it
slender iron
so fork and make changes to the branch that has an open PR?
I can give you access if you can't push to the existing branch. I just replied on the PR
short tendon
I can give you access if you can't push to the existing branch. I just replied o...
Looks like I can. Responded back on comments
short tendon
@lone axle so excited to see the stubs in there! Can't wait for the next beta release
manic glacierBOT
short tendon
@candid sun I was looking for the link in my comment to you to check something and realized I forgot to add it. Feel dumb...
Here it is:
https://gist.github.com/justmobilize/0a61e290fbb946815d172f58bb880b7c
hollow gazelle
hopefully a quick question -
I clicked on https://circuitpython.org/board/feather_m7_1011/
and it has a picture of some Feather boards, but the product link goes to a different shaped board
maybe there is a new feather M7 product in the works - but it is confusing to have a link go to another product
slender iron
hopefully a quick question -
I clicked on https://circuitpython.org/board/feath...
Maybe the product number was swapped for the metro. I think the feather was designed by a contractor first. It is on the TODO list but not soon.
plucky tulip
EDIT: Already merged
May I ask any of the CircuitPython developers to take a look at latest adafruit_httpserver PR, current release has a serious flaw which prevents usage.
The faster the fix is merged the fewer people will download the non-working version.
https://github.com/adafruit/Adafruit_CircuitPython_HTTPServer/pull/80
candid sun
<@347203168121126912> I was looking for the link in my comment to you to check s...
no prob, thanks for the link! i'll try to try this out later today
orchid basinBOT
devout jolt
Would it be reasonable to let fourwire.Fourwire allow for optional cs_pin?
Many times when using little TFT displays, I don't need or want to run a CS line. I'm devoting the SPI to the display, I tie the CS line to ground at the display, yet I need to use a cs_pin (I'll sometimes just stick it into an onboard LED if I'm not using that)
slender iron
Would it be reasonable to let `fourwire.Fourwire` allow for optional `cs_pin`?
...
Yup! I think so
slender iron
@tulip sleet I'm digging into why test-stubs.sh is only failing once when the message says it should fail twice. It looks like the two string type names in WriteableBuffer are preventing the test from erroring when passing in a bytes object
tulip sleet
<@329766224093249548> I'm digging into why test-stubs.sh is only failing once wh...
I see, but not sure why it should not just ignore those, because neither of them is bytes
I don't really know anything about mypy
so if you comment them out, it works?
slender iron
I agree. I don't know why it breaks. I just removed them and they work
there aren't any warnings about them
tulip sleet
but then that will break some library typechecking maybe
slender iron
circuitpython_typing/http.py:16: error: Skipping analyzing "adafruit_requests": module is installed, but missing library stubs or py.typed marker [import-untyped]
circuitpython_typing/device_drivers.py:12: error: Skipping analyzing "adafruit_bus_device.i2c_device": module is installed, but missing library stubs or py.typed marker [import-untyped]
circuitpython_typing/device_drivers.py:12: note: See https://mypy.readthedocs.io/en/stable/running_mypy.html#missing-imports
circuitpython_typing/__init__.py:39: error: Name "rgbmatrix" is not defined [name-defined]
circuitpython_typing/__init__.py:40: error: Name "ulab" is not defined [name-defined]
circuitpython_typing/__init__.py:48: error: Name "rgbmatrix" is not defined [name-defined]
circuitpython_typing/__init__.py:49: error: Name "ulab" is not defined [name-defined]
circuitpython_typing/__init__.py:131: error: Name "audiocore" is not defined [name-defined]
circuitpython_typing/__init__.py:132: error: Name "audiocore" is not defined [name-defined]
circuitpython_typing/__init__.py:133: error: Name "audiomixer" is not defined [name-defined]
circuitpython_typing/__init__.py:134: error: Name "audiomp3" is not defined [name-defined]
circuitpython_typing/__init__.py:135: error: Name "synthio" is not defined [name-defined]
circuitpython_typing/__init__.py:141: error: Name "rgbmatrix" is not defined [name-defined]
circuitpython_typing/__init__.py:144: error: Name "alarm" is not defined [name-defined]
circuitpython_typing/socket.py:129: error: Name "_FakeSSLContext" is not defined [name-defined]
Found 14 errors in 4 files (checked 8 source files)```maybe @short tendon understands python typing
short tendon
maybe <@1152454697614069811> understands python typing
Is this from the new stub work, or a big that's been around for a while?
