#circuitpython-dev
manic glacierBOT
willow totem
I'd like to propose a docs change, or a functionality change. The docs currently say supervisor.reload performs the equivalent of ctrl+d. It's right above supervisor.set_next_code_file which is what I wanted. I read that and then ran something to set code2.py next. Then I typed supervisor.reload in the repl and nothing happened. Can we add to the supervisor.reload section (or maybe as a last point to set_next_code_file) that supervisor.reload doesnt work from the repl and you have to use ctrl+d.
Instead of the docs change can I suggest supervisor.reload could be made to work from the REPL too. It was explained to me there are possibly good exception catching reasons that are useful to the REPL that maybe prevent it working, so path of least resistence is probably a docs warning that could prevent other tired/confused users from losing time over it.
tulip sleet
I'd like to propose a docs change, or a functionality change. The docs currently...
It's certainly confusing! Could you open an issue? Then it gets on the to-do list. You could submit a doc PR too, if you feel up to it.
lone sandal
manic glacierBOT
manic glacierBOT
thorny jay
I just received the Python on Hardware mailing: "Python on Microcontrollers Newsletter: CircuitPython 8.2.0 Stable Released" but that is like one day too early. Is there something special, or it is the 8.2.0 special edition?
random junco
I just received the Python on Hardware mailing: "Python on Microcontrollers News...
Yup, it's a day early on purpose. Enjoy! 🙂
thorny jay
I was ready to announce you all and Anne about my new project: https://github.com/dglaude/Mouse_Jiggler_Trinket_M0/tree/main
Given the plan for this evening (I am in Europe), I don't even have time to finish or fill the meeting notes.
But if anybody want to have a sneak preview or has comment on the code, I can improve that for next newsletter and do my usual Mastodon / Twitter post about it.
It is the ultimate M0 based mouse jiggler.
random junco
I was ready to announce you all and Anne about my new project: https://github.co...
I'm happy to include it in next week's newsletter!
lone axle
<@&356864093652516868> We look forward to hearing from everyone during the weekly meeting after the week off last week. The meeting will be held here on discord in the CircuitPython voice channel at the usual time of 2pm Eastern / 11am Pacific which is just over 1 hour from now. You can fill in any notes you have to the shared document: https://docs.google.com/document/d/1ajdwXoubhG86rUVabqTcEzHvOKfUgK9TgPdDxqBDlSc/edit?usp=sharing. See you there.
onyx hinge
@candid sun btw there is a way to do "more natural" (and more efficient) asyncio on keypad events but it needs a weird block of code.. https://github.com/adafruit/circuitpython/pull/6712
The weird block of code has never been put in a library/module/package or anything
If you are interested in this and want to learn more feel free to ask, but you don't necessarily need to do it that way.
candid sun
cool, thanks @onyx hinge !
onyx hinge
is there still time for a mic check? I'm on a different computer.
idle owl
@onyx hinge I think you had an echo, and you're a bit loud.
onyx hinge
I tried turning on echo cancellation, hopefully it's still good
idle owl
Not a preview for this week's newsletter and the next two issues! (They will go out at 11am on Monday.)
idle owl
Hooray! 😊 🎉
slender iron
gotta start somewhere
lone axle
I started a basic input text for displayio at one point. But realized how complicated cursors really are once I got into it. Mine ended up very wonky.
ember iris
I have two SCD41's, and one SCD40
onyx hinge
thanks all!
gilded cradle
Thanks
idle owl
Thanks, everyone!
ember iris
Thanks all! Have a good week!
slender iron
I wonder if new boards are enough of a reason to do 9.0 off main
lone axle
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/10AzCwaac8jCgfUiFIPchwTghFqHptCV99iMCsZfxAwo/edit?usp=sharing
slender iron
are we ready for an 8.2.x milestone? seems like a couple of these new issue could go there
onyx hinge
@slender iron seems fine by me!
tulip sleet
are we ready for an 8.2.x milestone? seems like a couple of these new issue coul...
sure - i wasn't sure, but I see several
I'll make one
slender iron
thanks!
mental nexus
I started a basic input text for displayio at one point. But realized how compli...
Have been text only watching the USB host work and drooling at the new possibilities. As for cursors for input, I made one for the pye editor for the cyberDUCK but just made the letter a simple inverse. Never tried for a real flashing “insert here” type of cursor. https://hackaday.io/project/171269-cyberdck-a-circuitpython-anatidae
slender iron
@mental nexus how did you handle switching between running code and editing it?
onyx hinge
ooh pye doesn't look too shabby
@mental nexus how'd your cursor work? A non-flashing block cursor would be fine as a starting point for me.
I guess that project's not displayio based is it
onyx hinge
…/capablerobot_usbhub/mpconfigboard.mk:EXTERNAL_FLASH_DEVICES = "GD25Q16C, W25Q16JVxQ"
…/stm32f411ce_blackpill_with_flash/mpconfigboard.mk:EXTERNAL_FLASH_DEVICES = GD25Q16C,W25Q64FV,W25Q32JVxQ,W25Q64JVxQ
```It's probably not HURTING anything that we use all these syntaxes, assuming they work ...¯_(ツ)_/¯
@onyx hinge any thoughts on how to handle a usb_host.Port living longer than a VM?
with displays we require you to release them
I wonder if we should just automatically release at next use
or something like that
onyx hinge
can they just be created each time the VM starts?
slender iron
they could but it throws an in-use error now
mental nexus
<@366059101203202052> how did you handle switching between running code and edit...
This is ancient history so just relying on memory. I started everything from the REPL. Just exec the editor to edit my file and then exec the edited file.
onyx hinge
I have had some Ideas about replacing supervisor allocations with objects that can live through VM resets, and maybe can re-locate/compcat when the VM is reset, but it is in the "this is an idea" stage rather than something that's implemented.
onyx hinge
and then of course there are the way that board.SPI / board.I2C work, would that model fit at all?
slender iron
if the board has usb host natively, it can do it in board_init()
onyx hinge
since I've concentrated on the host keyboard stuff I haven't actually used usb_host.Port
mental nexus
I guess that project's not displayio based is it
I made a terminal out of displayio. Can’t remember how I did the cursor will have to look at my code to remember. https://github.com/kmatch98/simpleTerminal
slender iron
@onyx hinge you could have a cursor tilegrid that you blink on top of the regular terminal
mental nexus
Yep looks like I just have a white rectangle behind a black character.
manic glacierBOT
manic glacierBOT
onyx hinge
@slender iron a different, wired keyboard was still working (accepting keystrokes) after running overnight.
supple gale
That makes sense. ‘’’ Note
Personal Hotspot uses a 5 GHz connection by default. On iPhone 12 or later, you can turn on Maximize Compatibility for Personal Hotspot to use a 2.4 GHz connection. Turning on Maximize Compatibility might reduce internet performance and Wi-Fi security for devices connected to the hotspot. For more information, refer to If Personal Hotspot is not working on your iPhone or iPad (Wi-Fi + Cellular).
onyx hinge
@slender iron I noticed that e.g., the "product" string was shown with a trailing NUL '\0'. fix at https://gist.github.com/jepler/6db3b04596ce9dd7c6c0fa5482439b3c
do you have a philosophy about whether supporting e.g. USB MIDI would be done using pure Python code, or if it'd have C code in the core?
onyx hinge
@slender iron I tried to use my USB MIDI device from CP but didn't have success.
Here's the code I can run on my host computer: py import array from usb.core import find, USBTimeoutError dev = find(idVendor=0x1c75) b = array.array('B', [0] * 8) while True: try: r = dev.read(0x81, b, 10000) print(r, b) except USBTimeoutError: print(end="...") (I have to run it as root and also "unbind" the kernel driver in the linux way: root@bert:/sys/bus/usb/drivers/snd-usb-audio# echo 1-1:1.2 > unbind)
On CP with rp2040 usb host feather, it just hangs at the dev.read call.
On Linux it prints something whenever I press a key on the MIDI keyboard, turn a knob, etc.
slender iron
<@252717193496756235> I tried to use my USB MIDI device from CP but didn't have ...
Good to try! I was running into an issue with tinyusb not calling us back.
manic glacierBOT
slender iron
@onyx hinge what device is it exactly?
onyx hinge
@slender iron it's an Arturia minilab 3 midi keyboard, it was recommended by todbot and jp for synth stuff
slender iron
do you have a philosophy about whether supporting e.g. USB MIDI would be done us...
I was thinking it'd be done from Python. I'd only do "workflow" related things natively
👍 I have some midi keyboard too I think
onyx hinge
I'm about to head out to help a friend, so itmya be the afternoon before I'm back
and when I come back I want to test the rp2040 metro prototype, probably won't get back to USB immediately
slender iron
I just got back from the daycare run so I'm just getting going
onyx hinge
we can talk more later, whether today or later in the week!
tulip sleet
any objection to me merging in the current usb host code?
Nope!
manic glacierBOT
orchid basin
warm stump
The USB-C port on my Macropad just fell off…it had a little bit of a rough time in the theatre.