slender iron
it has been around a while I think
short tendon
Okay. I'm happy to take a peek tomorrow
slender iron
Okay. I'm happy to take a peek tomorrow
thank you!
tulip sleet
slender iron
<@252717193496756235> <https://docs.espressif.com/projects/esp-idf/en/latest/esp...
gah. don't tempt me
manic glacierBOT
tulip sleet
@slender iron I have some "which modules on/off" for ESP32-S3 to discuss when you have time. User in forums noticed gifio was not on for ESP32-S3 TFT boards. There are some odd linkages between modules.
is off for space reasons
slender iron
Ok I’ll ping once I’m at my desk. Have a dermatology appointment this morning
Generally 4mb flash boards have less
tulip sleet
yup, i am seeing some better tradeffs, though, maybe
mortal kernel
Bluetooth LE question for Pico W: When setting an address of type BD_ADDR_TYPE_LE_RANDOM I am allowing the caller to set a random address or have a random address generated automatically. The cases are differentiated by the low-order 2 bits of the high-order byte of the caller provided address. If the bits indicate the address is locally administered/individual (L=1, G=0) the user provided address is used; otherwise, a random address that is locally administered/individual is generated and used.
This "feels" right, although I cannot locate a passage in the BT specs that says it is. Any opinions?
tulip sleet
i thought the differentiation would be between a user-supplied address and a generated random address. Does the nRF _bleio handle this?
a user supplied address is not "random", I would think
mortal kernel
if (local_address[0] & 0x02) {
// Set adapter to the caller-provided random address.
gap_random_address_set_mode(GAP_RANDOM_ADDRESS_TYPE_STATIC);
gap_random_address_set(local_address);
} else {
// Set adapter to a randomly generated address.
btstack_crypto_random_t request;
volatile bool crypto_done = false;
btstack_crypto_random_generate(&request, local_address, NUM_BLEIO_ADDRESS_BYTES, &random_addr_ready, (void *)&crypto_done);
// Synchronize with random address generation completion.
while (!crypto_done) {
RUN_BACKGROUND_TASKS;
}
// Set the L bit and clear the G bit (locally administered, individual address)
local_address[0] = (local_address[0] | 0x02) & 0xFE;
gap_random_address_set_mode(GAP_RANDOM_ADDRESS_TYPE_STATIC);
gap_random_address_set(local_address);
}
that's your code or current nrf code?
mortal kernel
Right, arbitrary. It looks like the SD does the work for nRF, common_hal_bleio_adapter_set_address() , all CP does is pass the user provided type and address to sd_ble_gap_addr_set().
What happens from there is opaque and not documented.
tulip sleet
yes, just looking at that
but I don't remember anything about fiddling with those bits to get it to do something else
Are you adding a capability that the nRF impl doesn't have?
mortal kernel
Yes. I'm concerned about presenting a valid address given its type.
mortal kernel
Not really. It's crypto unit can generate random bytes.
In this case I ask it for six bytes.
The Nordic doc is interesting. Are there rules coming from the BT spec or are they Nordic's own?
tulip sleet
I would say that you don't need to do this. It's up to the user. In the nRF world, you can use sd_ble_gap_privacy_set(), which will use a dynamic random address that is refreshed at a specified internval
is there something like that in the CYW API?
mortal kernel
No. The Nordic SD seems to be half a driver on top of the hardware.
tulip sleet
ok, looked more ... we have an anonymous flag in start_advertising(), that will turn on privacy to use a changing really random address. If CYW doesn't have something like sd_ble_gap_privacy_set(), then that needs to be coded explicitly. But I'd check whether there's a CYW advertising parameter that does that, maybe
don't make Adapter.set_address() do something tricky. If you want to reject an address with the wrong bits set, that's fine. CYW may do that. I don't know whether nRF SD checks the validity of those bits .... I'll look
mortal kernel
BTW, your CP code for nRF code looks really good.
tulip sleet
sd_ble_gap_set_addr() checks the top two bits and will complain with BLE_ERROR_GAP_INVALID_BLE_ADDR
thanks, it was inspired by our earlier Arduino BLE code, which was done by some real BLE experts. It took me a long time to wrap my head around BLE, which I feel is not great at clean layering of abstractions
mortal kernel
Good, I'll enforce nRF rules for the high-order address byte.
tulip sleet
the CYW equivalent of set-address may do that check too?
mortal kernel
the CYW equivalent of set-address may do that check too?
No check in BTstack or cyw43-driver that I can find. Caveat emptor code!
So, to summarize: Enforce nRF rules for high-order address byte in common_hal_bleio_adapter_set_address. Also, prohibit setting private address here, private addresses may only be set in start_advertising.
tulip sleet
So, to summarize: Enforce nRF rules for high-order address byte in `common_hal_b...
uint32_t sd_ble_gap_addr_set ( ble_gap_addr_t const * p_addr )
Set the local Bluetooth identity address.The local Bluetooth identity address is the address that identifies this device to other peers.
The address type must be either @ref BLE_GAP_ADDR_TYPE_PUBLIC or @ref BLE_GAP_ADDR_TYPE_RANDOM_STATIC.
mortal kernel
Still, I remain mystified by the high-order address byte.
tulip sleet
Still, I remain mystified by the high-order address byte.
see (old) post here, first reply: https://devzone.nordicsemi.com/f/nordic-q-a/10758/why-is-hardware-address-of-bluetooth-device-different-than-device-address
it points to somewhere in the BT spec (probably obsolete chapter number, etc.)
manic glacierBOT
mortal kernel
Yes! That is the reference I've been seeking. Huzzah!
short tendon
@slender iron 2 things:
- have some branches up and comments for the typing test bug
- also have a question on the ConnectionManager PR about some code that really isn't right (pulled from requests) and trying to see what should be done
mortal kernel
(I'd been extrapolating from IEEE 802-2001, which is how I wound up where I was.)
tulip sleet
i did see that mentioned in a less useful search result
mortal kernel
As always, our chat has been informative and helpful. Thank you, Dan!
tulip sleet
As always, our chat has been informative and helpful. Thank you, Dan!
You're welcome, and thanks so much for working on this!
tidal kiln
is there anyway to skip the 1 second safe mode blinks?
On start up, the LED will blink YELLOW multiple times for 1 second.
for faster boot
stuck elbow
yes, there is a compile time option
slender iron
yup, i am seeing some better tradeffs, though, maybe
I'm at my desk now
slender iron
<@252717193496756235> 2 things:
1. have some branches up and comments for the ty...
I'm here now
slender iron
anything more for end users? like something in boot.py?
It'll be skipped when booting back up from deep sleep
tulip sleet
I'm here now
in a call - will ping
tidal kiln
It'll be skipped when booting back up from deep sleep
thanks. i think they're use case is more simple. if you've got suggestions:
https://forums.adafruit.com/viewtopic.php?p=1002948
manic glacierBOT
slender iron
thanks. i think they're use case is more simple. if you've got suggestions:
<htt...
Seems like they just want to know why. The UF2 bootloader also has a delay for the same reason
that's how the double tap works
tidal kiln
yep, but PWMing in CP is much easier. so if there had been any way to reduce that delay, could've been useful for them.
onyx hinge
"posix" port advances. All "basics" tests now pass locally. ```jepler@bert:~/src/circuitpython/ports/posix$ make test TEST_ARGS="-d basics"
[...]