All is good though, as I repaired it as soon as I got home with some lollipop sticks.
It needed a quick reminder of all the sketches that were on it but it’s running again
meanwhile in the theatre…a splash screen?
onyx hinge
@slender iron using code similar to what I pasted?
slender iron
yup. it is an arturia keystep
(I removed the time out)
lemme see how many changes I have on my branch
onyx hinge
On branch usb_host_keyboard
Your branch is ahead of 'jepler/imx_usb_host' by 1 commit.
(use "git push" to publish your local commits)
I have a commit that fixes the trailing '\0' in the device strings, that's all. I made sure submodules were up to date.
I was entering the code in the REPL since it tended to freeze the device and be difficult to recover from
onyx hinge
I also don't know the correct way to find the endpoint number..
slender iron
it involves fetching the descriptors and parsing them
it should be the same as it is in linux
onyx hinge
yeah, lsusb -vv is how I found the endpoint number in linux. but what I meant was, I don't know your plan for how CP code would discover that information..
anyway I'll try that code tomorrow in code.py and see if it makes a difference. otherwise, 🤷 , I can just move on to trying something else I guess
slender iron
yeah, lsusb -vv is how I found the endpoint number in linux. but what I meant wa...
cp code can fetch and parse the descriptors
or you hard code it
here is my branch: https://github.com/tannewt/circuitpython/tree/usb_sharing
tulip sleet
@onyx hinge do you remember if we adopted some qstr changes from MicroPython early? I am seeing some things in the 1.19.1 merge that seem to be already merged, but the three-way merge is also showing "old" code that is not in the current adafruit/circuitpython. I looked in our PR's but could not find anything. There is a lot of restructuring of qstrs and their generation in this merge (separating out the hashes and length, for instance).
the last stuff is not our code. It's some other stuff that seems to be already incorporated
maybe this is just some peculiarity of three-way merge going back in time because there's a merge conflict
onyx hinge
@tulip sleet no that doesn't ring a bell right now
tulip sleet
thanks anyway
proven garnet
Looking at patching up the Learn Guide repo for a few things, among them the releases. I happened to notice that releases aren't being created for it, and it lines up with with the permissions update you did @tulip sleet for preventing non-PR pushes. Can we make adafruit-adabot have Maintain status for that repo? That should fix it I think.
manic glacierBOT
tulip sleet
Looking at patching up the Learn Guide repo for a few things, among them the rel...
Thanks for spotting! I switched adafruit-adabot to maintain. If necessary, I can turn off the branch protection rule again.
proven garnet
Thanks for spotting! I switched adafruit-adabot to maintain. If necessary, I can...
Thank you! I'll keep an eye out for things over the coming days 😁
candid sun
<@347203168121126912> btw there is a way to do "more natural" (and more efficien...
hihi @onyx hinge looking into this and hitting a wall with keypad and seesaw. it doesn't seem like there's a way to pass seesaw pins like GPIO pins to keypad? there is a seesaw keypad class (https://docs.circuitpython.org/projects/seesaw/en/latest/api.html#adafruit_seesaw.keypad.Keypad) but i think it has to be part of the device firmware, which the quad rotary encoder doesn't have or at least i wasn't able to get it working
onyx hinge
@candid sun the idea was that the interrupt pins, which would be a Pin object from the board module, are what you'd be giving to the Keypad constructor .. maybe I don't know how the interrupt works, though. If it's not working, maybe better to move on than try to make it fit my half-baked ideas.
not the "press in to click" functionality
I don't have any of those boards so I can't do any testing/playing of my own
candid sun
oh i see, i'll try the interrupts. thanks!
onyx hinge
don't spend too much time on doing it "my way", if you see other ways of getting to your goals
candid sun
right now i kind of feel like my brain is getting turned inside out trying to understand asyncio so i might also need a lunch break haha
onyx hinge
that state does wear off .. eventually. and I still never reach for asyncio "by default", so in that sense I've never gotten super comfortable with it.
tulip sleet
right now i kind of feel like my brain is getting turned inside out trying to un...
any suggestions for asyncio guide improvements welcome
tulip sleet
@slender iron @onyx hinge I am almost finished with the initial v1.19.1 file merging. I haven't even attempted a compile yet. This has been frustrating because some merges, especially in py/, especially the three-way merges, show many changes whose origin is obscure:
- who deleted these lines? git blame is very bad at helping you find this out.
- was this change something we did, diverging from the original, or does it just reflect some change we haven't picked up yet?
- "SPDX" changes that cause a merge conflict but are otherwise not useful. MPy has never warmed to the idea of SPDX (I looked at some closed PR's), and I'm not sure we should bother doing this in files we shared with them. It's not completely done anyway, so automated checking is not possible. I may change these back to minimize the churn in future merges.
- We really need to PR back simple changes that MPy would probably be fine with, such as spelling errors, and things like
inline-> ``MP_INLINE`. - because of 1 and 2 especially above, I am considering that we might label our changes with at least
// CIRCUITPY(possibly bracketed with// CIRCUITPY endor similar and making some further comments as necessary. This makes it clear where the origin of the change came from. I am constantly guessing, relying on imperfect memory, or plowing through github blame to figure these out, and I have to keep repeating that. I think this is really only mostly necessary for core stuff inpy/where we have diverged and may never converge again. It's a waste to time to keep rediscovering who made a change. - I think I will do another difftool merge between the current merge and pristine
maintip to handle 3 and 5 before attempting to compile.
onyx hinge
sounds like it's been a real slog. I especially agree with trying to upstream more stuff.
slender iron
we could upstream stuff until we can submodule mp for the vm
onyx hinge
we could upstream stuff until we can submodule mp for the vm
that would be quite an achievement.
slender iron
any more than doing three merges?
tulip sleet
i don't know if we will ever get to the submoduling possibility. there are some changes we've made they won't want back. I think since the last merge we have made more significant py/ changes than usual
The current divergences that I'm not sure about are the space squeezing you did, Jeff, in the type tables, traceback improvements, some different handling of properties, long-lived, some generator changes that may be partly related to async differences, and maybe a few other things. I still have to revisit objgenerator.c because it is so different. That is the last file that is not really merged.
also not relying on extmod/ at all would be great: there was a lot of churn in os and sys this merge. Of course that means we also can't "just merge" the imrpvoements from upstream
also there were a number of specious merges, where some random file that vaguely resembled our file was merged, despite not even being in the same directory. This may hav been due to a file move that git somehow was aware of, but not all seem like that. I don't remember specious merges like these from previous merges
onyx hinge
long-lived allocators & the stricter generator/async distinction are two core differences I don't think we'd ever sell them
onyx hinge
I've always wanted to try this on a "real situation" but ... it looks like it'll perform about 600,000 individual merge attempts in order to reconcile us with micropython 1.19 !? https://github.com/mhagger/git-imerge
tulip sleet
hunh! I had not heard of this.
meld is my friend 🙂
but still hard to track things. I have clicked on arrows a few thousand times, but nothing like this
onyx hinge
this doesn't replace the "mergetool", it just a way of scripting the finding of "smallest possible conflicts" by going along the changes on either side of the merge
there are apparently ~600 merge commits on the micropython side since our last merge and ~1000 merge commits on our side, and it wants to try every pair and have you resolve each conflict that arises. It appears not to scale well.
tulip sleet
got it, just not sure whether this would help. I'm confused about whether it shows intermediate merges or not
onyx hinge
it does but almost each micropython change has some conflict, of a "we deleted, they changed" type.
tulip sleet
what I would really like is to be able to click on a conflict and get directly to the commits that caused the conflict
onyx hinge
CONFLICT (modify/delete): ports/rp2/machine_i2c.c deleted in HEAD and modified in 7b0a42374e1ae6a4ab3f756090c6f3b66e82acb5. Version 7b0a42374e1ae6a4ab3f756090c6f3b66e82acb5 of ports/rp2/machine_i2c.c left in tree.
Automatic merge failed; fix conflicts and then commit the result.
Original first commit:
commit 01cabb0324186808c14ec1b02fb8a3be28837160
Merge: ddfcfca65d da4b38e756
Author: Jeff Epler <jepler@gmail.com>
Date: Tue Feb 15 12:36:26 2022 -0600
Merge tag 'v1.18'
[long log deleted]
Original second commit:
commit 7b0a42374e1ae6a4ab3f756090c6f3b66e82acb5
Author: Damien George <damien@micropython.org>
Date: Thu Jan 20 16:43:55 2022 +1100
rp2/machine_i2c: Provide more specific error codes from I2C transfer.
for instance it'll stop like this, after trying ~1000 individual merges .. After accepting the deletion with git mergetool, it starts again with the next commit from micropython, which will also have a conflict 😕
it's a cute idea but I don't think it works out with CP's git history
mmmmaybe if it was modified to run scott's conflict-fixing script and only treat the merge as failed if there's still a conflict
tulip sleet
I don't think that more automation is going to help me. It is mostly more information I need. Then I need to study the breaking changes and figure out how to incorporate them into code we have already changed.
onyx hinge
in principle this tool tells you what you want: this conflict is caused by THIS commit on their side and THIS commit on our side... but the reality is different
tulip sleet
yes, I would live a GUI difftool that does that.
manic glacierBOT
manic glacierBOT
rancid pilot
Quick question: is there a way to expand the "default" ringbuffer? It can be found at "/py/ringbuf.h". It stores type uint8_t.
So far I made my own but it might not be needed?
Is it okay to add a second ringbuffer impl?
spare jacinth
That is not storing a u8, but a pointer to u8, namely an array with many u8's allocated. Based on the other fields being u32's, its likely to have a big size (or at least being able to have it)
What's exactly your intention/change with the custom implementation?
rancid pilot
I would like to store canio_message_obj_t.
CAN.c creates an rx fifo via the FlexCan SDK. It automatically uses message buffers and can id filters and gives me back a FlexCan frame in a callback (interrupt) where I convert the frames into canio_message and place them in a ringbuffer. Then the Listener.c in common_hal_canio_listener_receive would just grab the next element from this ringbuffer.
The problem I would like to solve is that the SDK expects to process the frame/message in the callback, but canio interface uses polling to consume the messages with the Listener method(s).
manic glacierBOT
spice gate
hello, I'm having a hard time making CircuitPython run on a custom board based on the nRF52840 (MicroPython works without problems though, so it's not something hardware-related). CircuitPython boots but never gets to the USB CDC serial console. I've tried both with internal and external flash storage, with or without HF/LF crystals. It seems to hang the moment it gets to setting up the filesystem. Any ideas on what to try next ?
(this is 8.2.0, up-to-date build system on OSX based on ARM GCC toolchain , Adafruit Bootloader running the show and working just fine )
manic glacierBOT
midnight ember
I wrote a display driver for a non-adafruit st7796 display for circuit python. #help-with-circuitpython message
Based on the adafruit_st7789 circuit python driver. It does not exist in the adafruit 8x or community bundle.
Currently named adafruit_st7796, should i name it something else to submit to the community bundle?
or submit to the circuit python library as adafruit_st7796?
It's my first display driver so unsure how to proceed. It's not an adafruit display it's something I got from aliexpress.
So maybe circuitpython_st7796 instead and submit to community bundle?
idle owl
Currently named adafruit_st7796, should i name it something else to submit to th...
CircuitPython_ should be the prefix. Avoid including Adafruit as it is not an official Adafruit-supported library.
midnight ember
Should I create my own CircuitPython_ST7796 repo first? I'm looking at an example for the community library using the GC9A01 here https://github.com/tylercrumpton/CircuitPython_GC9A01/tree/e31bc61f990f4d52b3b6b1a90eda3090d71f100a
Use that format structure?
Do I need to install cookie cutter? This is becoming more complex than I thought.
Is this only for contributing to adafruit repo or also the community bundle? https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/creating-a-library
idle owl
Do I need to install cookie cutter? This is becoming more complex than I though...
Using cookiecutter is ideal, yes.
idle owl
Is this only for contributing to adafruit repo or also the community bundle? htt...
This guide kind of applies across the board, Adafruit and Community. There are obviously differences in what you would do for a Community library, but I think it's meant to work for both. I know cookiecutter certainly has a Community library setting when you run it.
midnight ember
I just copy/pasted from the GC9A01 example. It's not from cookie cutter but looks relatively enough the same.
idle owl
@onyx hinge Do you have a moment for a question?
onyx hinge
sure
onyx hinge
probably intended to convert from seconds to milliseconds
idle owl
Ohhhh.
onyx hinge
but of course ping can return None ... in which case that's not so great
idle owl
Right.
It's fine, it prints "none" I think
IS ping seriously returning seconds by default?
I thought it was ms?
See: "Ping google.com: None ms"
is what printed this last run.
I can add code to make it say it failed.
This page and code is outdated. I'm updating it. https://learn.adafruit.com/adafruit-metro-esp32-s2/circuitpython-internet-test
Not certain how complicated I want to get here.
midnight ember
If I don't use cookie cutter will circup still work with it? I don't use it just thinking about others.
I'll try to sit down and do a real cookie cutter version of it someday. The way it's formatted right now doesn't play nice with windows Git desktop application. Git says it's can't find my repo, I think might have something to do with the copy/pasted .git ignores.
Is there a learn guide on how to submit a new library to the community bundle?
idle owl
Is there a learn guide on how to submit a new library to the community bundle?
The guide you linked earlier covers "Sharing in a Bundle" or something close to that.
If any of the folks on the CircuitPython review team has a moment, I would appreciate a review on this PR: https://github.com/adafruit/Adafruit_Learning_System_Guides/pull/2561 It's mostly formatting changes, but also updating the file to use settings.toml and adding a settings.toml starter.
rough folio
@plucky tulip
Figured I should put thoughts on https://github.com/adafruit/Adafruit_CircuitPython_HTTPServer/pull/63 here and not in the help channel
Feels like this should definitely be two PRs, or a stacked PR though. Websockets are a pretty distinct functionality
But just addressing the routes aspect-
The key thing missing here is a way to get state in the route. For example-
import routes
something = Something()
server = Server()
server.route(...)
def handle(...)
something.do_thing()
server.add_routes([routes.otherroute])
def other_route(request):
<no way to access `something` or other objects not in scope>
otherroute=Route(..., other_route)
I also don't see that there's a clean way to define routes here either
like the ideal setup would be a route decorator that I can use in other modules
so the above example would be
@route(...)
def other_route(request):
FYI jimmo said on the MP discord they hope to release 1.21 in the next week
rough folio
the way I implemented my hacky version of external routes is here- https://mystb.in/HitsTucsonSatisfactory
the key thing is when calling the route it passes the server object not just the request
which allows me an injectable state
to refrain from a breaking change the state could be added to the request object or passed as a kwarg
also added some syntaxy sugary decos like @get("/path")
also instead of putting closes everywhere I bet you could change poll's context manager close so it figures out if it's doing a websocket and shouldn't close
still zephyr
If I don't use cookie cutter will circup still work with it? I don't use it jus...
Circup will work if the library is in the community bundle and it can build. This is because circup could use either mpy or py files.. One step that is needed is to add the library as a submodule to the .gitmodules following the instructions in the mentioned learn_guide.
midnight ember
Yeah just read that part. A lot more involved to contribute to the community bundle than I realized. Everything has to be done with CLI git really. Can't use desktop git for something like this.
Contributing adafruit_requests examples is a cakewalk compared to contributing a new library.
It's more like creating a custom build of circuit python, which I've done once. There's a lot of moving parts and hoops to jump through.
still zephyr
yes is a little different, but cookiecutter helps with all the behind the scenes, and if you follow the learn guide it will be ok. if you need help let me know :).
midnight ember
thank you, i don't have time to do this today. will likely give it a shot tomorrow because i do want to contribute my first real display driver.
still zephyr
If you dont want to use cookiecutter coy the .github/workfloes directory
Yeah ill be around 🙂 ping me when needed or DM if you want
slender iron
Gnu unifont emoji. Larger font 8x16 instead of 6x12
onyx hinge
emoji is kinda neat. are you hatching a plan to fix the horizontal space between the letters?
solar whale
I've run into a funky issue with the RP2040 RFM9x board and I'm wondering where to open an issue (with CP or the RFM9x library) I am experiencing USB disconnects when transmitting (especially at high power). On my MAC, the MAC will often freeze until I disconnect the board from USB, then it recovers. On a Raspberry Pi, l Ijust see brief disconnects on the USB (via tio)... Has anyone else seen anything like this? Any suggestion as to where to open an issue?
solar whale
This may well be a power supply issue... much more testing to do. I have never seen i with other RFM boards though. When I have a clean example, I'll post it in #help-with-radio -- already started a discussion there.
slender iron
emoji is kinda neat. are you hatching a plan to fix the horizontal space between...
Brainstorming yes. Debating how to store whether a unicode char is double width. May make it a binary search rather than linear unicode search.
we'll use more flash space in small builds to store the taller font
(and wider)
also brainstorming colors
less seriously
the monochrome terminal could be more endearing
onyx hinge
right now my big wants are:
- I don't think it's possible to print a character in the lower right of the screen and not scroll
- block cursor
- reverse video would be nice, and provides a way to do block cursor
Full 16bg + 16fg colors is way after that, but those three items would really enhance the usability of the screen-based editor
struggles to get discord to render a list of points
slender iron
what do you mean by reverse video?
idle owl
Am I seeing correctly that the settings.toml provided by CircuitPython on a fresh install is completely empty?
For some reason I thought it had starter boilerplate.
@slender iron Is that expected, or is there something odd about the install I did a minute ago?
slender iron
I think it is empty normally
slender iron
turns out I broke safe mode entry on rp2040 with my picodvi work
idle owl
OMG.
slender iron
(via reset pin)
idle owl
I have been fighting with that for so long..........