523 tests performed (16408 individual testcases)
523 tests passed
17 tests skipped: array_micropython builtin_next_arg2 builtin_range_binop class_delattr_setattr class_inplace_op2 class_notimpl fun_globals fun_name generator_name io_buffered_writer io_iobase memoryview_itemsize memoryview_slice_size namedtuple_asdict nanbox_smallint sys_getsizeof sys_tracebacklimit
overall 72 tests failed, 181 tests skipped.
empty salmon
Something has definitely changed WRT CircuitPython and integrated, on-board, ePaper displays - e.g. board.DISPLAY.
I have run several tests and I no longer get runtime error output on the display.
I assume this new behavior is intentional (and if so, thanks 😉)
tulip sleet
@slender iron ok, free now, if you are
slender iron
<@252717193496756235> ok, free now, if you are
just a couple minutes and I'll be ready
idle owl
Is there a place to put a demo that is specific to a particular board, but not a particular library?
onyx hinge
@idle owl good question. I don't think we have anywhere. There's the git repo for learn, of course, when it's related to a learn guide but I assume that's not at all your goal.
It's easy to share random files as github gists or create a repo under your own git user. I kinda like gists because of how they can be embedded on a blog entry with just a couple lines of javascript.
example of a blog post with an embedded github gist (python but not circuitpython): https://emergent.unpythonic.net/01705158260
embedding code: <script src="https://gist.github.com/jepler/ff55a475e9c90c8b1b7f00707799565c.js"></script>
idle owl
<@330227457296957440> good question. I don't think we have anywhere. There's the...
Valid. Thanks. There's an unusual thing about the buttons for this board in CircuitPython, so having a demo available to folks seemed like a good idea. I can host it though.
orchid basinBOT
manic glacierBOT
short tendon
@lone axle willing to test https://github.com/adafruit/circuitpython/pull/8894?
onyx hinge
@short tendon not asking you to change that PR -- you're clearly making a minimum change relative to what was previously there -- but just for your awareness: https://docs.python.org/3/library/importlib.resources.html#importlib.resources.files might be a less finicky way to access .pyi files that are bundled as part of a package. It's in 3.9 and newer (which is in debian oldstable stable and ubuntu LTS 22.04, so even the chronically slow updaters are well covered)
In the unlikely event support for python 3.8 is desired (EOL not until this october) there's https://docs.python.org/3/library/importlib.resources.html#importlib.resources.read_binary but this has been deprecated in 3.11.
short tendon
<@1152454697614069811> not asking you to change that PR -- you're clearly making...
For the circuitpython_setboard one (sorry have a couple in flight)?
onyx hinge
short tendon
yeah that makes sense. What version would you support? Since this is for the common circuit python developer, I could see them having a pretty old verion of Python installed...
mortal kernel
A thought to file away: The Pico W's CYW43439 has a cryptographic unit that may be useful for other CP purposes like TLS handshake.
short tendon
@onyx hinge I like that much better:
from importlib import resources
import os
import sys
import shutil
def set_board():
if len(sys.argv) != 2:
print(f"Incorrect args. Please call with: 'circuitpython_setboard chosen_board'")
return
chosen_board = sys.argv[1]
print(f"setting board: {chosen_board}")
board_stubs_file = resources.files('board-stubs').joinpath("__init__.pyi")
board_definitions_path = resources.files('board_definitions').joinpath(chosen_board)
board_definitions_file = board_definitions_path.joinpath("__init__.pyi")
if not board_definitions_file.is_file():
print(f"Board: '{chosen_board}' was not found")
return
shutil.copyfile(board_definitions_file, board_stubs_file)
Would you recommend this method or supporting 3.8?
onyx hinge
<@426200547024961539> I like that much better:
```
from importlib import resourc...
nah, I doubt supporting 3.8 for new optional features is important. I only mentioned the version 3.9 and newer limitation because i saw it in the docs.
onyx hinge
If so I'll update the PR
sure, though with the caveat that I haven't even touched this new functionality.
short tendon
It's working for me. And I know FoamyGuy knows how to test
manic glacierBOT
short tendon
@slender iron do you have enough info to know of the 4 exceptions currently tracked in the get_socket in requests, which should be retryable? I assume the memory error is, after we release any no longer used sockets. But what of the other 3?