I thought it was me. Dan kept telling me it was possible, but I couldn't get it to work lately.....
I have been flash_nuking all of my RP2040s as I keep messing up the disabling of CIRCUITPY.
slender iron
sorry!
idle owl
Nah, no worries, I'm simply hugely relieved over here to know it wasn't a me-problem. 😄
slender iron
feels an 8.2.1 coming on
pallid creek
right now my big wants are:
* I don't think it's possible to print a character i...
for reverse video, do you mean like ANSI terminal codes that can have different colors for bg vs text?
plucky tulip
<@699200371909853216>
Figured I should put thoughts on https://github.com/adafr...
I see what you mean, but in my opinion accessing variables that are not in scope is not really a good design practice, it would be better to have a server initialized in separate file, then importing it in code.py and routes.py. This way we can keep everything separated, while also giving access to the server for routes that are in other files, which is pretty much exactly the behaviour of decorator that you can use in other modules.
When it comes to passing server object and not only request, I am not sure whether I understand correctly, but you can access it now by using request.server. https://docs.circuitpython.org/projects/httpserver/en/latest/api.html#adafruit_httpserver.request.Request.server
rough folio
Didn't know that request.server existed whoops
That said I have no idea what you're saying in the first paragraph
plucky tulip
@rough folio Something like that, probably with more finesse, but it is just a proof of concept.
server.py
from adafruit_httpserver import Server
# connecting to wifi etc...
server = Server()
from server import server
something = Something()
@server.route(...)
def handler(...):
...
from server import server
import routes
server.serve_forever()
EDIT1: wow, Discords support markdown headers, didn't know about that
onyx hinge
@slender iron I mean that colors 0/1 would be swapped. It's a common way that cursors were implemented on old platforms, by setting an otherwise unused bit in the display RAM. Sometimes it was the high bit of the character (ASCII was assumed in those days) or a separate attribute bit.
and yeah you'd trigger it by sending a VT escape sequence
the 2nd row of text here has some "reverse video"
slender iron
tile grid can't do that unfortunately
probably simpler to have terminal manage a cursor tilegrid and move it around
onyx hinge
yeah that's another approach that would require less change in the core.
slender iron
otherwise you'd need pallete per tile indirect I think
onyx hinge
Or a pallette offset
rough folio
<@129855424198475776> Something like that, probably with more finesse, but it is...
I'd call this pretty odd honestly
like, every lightweight api framework out there I know of (flask (through blueprints), litestar, starlette, fastapi) doesn't do this but instead opts for the "define route object, import, add to server object" method. I've got no idea how this would play with lazy importing either- and it would preclude or make fairly difficult defining the same route for multiple server objects but IDK how realistic that need is. Oh, and adding routes dynamically would be a fair bit trickier too come to think.
Though my initial example missed a detail of adding something to server which I now realize would be injected in the request so you could do request.server.something in the route handler which I think is what you were talking about with variables-out-of-scope
For a direct example with the hack I did (pardon the lazy form parser...)
@post('/pid-update')
def pid_update(request:Request, server:Server):
req = request.body.decode()
match = re.match('Kp=(\d+)&Ki=(\d+)&Kd=(\d+)', req)
if not match:
raise Exception(f'failure to match, {req}')
pid = server.state.trickler.pid
pid.Kp = int(match.group(1))
pid.Ki = int(match.group(2))
pid.Kd = int(match.group(3))
print(f'updated pids to {pid.Kp=} {pid.Ki=} {pid.Kd=}')
return Response(request, '', content_type='text/html')```
and then I can just import `pid_update` and add it as a route whereever to whatever server I like
in this case `state` is a fairly thin class that I enclose the global state and various sensors in for access throughout the project.
plus this implementation is really straightforward because all you do is move the route decorator from the server and add a route-registration method - `post` is just a thin wrapper around the original deco that has the method pre-set for handiness@rough folio Nevertheless, you would have to use the state in routes and hope it will be populated somewhere else, in other module, there is also a problem with typing such class.
I never encountered something like adding state to server, seems to me like low cohesion of Server in this case, but I have the most experience with Django and some with Flask, don't know how Fastapi handles that.
crimson ferry
Would it be expected that espidf.MemoryError: survives a reload?
rough folio
it's not too hard to type- you can just do
if typing.TYPE_CHECKING:
class Request(Request):
state:State
Example from Flask docs, similar to my approach:
rough folio
from flask import Blueprint, render_template, abort
from jinja2 import TemplateNotFound
simple_page = Blueprint('simple_page', __name__,
template_folder='templates')
@simple_page.route('/', defaults={'page': 'index'})
@simple_page.route('/<page>')
def show(page):
try:
return render_template('pages/%s.html' % page)
except TemplateNotFound:
abort(404)```
from flask import Flask
from yourapplication.simple_page import simple_page
app = Flask(name)
app.register_blueprint(simple_page)```
plucky tulip
it's not too hard to type- you can just do
```
if typing.TYPE_CHECKING:
class...
I mean, if you add e.g. a sensor to your state, it will not be types unless you create a new class for State, because without it, you wont get hints for e.g. state.bno055
rough folio
as it so happens I did lol
the way I structure this is basically like
class State:
def __init__(self, sensor:Sensor):
self.sensor=sensor
.....
etc
makes a handy global state that I can fiddle with in a repl etc
plucky tulip
This is a problem with Django setting for example, where you use a child but it is not statically defined, so it works, but is not looking good
rough folio
I also tend to have a "default init" staticmethod that returns the class with the sensors initialized for use in a repl when debugging stuff
and then some .update() method which calls .update() on the state's children for some sort of async programming
but that's just how I style these
plucky tulip
I would have to take a look at these Blueprints, although I believe the approach with importing server into multiple files where it is needed is more Python like, not lib specific
rough folio
I can even point at non-web libs that use this pattern lol
like discord.py for example off the top with it's cogs
flask's default method assumes an amount of ownership over everything that the module is a webserver first and other stuff second imo
plucky tulip
Would you be open to discuss it later this week? Right now (~1AM where I am) I think I am unable to fully understand the point you are making.
rough folio
haha, fair enough
empty stirrup
In case someone sees something I messed up: from a fresh install of CP (commit-id: 8de9d5a52) I run make -C mpy-cross, then make fetch-all-submodules then cd ports/unix; make axtls then make micropython This results in a compile error:
In file included from ../../extmod/ulab/code/numpy/vector.c:15:0:
../../extmod/ulab/code/numpy/vector.c: In function ‘ulab_sinc1’:
../../extmod/ulab/code/numpy/vector.c:580:20: error: conversion to ‘float’ from ‘mp_float_t {aka double}’ may alter its value [-Werror=float-conversion]
if (fpclassify(x) == FP_ZERO) {
^
../../extmod/ulab/code/numpy/vector.c:580:20: error: conversion to ‘float’ from ‘mp_float_t {aka double}’ may alter its value [-Werror=float-conversion]
if (fpclassify(x) == FP_ZERO) {
^
cc1: all warnings being treated as errors
../../py/mkrules.mk:62: recipe for target 'build-standard/extmod/ulab/code/numpy/vector.o' failed
It seems that mp_float_t is typedef'ed to double and for whatever reason fpclassify wants floats
digital shoreBOT
adafruit
solar whale
FYI -- I'm not sure whose attention to bring this to, but I think there is a real issue with interference between the radio and the USB ports on the RP2040 RFM9x board -- see this thread #1129476234678632540 message It has been bugging me for quite awhile but I finally got around to doing some more testing and I think it is real. Comments/suggestions welcome. I have not tried it under Arduino yet to see if it occurs there -- Adding that to my "todo" list.
crimson ferry
USB 3 ports? they're an issue for 2.4GHz, at least the other way around, wouldn't be surprising with somewhat lower frequencies https://www.usb.org/sites/default/files/327216.pdf
solar whale
oooh! -- good point
I does appear to also be an issue under Arduino so not a CP issue.
slender iron
@solar whale I think we saw something similar on one of the other rp2040 boards
solar whale
<@343570430042308609> I think we saw something similar on one of the other rp204...
Good to know. I just wanted to get it “on the record”
solar whale
I just started the thread in #help-with-radio Where would you like an issue?
slender iron
I just started the thread in <#439293289259204619> Where would you like an issue...
circuitpython repo is fine. thanks!
manic glacierBOT
manic glacierBOT
solar whale
circuitpython repo is fine. thanks!
Created issue 8176 https://github.com/adafruit/circuitpython/issues/8176
lime trellis
Has anyone mentioned (or built themselves) a profiler for circuitpython like FunctionTrace?
https://functiontrace.com/
I could see target micropython/circuitpython audiences benefitting greatly from something like this
graceful rain
Hi all, I'm trying to rebuild CP after pulling latest changes (it's been about 2 months), and trying to make fetch-all-submodules but hitting this error. I'm on main. Any suggestions?
remote: Total 0 (delta 0), reused 0 (delta 0), pack-reused 0
Entering 'ports/mimxrt10xx/sdk'
fatal: couldn't find remote ref refs/heads/master
fatal: run_command returned non-zero status for ports/mimxrt10xx/sdk
.
ERROR: fetch-submodules.sh FAILED
vague thicket
Hi, is there way how can I clear displayio group/buffer? Is this (#circuitpython-dev message) only way ho to do it? I work with epaper display so I do not want to write something to the display, refresh it to clear it and then refresh it with new data. Thanks
empty stirrup
@graceful rain Iirc there was some move to move from master to main so this error fatal: couldn't find remote ref refs/heads/master seems a bit suspect. Might it be simplest to just start over from scratch?
manic glacierBOT
lone axle
Hi, is there way how can I clear displayio group/buffer? Is this (https://discor...
How do you mean clear the group / buffer?
If you intend to show noting on the display for some period of time (even a small peirod between changing from other things) then you would need to explicitly show "nothing" from the code (an empty group or None should work I think).
But you could also change from whatever is currently showing directly to the thing you want to show next without explicitly "blanking out" the screen inbetween.
Can you describe the type of behavior that you're hoping to achieve a bit more?
manic glacierBOT
sterile tundra
whats the reason that question mark (bool) is unsupported in struct.unpack format strings?
https://docs.circuitpython.org/en/latest/shared-bindings/struct/index.html
I did see some mention of if not available C99 talk on the python docs (screenshot), but couldn't mentally see why not supported.
manic glacierBOT
craggy plume
I am working with Adafruit CircuitPython 8.2.0 on 2023-07-05; Waveshare RP2040-Zero with rp2040.
circuitpython 8.2.0 supports alarm.sleep_memory. Documents state the data type of sleep_memory must be byte, which makes sense. But the type returned is int. >>> import alarm
dir(alarm)
['class', 'name', 'SleepMemory', 'exit_and_deep_sleep_until_alarms', 'light_sleep_until_alarms', 'pin', 'sleep_memory', 'time', 'touch', 'wake_alarm']type(alarm.sleep_memory[0])
<class 'int'>
Which is it?
empty stirrup
(I don't have a suitable uc to test) what does type(alarm.sleep_memory) return?
manic glacierBOT
empty stirrup
I am working with Adafruit CircuitPython 8.2.0 on 2023-07-05; Waveshare RP2040-Z...
I wonder if this is WAI. In python, accessing individual elements of a bytearray should return an int (and is admittedly confusing.) type(alarm.sleep_memory[:1]) should probably work as expected.
jaunty juniper
circuitpython 8.2.0 supports alarm.sleep_memory. Documents state the data type o...
that's normal, what's the issue ?
the bytes and bytearray types are arrays of bytes, they contain ints (limited to 0-255)
(and are represented as ASCII strings in the REPL, which does confuse people too sometimes)
craggy plume
(I don't have a suitable uc to test) what does type(alarm.sleep_memory) return?
import alarm
type(alarm.sleep_memory)
<class 'SleepMemory'>
OK. So it is not necessary to convert ints to bytes when assigning to alarm.sleep_memory? i.e. alarm.sleep_memory = 25 should work and only take up a single byte.
empty stirrup
yes - looking at the code, that should be fine
It's expecting an int between 0 and 255, and should throw an error if that isn't the case: https://github.com/adafruit/circuitpython/blob/8de9d5a52259cd0422a3633b6cf562e44d0cca9d/shared-bindings/alarm/SleepMemory.c#L161
empty stirrup
OK. So it is not necessary to convert ints to bytes when assigning to alarm.slee...
I'm guessing this might be a typo 🙂 but just to be clear - alarm.sleep_memory[0] = 25 should do the trick
tulip sleet
hello, I'm having a hard time making CircuitPython run on a custom board based ...
Could you link to your repo so we can do some code comparisons? Could you also point to a schematic? And have you tried Arduino (with a generic board profile) and writing to the USB serial console? If the latter does not work, I would suspect hw.
tulip sleet
In case someone sees something I messed up: from a fresh install of CP (commit-i...
This is apparently due to updating to the latest ulab. We don't build with ulab in the CI builds, because it's not needed for the tests. Could you file an issue in https://github.com/v923z/micropython-ulab/issues ? Thanks. I think Zoltán would be interested in this.
tulip sleet
Hi all, I'm trying to rebuild CP after pulling latest changes (it's been about 2...
Also can try make remove-all-submodules first.
empty stirrup
This is apparently due to updating to the latest ulab. We don't build with ulab ...
Ah, nice - thx for the tip! I'll add an issue as well.
tulip sleet
whats the reason that question mark (bool) is unsupported in struct.unpack forma...
This decision was made by the MicroPython folks: not sure why, maybe because using a b or B type will yield a "truthy" value anyway. We just inherited their implementation. I searched their github threads, found nothing on it.
tulip sleet
@jaunty juniper @drifting hearth @midnight ember I recorded the somewhat unhelpful Too many display busses discussed a few days ago in #help-with-circuitpython as an issue. ⬆️
midnight ember
Thank you. Also is it just me... when clicking on the help with alias channel comment links sometimes they take you to a place that's no where close to the actual comment it's referencing?
for example that link shows me this
which has no mention of display busses
but i know the conversation you're talking about and the opened issue 8178 is appreciated.
jaunty juniper
yeah it helps sometimes to go back and click the link again, I assume because the convo is already loaded then
tulip sleet
Thank you. Also is it just me... when clicking on the help with alias channel c...
I double-checked the link -- it's the right one, but you're right, it's going to something slightly above the actual post.
vague thicket
How do you mean clear the group / buffer?
If you intend to show noting on the ...
I´m trying to recreate Google calendar PyPortal project but with epaper display and it works this way now:
At start it will connect to Wifi, get user conde for device and show it on the display. Then when I approve/connect device, it will get items from the calendar and displays it on the screen, but now, it does not clear the screen, but just add calendar items "on top" of the previous information. I know that I´m appending calendar data to the buffer with information for device setup and I would like to clear the buffer/group before I would write calendar data to the screen.
I hope that makes sense and in few words, I want display 2 pages with one display group.
lone axle
I´m trying to recreate Google calendar PyPortal project but with epaper display ...
There is a PageLayout helper class that can help with that use case. https://docs.circuitpython.org/projects/displayio-layout/en/latest/api.html#adafruit_displayio_layout.layouts.page_layout.PageLayout There is a basic example of it's usage here: https://github.com/adafruit/Adafruit_CircuitPython_DisplayIO_Layout/blob/main/examples/displayio_layout_page_layout_simpletest.py and a more advanced one within the same directory in that repo.
If you can share the code you're using I could look at it later on and see if I can recomend changes that don't involve using the library, but it kind of depends on what you have so far. Generally, it sounds like you're adding a new group to the visible group without removing the pre-existing one. You could use pop() or remove() functions on Group to get rid of the previous group (or keep it in a variable, but not on the display). You could instead use the hidden property of your Groups to make one visible and the other one hidden. Or you could keep entirely seperate groups and change between them by calling show() or setting the root_group to each different one when needed.
sterile tundra
This decision was made by the MicroPython folks: not sure why, maybe because usi...
I'd like code to be closer to cpython compatible, to make my life porting things easier, and other peoples. Would it be better to add this desired functionality to circuitpython, or would it only be accepted via upstream micropython? Maybe I should at least pose the question on the micropython github issues, but I have the feeling (no idea) that micropython is less concerned with being cpython compatbile than circuitpython is, that may well be a mistaken impression gleaned from thin air...
I'm not really concerned with adding the functionality as it was only two lines to change to port a driver, but when things work unchanged like magic it's sooo good.
vague thicket
There is a PageLayout helper class that can help with that use case. https://doc...
Thanks for your help. You can look at my code (which is not working currently as I have issue with datetime and strftime) here (https://www.dropbox.com/s/92v09j2y54fsqo5/google_calendar.py?dl=0). I will look for sure at PageLayout as I think it will help me organize different items (calendar, weather) on one page. I did not know about pop()/remove() functions but I think they are what I was looking for.
spice gate
Could you link to your repo so we can do some code comparisons? Could you also p...
turns out it was the power failure management. Commenting out the nrfx_power_pofwarn_config stuff in port_init() allows this to proceed into boot and work well. I wonder why is so though.
question: when using external SPI flash with SPI_FLASH_FILESYSTEM = 1 and EXTERNAL_FLASH_DEVICES suitably defined, is there a general assumption that that SPI interface is exclusively used for the filesystem? What's the best practice in CircuitPython when you have a single SPI bus connected to the flash chip and to all other SPI devices on the board ?
tulip sleet
turns out it was the power failure management. Commenting out the nrfx_power_pof...
that is odd. Note there are some issues about what the default GPIO level is -- there's a non-volatile memory setting, and maybe for some other things. So could be the "fuse" settings.
tulip sleet
I'd like code to be closer to cpython compatible, to make my life porting things...