slender iron
I don't know. I'd go based on what minimqtt does
short tendon
K. Will dig in
slender iron
K. Will dig in
Thanks! Seems like the retry code needs a new set of eyes anyway
short tendon
Thanks! Seems like the retry code needs a new set of eyes anyway
Do you know who is most knowledgeable on minimqtt? Would love to chat with them a bit to find common ground
slender iron
I suspect it is brent
short tendon
Ah, that would make sense, since he pinged me on my last change in requests...
slender iron
I'd suggest checking the minimqtt commit history
slender iron
I usually look at the earliest commits
there is a lot of automated tooling and updating that happens as well
short tendon
Is brentru on GitHub the same Brent you were thinking?
desert gust
how functional is the usb_host for CircuitPython 9.0.0-beta.0 on the Feather RP2040 with USB Type A Host? and can the usb host be able to send serial commands to a marlen 3d printer usb controll port?
manic glacierBOT
slender iron
Is `brentru` on GitHub the same Brent you were thinking?
yup
slender iron
how functional is the usb_host for CircuitPython 9.0.0-beta.0 on the Feather RP2...
Yes but you'll need to do the low level endpoint writing
I don't think we have a host cdc library yet
desert gust
I don't think we have a host cdc library yet
what is cdc
because there is usb_cdc for both CircuitPython 8.2.9 and CircuitPython 9.0.0-beta.0 built-in to the circuitPython uf2 installer
tulip sleet
because there is usb_cdc for both CircuitPython 8.2.9 and CircuitPython 9.0.0-be...
That is CDC device support, not host support
CDC is USB serial (like COM ports, /dev/ttyACM0, etc)
desert gust
CDC is USB serial (like COM ports, /dev/ttyACM0, etc)
can cdc work for usb controlled 3D printer aka the Anycubic Kobra Go?
tulip sleet
if it uses USB serial, yes, but, again, we don't have USB CDC host code yet
desert gust
than what is this?
tulip sleet
than what is this?
That is for acting as a USB device, not a USB host. If your printer is a USB device, then those don't help to talk to it. e.g. usb_cdc is for plugging a CircuitPython device into a host computer (Windows, Mac, Linux, etc.) and talking to the host over its serial ports.
desert gust
dose cdc sand for "com device connetion"?
slender iron
what is cdc
cdc is how serial is done over usb. (it's an old modem protocol.) usb_cdc is for device mode. not host
slender iron
And is that `@ brent` here? Still learning who's who....
yup @prime flower
short tendon
Hello @prime flower, I had a quick question on MiniMQTT I was hoping you could answer.
In ._get_connect_socket during sock.connect if it raises a OSError, it sets backoff = True and then delays the next try.
Do you know why? I'm working on the new ConnectionManager and trying to make a unified method for getting sockets and between Requests and MiniMQTT this is the outlier...
desert gust
@slender iron but dose cdc sand for "com device connection"?
prime flower
Hello <@311339734339289090>, I had a quick question on `MiniMQTT` I was hoping y...
This might've been a change made by an external contributor, give me a moment to pull up the history on that file
Seems it was added by this PR, https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT/commit/7cab7d87b7821094ed15fd91c6d8058ff0741263. I understand why. It was added as a way to implement exponential backoffs for reconnections. These are used in order to avoid throttling by a MQTT server for too many retries during the auth process.
short tendon
Seems it was added by this PR, https://github.com/adafruit/Adafruit_CircuitPytho...
Do you feel it's needed in that particular case? When the socket can't even physically connect?
And another question, what reason does socket.connect((host, port)) raise an OSError?
prime flower
Do you feel it's needed in that particular case? When the socket can't even phys...
Don't think so, it should required when the broker sends back an unauthorized message to the client (I think this is a CONACK msg in MQTT spec. parlance)
prime flower
And another question, what reason does `socket.connect((host, port))` raise an `...
Could you link me to the GitHub page with the line number, like https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT/blob/main/adafruit_minimqtt/adafruit_minimqtt.py#L14
short tendon
Could you link me to the GitHub page with the line number, like https://github.c...