I think you could start by opening an issue in the MicroPython repo (or a PR). We'd be happy to add it, and if it were added upstream we'd take that too.
spice gate
that is odd. Note there are some issues about what the default GPIO level is -- ...
lowering the POF threshold from NRF_POWER_POFTHR_V27 to NRF_POWER_POFTHR_V25 also fixes this, so it might be that the default threshold is too high for when I run off battery
tulip sleet
what kind of battery? a 3.7V lipo?
spice gate
yes
tulip sleet
hmm, our code is assuming 3.3V. (so that should match your use case) ... I seem to remember that there are several ways of wiring up power for the nRF52840, and the use cases are not completely obvious. You might compare your schematic with ours.
VDD and VDDH, etc.
spice gate
yeah my setup is pretty complicated in that respect due to USB, LiPo, solar panel charger ... I'll look into it. Thanks a lot Danh for your time and feedback.
manic glacierBOT
slender iron
Has anyone mentioned (or built themselves) a profiler for circuitpython like Fun...
I've played around at the c function level but not the python one. I used the compiler's entry and exit annotations to do it.
manic glacierBOT
I'm just using the spring antenna in this cases.
Quick test does seem better on the Mac... I'll run a bunch more tests. Thanks!
manic glacierBOT
random junco
<@&356864093652516868>: The weekly meeting is in about 90 minutes at 2pm EST / 11am PST. I look forward to hearing your hug reports and status updates. Here is the link to this week's notes doc: https://docs.google.com/document/d/10AzCwaac8jCgfUiFIPchwTghFqHptCV99iMCsZfxAwo/edit?usp=sharing
proven garnet
Heads up that Node12 is gone for good on GitHub Actions runners starting August 14th! I think we're all set at this point across CircuitPython and Arduino but heads up in general (and your own projects!)
royal rover
don't forget timestamps
onyx hinge
that is the shortest list of library activity I remember seeing! I wonder if it's correct.
onyx hinge
thanks for doing stats <@426200547024961539>
you're welcome. was pretty sure you wouldn't mind if I offered to do it
idle owl
that is the shortest list of library activity I remember seeing! I wonder if it'...
It is. I thought the same thing last week, but Tekktrik verified it.
onyx hinge
that is the shortest list of library activity I remember seeing! I wonder if it'...
I guess things are just quiet.
midnight ember
Nice to see foamyguy back online
onyx hinge
specifically struct "?" format character
onyx hinge
newsletter editors are basically Audrey saying "Feed me, Seymour"
mental nexus
Online hug for all the efforts on USB host and a CircuitPython editor.
slender iron
👍
midnight ember
Hope Ari gets well soon.
Links to examples of the fonts?
a font with more language support sounds good
silver tapir
Those fonts sound interesting, I'll have to test them later with the bribri/cabécar update I have.
onyx hinge
@mental nexus nice to see you drop by!
mental nexus
<@366059101203202052> nice to see you drop by!
Thanks, I’m still keeping an eye on all the great new things in CP.
midnight ember
Cozette doesn't look too different from terminalio so wouldn't be a drastic change, plus the extra language support, looks good to me.
onyx hinge
slender iron
cozette doesn't really do more languages but it does do emoji
idle owl
I have to run.
midnight ember
Unifont looks more industrial, my preference is cozette with all things being equal?
if a language isn't supported by terminalio by default i could see how that could be a big problem for certain language
midnight ember
i would actually like to see a 5x5 supported by default too
smaller text is better for smaller displays and matrix panels.
silver tapir
Thanks for hosting Paul. Nice meeting.
lone axle
Thanks for hosting Paul! Have a great week everyone 🎉
midnight ember
Thank you for hosting @random junco Great job!
slender iron
our current font, terminus is 6x12
onyx hinge
http://wenq.org/wqy2/index.cgi?BitmapSong_en
The Wen Quan Yi bitmap font includes complete CJK Unified Ideograph (U4E00 - U9FA5) glyphs at four different sizes (9pt-12X12 pixel, 10pt-13X13 pixel, 11pt-15X15 pixel, 12pt-16x16 pixel) and two weights (medium and bold). Use of this bitmap font for on-screen display of Chinese (traditional and simplified) in web pages and elsewhere eliminates the annoying "blurring" problems caused by the high stroke density of many Chinese characters and insufficient "hinting" of anti-aliased Chinese fonts. This font also provides bitmap glyphs for Japanese Hiragana (U3040 - U309F), Katakana (U30A0 - U30FF) and for Korean Hangul (UAC00 - UD7A3).
midnight ember
yup and with smaller font you can fit more lines on any display. you can fit one more horizontal line on a matrix panel using a 5x5 font.
you'll be squinting on a tiny SSD1306 with a 5x5 but on a matrix panel it's still quite large.
random junco
@lone axle any chance you made a backup recording? I only have my sound, not the speakers.
I even thought I tested it before the meeting. 😦
lone axle
<@382939733107408897> any chance you made a backup recording? I only have my so...
I do! I'll get it uploaded to dropbox and send a link in a moment
random junco
Thank goodness, I've got an early hug report. Thanks so much
lone axle
@random junco https://www.dropbox.com/scl/fi/brr190mbjd7fqe9z470da/CircuitPython_Weekly_Meeting_7_17_23.mkv?rlkey=0ak6gd92i9chjttlnypzba8mg&dl=0 here we go. I had a quick scan through to make sure everything was captured, and I believe it was. But have not listened to the whole thing of course.
random junco
<@273162248615100417> https://www.dropbox.com/scl/fi/brr190mbjd7fqe9z470da/Circu...
thanks! downloading now and I'll give it a run through
random junco
Can someone approve my PR to add the meeting notes? Then I can add the link to the video. Thanks! https://github.com/adafruit/adafruit-circuitpython-weekly-meeting/pull/125
thanks dan!
random junco
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/1Zx4UF48QN6YoRq-qVAJgdsfr-5KW401aJPTYwjxMQns/edit?usp=sharing
spare jacinth
hey, i have a couple of aliexpress boards:
- https://es.aliexpress.com/item/1005005592730189.html
- https://es.aliexpress.com/item/1005003796653297.html black 16MB one with RGB
- https://es.aliexpress.com/item/1005004539320086.html "classic" model (not received yet)
...which i would like to "port". I dont mind having to fiddle looking for datasheets and whatnot, but i thought it would be somewhat easy to add them to the supported boards list, so less-experienced users could just download a fine-tuned firmware (and can find it with an image of the product instead of matching with similar stuff)
my assumption is that creating a "board" would be pretty much assigning the several aliases for pins, flash size, and perhaps some other config and that i dont need to be affiliated with the companies in any way, but can do it in a community-project way. i reckon that they would probably need unique vid/pid pairs, how to go about that?
confused about the 1st product, as the chip on it is a "regular" one, while all ESP32-S3 images on the download section are enclosed in a metal box, does that mean it is very different? 🤔
second one seems like it could be a clone of https://circuitpython.org/board/weact_studio_pico_16mb/, if it really is, is there some sort of aliasing method or would it require duplicating code?
slender iron
hey, i have a couple of aliexpress boards:
1) <https://es.aliexpress.com/item/10...
This is the best place to start on adding a new board: https://learn.adafruit.com/how-to-add-a-new-board-to-circuitpython
tulip sleet
hey, i have a couple of aliexpress boards:
1) <https://es.aliexpress.com/item/10...
There are hundreds of these ESP32-clone-ish boards on aliexpress. The manufacturer provides little to no support, and they come and go. A number have been "cost reduced" to the point that their functionality suffers. If you can find an existing build that works, I'd recommend just using that. If there is no generic build that works, then creating a generic build that would work for a certain class of board (combo of chip, flash chip, PSRAM) I think would be best.
spare jacinth
ugh, that's definitely a stopper indeed, wouldnt be nice setting up building a board to then get some of its components changed without any warning, guess i will not make the ports then
but will hopefully be able to find a working build, so far i think it came with nothing flashed on it, unsure what/how to flash it - probably tinyuf2 shall be the 1st thing to throw at it
tulip sleet
For instance, there are Espressif-built dev boards for all the different modules, and we have builds for many of them. They break out all the pins, and are pretty generic.
spare jacinth
i almost forgot... thx both! 😄
crimson ferry
UnexpectedMaker S3 boards are N16R8, not sure if the flash and PSRAM types would match though
spare jacinth
i dont even know what that means rigth now, lol.... S3 are a "series" and thats one of the actual devices?
crimson ferry
like board (1) above ...16MB flash, 8MB PSRAM
https://circuitpython.org/board/unexpectedmaker_feathers3/ (also proS3 and tinyS3)
tulip sleet
for the RP2040 boards, the Pi Pico build would probably be just fine, it loks like the pinout is the same
spare jacinth
for the RP2040 boards, the Pi Pico build would probably be just fine, it loks li...
i flashed it a couple months ago with i-dont-remember-what and i could work with it no issue, but im rather familiar with RP2040... im not with ESP
tulip sleet
ESP boards are a Cartesian product of a bunch of choices.