Yup it's here: https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT/blob/a05b19f522fae5232f7d92edcd88160c4de4ed28/adafruit_minimqtt/adafruit_minimqtt.py#L365. That's the only one in the get/connect socket flow that isn't a TemporaryError that would then be retried right away
prime flower
Yup it's here: https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT/blob/...
I am not sure why it would raise an OSError
short tendon
I am not sure why it would raise an OSError
Do you think it's safe to not have that one cause the delay? Allowing ConnectionManager to just have it's own retry logic to connect and either raise an exception on return a socket?
manic glacierBOT
short tendon
Am I correct that here https://github.com/adafruit/Adafruit_CircuitPython_Typing/blob/0dc12dfb696a5f8c9174ccda0972404338984091/circuitpython_typing/io.py#L43 the , / shouldn't be there? My PR is failing because:
$ /c/tools/mpy-cross/mpy-cross-8.2.9.exe circuitpython_typing/io.py -o output.mpy
Traceback (most recent call last):
File "circuitpython_typing/io.py", line 43
SyntaxError: invalid syntax
That is a missing feature in micropython/circuitpython. It works ok in CPython. But circuitpython_typing does not need to be importable in circuitpython, right? I'm surprised we're trying to mpy-cross it, but I guess it is like any other library
short tendon
That is a missing feature in micropython/circuitpython. It works ok in CPython. ...
Since it's been there forever, I assume originally it wasn't. What would you like me to do? No PR will work until that's changed...
And no, it's only for type checking your development work...
prime flower
Do you think it's safe to not have that one cause the delay? Allowing `Connectio...
You could try it. I'd remove it first, and see what happens (expected behavior) before re-implementing with this new connection manager
tulip sleet
I think this is a CI problem related to https://github.com/adafruit/Adafruit_CircuitPython_Typing/issues/34
short tendon
You could try it. I'd remove it first, and see what happens (expected behavior) ...
Sounds good. I image a test might fail depending on how in-depth they went when adding them.
tulip sleet
we need to suppress building the .mpy's at all
short tendon
we need to suppress building the .mpy's at all
Is there something I can change in the workflow?
tulip sleet
i'm looking
stoic swallow
Is it possible to get adafruit_board_toolkit.circuitpython_serial.data_comports()[0].product (in this case, M4 express running CP 8 with data serial enabled) to return a custom string?
tulip sleet
Is it possible to get `adafruit_board_toolkit.circuitpython_serial.data_comports...
there is some customization of USB ID strings but I'm not sure if it's this one or not. What is it returning now?
stoic swallow
None
And my ignorant attempt at setting it in boot.py didn't work.
I'm not wedded to product but would like to otherwise identical uCs connected and having a programmatic way of telling them apart.
slender iron
<@252717193496756235> but dose cdc sand for "com device connection"?
Looks like "communications device class" https://en.wikipedia.org/wiki/USB_communications_device_class
tulip sleet
I'm not wedded to `product` but would like to otherwise identical uCs connected ...
there are other names that say "CircuitPython"
if you don't enable the data comport it will not be there.
look at .interface instead
stoic swallow
Yes, and the adafruit_board_toolkit is getting me 90% of the way there. I have a modular robot project where each module is controlled by a uC. I would like a user-friendly way for the host to know what modules are connected.
manic glacierBOT
tulip sleet
the fact that data_comports returns a non-empty tuple at all means it's a CircuitPython device, because it's looking for "CircuitPython CDC2 ..."
in the .interface name
stoic swallow
Right, and if I have 2 M4 express uCs, both with data enabled, connected to host, I should get 2 entries in that tuple. Right now, the only way I can differentiate them is if I have a host-side database that ties something like .serial_number to the function of the device.
tulip sleet
aha, ok I thought this was CPy vs non-CPy boards
tulip sleet
Right, and if I have 2 M4 express uCs, both with data enabled, connected to host...
https://docs.circuitpython.org/en/latest/shared-bindings/supervisor/index.html#supervisor.set_usb_identification Note that this is the whole USB device, not the COM port. There are ways to get the device identification, but adafruit_board_toolkit does not provide them. Take a look at https://github.com/Neradoc/discotool
stoic swallow
Thanks for the refs.
tulip sleet
Thanks for the refs.
makes me think I should add USB device info fetching to board toolkit
stoic swallow
Definitely wasn't intending to make more work for you 🙂
tulip sleet
@short tendon I'm still looking at the workflow stuff too 🙂
short tendon
Thanks!
tulip sleet
@short tendon it does not look like a simple fix for the mpy-cross problem. See https://github.com/adafruit/Adafruit_CircuitPython_Typing/issues/36 which I just opened. So how about removing the , / in that signature for now? Maybe comment it out with a TODO to put it back when the CI workflows can not do the mpy-cross build.
short tendon
<@1152454697614069811> it does not look like a simple fix for the mpy-cross prob...
Can do. Am I good to reference the issue in the TODO?
tulip sleet
oh sure!
ornate breach
I noticed that MicroPython has a zephyr port for zephyr compatible boards. Would that be a reasonable template to try circuitpython with zephyr?
slender iron
I noticed that MicroPython has a zephyr port for zephyr compatible boards. Would...
Ya, though I think they are still changing it. What board are you wanting to run CP on?
ornate breach
Ya, though I think they are still changing it. What board are you wanting to run...
I was thinking an nRF board
Something simple like an nRF52840
tulip sleet
I noticed that MicroPython has a zephyr port for zephyr compatible boards. Would...
were you prompted by my discussion with JLA in #general-tech ?
ornate breach
were you prompted by my discussion with JLA in <#327254708534116352> ?
Yeah,I was curious what the path to use zephyr looked like.
manic glacierBOT
ornate breach
Just a quick glance I understand that it’s going to be specific to zephyr supported boards, likely at this point being ones that fit the cross section of circuitpython and zephyr
tulip sleet
Just a quick glance I understand that it’s going to be specific to zephyr suppor...
See https://github.com/adafruit/circuitpython/issues/8135 if you have not already.
ornate breach
See <https://github.com/adafruit/circuitpython/issues/8135> if you have not alre...
Seems like the main zephyr issues in waiting have closed. I’m not confident that indicates progress can go forward but perhaps it does 🤷♂️
lone axle
let me know when yr around
I am now 👋
meager fog
@lone axle ok plz log into adafruit account, get cart with anything you want ready, adn go to the memtno page
ill relase one of our sekret stash
lone axle
<@382939733107408897> ok plz log into adafruit account, get cart with anything y...
okay I am logged in, cart ready and on momento product page
meager fog
@lone axle GO GO GO
lone axle
<@382939733107408897> GO GO GO
okay got one in cart, checking out
meager fog
@lone axle ok i restashed 🙂
lone axle
checkout finished. Thank you!
meager fog
ok i tagged ya on github with the idea
short tendon
@slender iron thank you for pinging me on that type issue with test-stubs.sh. That lead me down quite a handful of rabbit holes, but also helped me discover some other parts of the infra, which let me kill 933 bytes out of ConnectionManager AND remove all the types from the docs.
short tendon
You could try it. I'd remove it first, and see what happens (expected behavior) ...
Not seeing any issues. I needed to update one test. I'll check a few things and open a PR to MiniMQTT with the changes so others can look
slender iron
<@252717193496756235> thank you for pinging me on that type issue with `test-stu...
Thanks for jumping into the rabbit hole. 🙂
short tendon
Quite welcome. In the end it was net positive for me.
I also feel I have a path forward on ConnetionManager for retries. Will be similar as I work through some PRs in other places first
modern star
Hello i just recently got a adafurit rp2040 for using it for my macros like opening up google and discord and ect.. but i am having trouble putting the code on there. There isnt good vids or guide to help do that i am using Mu to edit my code
mortal kernel
@danh or @slender iron Similar to cyw43-driver, I need an Adafruit fork of BTstack that I can use for patches awaiting incorporation upstream. The specific case I have with BTstack is a pair of missing function prototypes that prevent extended advertising from building.
willow totem
@prime flower / all - finally tracked down the page on pin config in UF2 files, that I saw in some MakeCode article, called Binary Bootloader-embedded Config (CF2):
https://github.com/Microsoft/uf2/blob/master/cf2.md
Basically adjust pin numbering/peripherals if same mcu without adjusting main code, just patch the uf2 CF2 section.