ESP32, ESP32-S2, ESP32-S3, ESP32-C3 are all different chips. only S2 and S3 have full native USB support
ESP32 has no USB support; ESP32-C3 has only USB serial on chip
-S2 and -S3 are a lot easier to deal with for CPy purposes
spare jacinth
cool, that make things slightly more clear
spare jacinth
<https://circuitpython.org/board/unexpectedmaker_feathers3/> (also proS3 ~~and t...
is this a known issue? i dont even know if my hw matches what that firmware is setup for... but the drive name does match
(time to manually copy the uf2)
spare jacinth
is this a known issue? i dont even know if my hw matches what that firmware is s...
no circuitpy drive for me, neither with that firmware nor the Pro one, time for a ride to compile some custom stuff?
crimson ferry
@spare jacinth what tools produced those screenshots? You should be able to just drag the UF2, and it shouldn't need to match
but yeah, no surprise if the flash and / or PSRAM don't match
spare jacinth
<@168269895988019200> what tools produced those screenshots? You should be able ...
that was the experimental installer (from Chrome), gave it a try for the bootloader which i was not sure how to get into it (was reading on esptool at the same time) + give some feedback
bootloader seems to be working tho, i can see the external drive after double-pressing boot button, with what im guessing are the 3 expected files on it
crimson ferry
but likely no bootloader on those boards, esptool can write the .bin file directly
ah, ok, there is a bootloader
spare jacinth
the web installer did flash some tinyuf2 which is apparently working fine, yeah
it probably used esptool under the hood im assuming
crimson ferry
not sure, never used that one (don't have Chrome)
spare jacinth
(shall move this convo elsewhere?)
tulip sleet
https://circuitpython.org/board/espressif_esp32s3_devkitc_1_n8r8/ might work, pretending to only have 8GB flash
maybe start a thread
spare jacinth
ESP32-S3 setup issues
devout jolt
oh hai @onyx hinge, I just heard you mention this in the meeting notes. I will definitely like to help test sdioio support on RP2040!
onyx hinge
@devout jolt keep an eye open for the pull request, hopefully this week
manic glacierBOT
onyx hinge
@devout jolt It's not ready to PR yet (because capacity is not detected & write is un-tested, mainly), but it can read a FAT filesystem on the adafruit metro rp2040 prototype board I'm using for testing. in theory can work with other setups. main requirement is that DAT0..3 are sequential in GPIO numbering https://github.com/adafruit/circuitpython/compare/main...jepler:circuitpython:rp2-sdioio?expand=1
tulip sleet
@onyx hinge could I talk to you on audio or video sometime today about some strategies about doing this merge? I don't want to give up, but I'm trying to figure out how to review all the changes more precisely and with less chance for error. I've done some studying of "review every change" kind of stuff in stackoverflow, and none seem quite right. I can explain more orally.
onyx hinge
@tulip sleet sure. After 1:30 CT / 2:30 ET would be better for me.
tulip sleet
sure, thanks, I'll ping you then, and do something else for the nonce.
abstract oak
I know there's no actual Doxygen or anything, but are doc-comments on CircuitPython internal C functions accepted?
slender iron
I know there's no actual Doxygen or anything, but are doc-comments on CircuitPyt...
what do you mean by doc-comments? I don't see why not
abstract oak
Doxygen-style comments for annotating parameters, mostly
there are a few cases of that already in the source tree, but very few
abstract oak
onyx hinge
@tulip sleet I'm back home and getting settled in, can video in ~10 minutes from now or maybe a bit earlier
tulip sleet
sure - finishing up a conversation
@onyx hinge Kattni needs me to review something quick, I'll get back to you, if that's ok
onyx hinge
that's fine
tulip sleet
@onyx hinge ok, done with that for now. (I was doing a sample review for a library so Kattni could screenshot it to update a git guide.)
tulip sleet
devout jolt
<@352910176736772096> It's not ready to PR yet (because capacity is not detected...
nice! I think we can also test this with the Pimoroni Pico DV Demo Base, which has SD card wired for SDIO
onyx hinge
I didn't check but I think it's a good assumption that the Pico DV Demo Base is wired for this code to work.
manic glacierBOT
spare jacinth
ESP32-S3 setup issues
this got somewhat buried..
i've just posted a little extra info, if anyone can/want to help, any guidance is welcome 😄
tulip sleet
@onyx hinge thanks for the talk yesterday; I'm going to use difftool between our tip of main and the merged fileson some of the changed files to re-vet some changes, but probably not starting over. But I found some useful git log flags and args, e.g. git log --oneline --name-status v1.18..v1.19.1 lists all commits between the tags, and also changed files. Damien is very systematic about his commit messages, and that's very helpful.
@onyx hinge @slender iron my recollection (and what ag seems to tell me) is that these kinds of lines in the board-specific sdkconfig only act as arguments to some Makefile flashing targets. But am I wrong and they influence the build?
CIRCUITPY_ESP_FLASH_MODE = dio
CIRCUITPY_ESP_FLASH_FREQ = 40m
CIRCUITPY_ESP_FLASH_SIZE = 4MB
as far as I can tell that's correct, is there a reason you're doubting it?
ports/espressif/Makefile: FLASH_SDKCONFIG = esp-idf-config/sdkconfig-$(CIRCUITPY_ESP_FLASH_SIZE)-no-ota-no-uf2.defaults```
flash size is used to select the sdkconfig filei meant esptool args... ah, you're right, it selects the sdkconfig
trim elm
@lone axle hey so I'm up to 5750 on the OSHWA stuff
I heard you were starting at higher numbers and working back
idle owl
<@382939733107408897> hey so I'm up to 5750 on the OSHWA stuff
Please keep this discussion in the email thread as Phil requested. We cannot all remain on the same page if you spread it over multiple platforms. @lone axle
trim elm
Please keep this discussion in the email thread as Phil requested. We cannot all...
Will do. Sorry about that
rancid pilot
I think i am done with the canio support for the port mimxrt10xx. I would like to make a build to test and verify. However I get the following error messages. What step am I missing?
lone axle
I think i am done with the canio support for the port mimxrt10xx. I would like t...
maybe try make BOARD=whatever clean to clear out prior builds and force it to generate everything fresh.
rancid pilot
i think i am doing that already
tulip sleet
i think i am doing that already
did you set CIRCUITPY_CANIO = 1 ? If so where?
lone axle
okay, someone else may have better ideas for you. I don't recall needing to do anything specific, but I've mostly copy/pasted the boilerplate, possibly I've been doing a step that I overlooked.
onyx hinge
@devout jolt I updated my branch for sdio .. reading is pretty solid, writing is causing a lot of hangs & FAT corruption 😞 and I failed at figuring out how to get the "CSD" which is where the card capacity info is stored.
spare jacinth
not sure if this is the place for this kind of reports, but:
- https://learn.adafruit.com/mqtt-in-circuitpython/circuitpython-wifi-usage links to another tutorial for ESP32-S2 boards (thought i may need something different for S3, but code seems to work just fine)
- i guess S3 support is still not mature enough to want it listed on docs and/or they aren't not fully updated yet
- the code on such tutorial (https://learn.adafruit.com/adafruit-metro-esp32-s2/circuitpython-internet-test) does
ping = wifi.radio.ping(ip=ping_ip) * 1000which resulted in
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
TypeError: unsupported types for __mul__: 'NoneType', 'int'
```not always tho, but the `* 1000` could be moved to the `if ping is not None` block below i guess@onyx hinge i used SD_Card_Size = os.statvfs(virtual_root) to pulll the card capacity
Your issue is likely much lower level than I can dive but you've been of tremendous help so if there's even a sliver of way I can help you back I'm taking it. 😉
onyx hinge
@midnight ember that's good information, thanks. In this case, I need the number of sectors on the whole SD card (low level information) not details about a filesystem, so that I can implement https://docs.circuitpython.org/en/latest/shared-bindings/sdioio/index.html#sdioio.SDCard.count
that method should work whether or not there's a proper FAT filesystem on the SD card
but yes, mounting and reading the device's FAT filesystem works OK even if the "capacity" number is returned as 0
midnight ember
it was worth a shot. wish you the best of luck!
crimson ferry
@spare jacinth code should run similarly on ESP32-S2 and -S3, with the noted exceptions in the release notes (plus any open issues) https://github.com/adafruit/circuitpython/releases/tag/8.2.0. ping on espressif in general has some limitations https://github.com/adafruit/circuitpython/issues/7940
jaunty juniper
not sure if this is the place for this kind of reports, but:
- <https://learn.ad...
you can leave feedback on a learn guide via the "feedback" link on the right of it
you are correct that the internet test should * 1000 in the if
(the github link is at the bottom od the code block on inserts like that)
spare jacinth
PR on the way 😉
update: there it is
spice gate
Hello. When porting CircuitPython to a new board, is it OK to share the same SPI lines / interface between external SPI flash and other peripherals?
slender iron
Hello. When porting CircuitPython to a new board, is it OK to share the same SPI...