Also online patcher / example config https://microsoft.github.io/uf2/patcher/
example keys for CF2 (from patcher site)
tulip sleet
@danh or <@252717193496756235> Similar to `cyw43-driver`, I need an Adafruit for...
which repo do you need us to fork?
lone sandalBOT
mortal kernel
which repo do you need us to fork?
bluekitchen/btstack
tulip sleet
`bluekitchen/btstack`
The licensing of btstack is a bit tricky. Raspberry Pi got a supplemental license for Pico W: https://github.com/raspberrypi/pico-sdk/blob/master/src/rp2_common/pico_btstack/LICENSE.RP. I It may be OK. But I think I will check with management about our forking of it before we do that. As I understand you are just getting around some compiler warnings or lack of protoype declarations? Could you add those by hand somewhere or turn off some warnings for certain source files?
mortal kernel
The licensing of btstack is a bit tricky. Raspberry Pi got a supplemental licens...
I'll see what I can do. Jamming the prototypes into my btstack_config.h will be the easiest. The affected module hci.c is several thousand lines of code, so I hesitate to disable warnings. The functions in question should be static, the prototypes are only needed since they're not. The licensing of BTstack has worried me, too. Perhaps we should think about finding an alternative protocol stack?
Also, I'll likely find more breakage in BTstack as I progress.
manic glacierBOT
mortal kernel
The licensing of btstack is a bit tricky. Raspberry Pi got a supplemental licens...
It's not at all elegant, but I was able to work around the problem by jamming the prototypes into btstack_config.h so we can put the issue of forking BTstack to the side for now.
// The following prototypes work around missing prototypes in btstack/src/hci.c:
#include <stdint.h>
uint8_t hci_le_extended_advertising_operation_for_chunk(uint16_t pos, uint16_t len);
void le_handle_extended_advertisement_report(uint8_t *packet, uint16_t size);
solar whale
Just a heads up -- I don't have time to file an issue now, but since CP9 beta.0 something has broken adafruit requests. With recent builds call to get an AIO feed reult in ans "Out of Retries" error in adafruit_requests lin 696. I'll create a proper issue later or more likely tomorrow but I was wondering if anyone ese was seeing issues in the last few days.
short tendon
Just a heads up -- I don't have time to file an issue now, but since CP9 beta.0 ...
I've seen a bunch of SSL Handshake issues on ESP32S2s on CP9. Is that what you are using?
solar whale
I've seen a bunch of SSL Handshake issues on ESP32S2s on CP9. Is that what you a...
I’ve been using esp32s3s since I noticed this.
On latest CP9
Works fine with Beta.0…
short tendon
On latest CP9
What's the actual error you're getting?
The one hidden behind the max retries
solar whale
I’ll post more details later. Tied up now.
orchid basinBOT
tulip sleet
Just a heads up -- I don't have time to file an issue now, but since CP9 beta.0 ...
If you roll back to the last alpha does it work with the same version of the library? I would also check to see whether there was a library regresion.
solar whale
If you roll back to the last alpha does it work with the same version of the lib...
I started going back from the latest and it was failing but I ran out of time. It works with Beta.0 -- not usre when after taht it quit -- I will investigate it tomorrow. Here is the full traceback ```Adafruit CircuitPython 9.0.0-beta.0-23-g8af780c7fa-dirty on 2024-02-11; ESP32-S3-EYE with ESP32S3
import aio
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "aio.py", line 29, in <module>
File "adafruit_io/adafruit_io.py", line 748, in get_feed
File "adafruit_io/adafruit_io.py", line 578, in _get
File "adafruit_requests.py", line 732, in get
File "adafruit_requests.py", line 696, in request
OutOfRetries: Repeated socket failures