Yes, but if you can make it not shared then that'd be better
this is why busio.SPI has try_lock
@tulip sleet @onyx hinge what do you think about me switching the heap allocation size for imx to 32 from 16. that way buffers will align with the cache lines
(I'm seeing tinyusb invalidate memory at a +16 offset and destroying the obj struct before it.)
it'll make allocating stuff on the heap tighter
(but maybe not by much)
tulip sleet
<@329766224093249548> <@426200547024961539> what do you think about me switching...
worth trying. I wonder if it might improve available heap on small boards too. Or is there gc storage that gets bigger?
everything is 32 right now? Does MicroPython ever use 16 for any boards?
tulip sleet
oh sorry, I misread
slender iron
but will also reduce the bookkeeping overhead
tulip sleet
the 1011 is pretty tight, maybe try synthio examples with 32 and see if we run into trouble
trying to think of what the main use cases for i.MX are
slender iron
I can make it conditional on usb host
tulip sleet
MIDI will want USB host so you can plug in a controller
slender iron
though I wonder if it causes issues with device too
spice gate
Yes, but if you can make it not shared then that'd be better
Thank you 👍
spice gate
Yes, but if you can make it not shared then that'd be better
one issue I see (v8.2.0) when running with the CIRCUITPYTHON drive backed by the SPI flash chip, is that when I invoke the busio.SPI() constructor I get an error stating that the SPI signals (SCK, others) are already in use. If I use board.SPI(), I get the same error. Is there something I'm missing ? Thanks in advance.
onyx hinge
@spice gate @slender iron I don't know that we have any boards that operate that way now. And I'm not sure it'll operate well, even once you get past that error.
A write to CIRCUITPY from the host computer via USB can happen at any time. If the SPI bus is locked by some Python code, I don't think we have an effective way to defer the write from USB until it's unlocked; the only possibility I see is to fail it when the bus's try_lock fails.
Anyway, when creating the SPI bus for the CIRCUITPY storage I think you'd need to arrange for common_hal_board_get_spi(0) to return supervisor_flash_spi_bus or a way for the routines in supervisor/spi_flash_api.h to use the SPI bus common_hal_board_get_spi(0) instead.
spice gate
<@721997443188654080> <@252717193496756235> I don't know that we have any boards...
thanks @onyx hinge , this is insightful. Do I understand it correctly, that all of the currently supported boards use different SPI interfaces for the external SPI flash filesystem (when present) and for all the other SPI peripherals? Since this is a board for which I have all the peripherals (serial flash included) connected to the same SPI lines and MCU interface, what's the best approach for porting CircuitPython? Shall I simply fall back to the internal flash, and use the SPI bus for non-storage peripherals only?
onyx hinge
thanks <@426200547024961539> , this is insightful. Do I understand it correctly,...
of the currently supported boards use different SPI interfaces for the external SPI flash filesystem (when present) and for all the other SPI peripherals
From what I can see that's the case, but I sure don't know all of our hundreds of supported boards anymore. microbit_v2 is an example of an nrf board that uses a portion of internal flash for the CIRCUITPY filesystem.
fwiw it'd probably be a neat addition if similar to "sdcardio" it was possible to mount an external flash chip.
spice gate
> of the currently supported boards use different SPI interfaces for the externa...
yeah I was thiking of that. Using the internal flash for the main "CIRCUITPY" filesystem, and then mounting the external flash over SPI using something like sdcardio. Thanks a lot for this feedback - much appreciated !
manic glacierBOT
This documents which boards support the following items that are internal to a module, and can be separate enabled/disabled:
- builtins.pow3 (only in support matrix, there does not seem to be docs for this function)
- busio.UART
- busio.SPI
- keypad.Keys
- keypad.KeyMatrix
- keypad.ShiftRegisterKeys
- os.getenv
Closes #8168
timid bolt
It seems that USB host on RP2 uses PIO. However the RP2 datasheet says it has a USB host controller on chip. Is that true so far? If so why the choice not to use the on-chip controller?
slender iron
It seems that USB host on RP2 uses PIO. However the RP2 datasheet says it has a ...
we want to use the native peripheral for device still
timid bolt
Tinyusb has these examples of dual host/device mode. I haven't looked into it, but does that help you?
slender iron
it's what we're doing with cp already
timid bolt
Dual mode requires separate physical ports?, Or dual mode is already available on the native port?
timid bolt
we want to use the native peripheral for device still
Do you want device and host simultaneously? For example, phones are both device and host on one physical port but not at the same time.
slender iron
🙄
timid bolt
Has anyone tried to use a Tinyusb audio device? For me, Windows at least doesn't accept the device.
slender iron
Has anyone tried to use a Tinyusb audio device? For me, Windows at least doesn't...
I've only done the midi
timid bolt
Because I notice there isn't usb audio device support in CircuitPython and I'm wondering is there's maybe an issue with Tinyusb.
timid bolt
I've only done the midi
Idk if you've looked at audio, but the usb descriptor configuration is amazingly complicated 😵💫
slender iron
not really. It'd be cool to have audio in and out to cp though
timid bolt
Agreed. Does Tinyusb have a Discord?
devout jolt
Has anyone seen the error when trying to build for ESP32 (S3 in this case) of
-- The C compiler identification is unknown
-- The CXX compiler identification is unknown
-- The ASM compiler identification is unknown
-- Found assembler: xtensa-esp32s3-elf-gcc
CMake Error at esp-idf/tools/cmake/project.cmake:296 (__project):
The CMAKE_C_COMPILER:
xtensa-esp32s3-elf-gcc
is not a full path and was not found in the PATH.
The process I used to build was, from a fresh checkout and after installing all prerequisites:
cd circuitpython/ports/espressif
make fetch-port-submodules
make BOARD=adafruit_qtpy_esp32s3_nopsram
ahh nevermind. the new make fetch-port-submodules does not invoke the esp-idf/install.sh.
jaunty juniper
yeah that was always done by hand
devout jolt
yep yep, haven't built for esp32 in a while
plucky tulip
haha, fair enough
Took me a bit longer, but I checked out the Blueprints and I managed to create a very similar behavior with minimal amount of added code. I already pushed these changes to the PR (https://github.com/adafruit/Adafruit_CircuitPython_HTTPServer/pull/63). I does not work exactly like Flask, I skipped the intermediate Blueprint object, mainly in order to save memory and to keep the codebase smaller, but you can import a Route the same way as you would import a Flask's Blueprint from external file, and register it to the Server.
rough folio
Yeah I think the flask blueprint is a bit of a curious design lel
I'll take a looksy in a bit
plucky tulip
Yeah I think the flask blueprint is a bit of a curious design lel
I'll take a ...
They might be more intuitive in the large webapp with lots of routes, but I do not think that someone will reach this kind of scale on a microcontroller. 😛
rough folio
yeah this is pretty much all you need I think- more or less the same api I used
lg2m
It is a shorthand for manually creating a ``Route`` object, that can be used only one time
per function.```
this wording is kind of weird though. First read it implies that the route can only be used once or something- but I'm pretty sure you mean you can only stack the deco once which I'd hope would be kind of implicit
idk minor nit doesn't really matter
something that might be worth thinking about is a route.add_route which would add a subroute but that's probably more than needs doing- ie
@as_route(/path)
def func(...)
...
@func.as_route('/subpath')
def xyzzy(..)
....
xyzzy could be accessed as /path/subpath
I don't _think_ this would be arduous to add?whoops forgot to reply here's a ping @plucky tulip
plucky tulip
yeah this is pretty much all you need I think- more or less the same api I used ...
Thanks for review
manic glacierBOT
plucky tulip
Would it possible to compile CircuitPython in a way that keeps track of symbolic local environment, MicroPython docs says it is a optimization (https://docs.micropython.org/en/latest/genrst/core_language.html#local-variables-aren-t-included-in-locals-result).
Is it only the matter of changing a variable from OFF to ON or is it implemented deeply inside core?
Is it be possible to add such feature to CP? Would it be hard, or even possible?
digital shoreBOT
adafruit
digital shoreBOT
adafruit
tulip sleet
Would it possible to compile CircuitPython in a way that keeps track of symbolic...
This is a deep choice in the implementation: the locals are internally referred to with a slot number or something like that, so the name information is lost, sadly.
slender iron
Would it possible to compile CircuitPython in a way that keeps track of symbolic...
Probably better to ask the MP folks about difficulty too.
tulip sleet
cunning orchid
Quick question: I'm awaiting a review with my company's Legal team next week to get the official go-ahead to contribute (mainly a formality as they're pretty open-source friendly). In the meantime, I'm looking for places to start nibbling to get a better feel for the codebase before poking at multicore. Looking at the issues in GitHub, I see tags "help wanted" and "good first issue" am I correct in understanding that you think that these are good starting issues for new to the project contributors?
idle owl
Quick question: I'm awaiting a review with my company's Legal team next week to ...
Hello and welcome! The Good First Issue label is definitely meant for folks who are new to contributing to open source in general, or those who are new to our project. Are you familiar with Git and GitHub?
cunning orchid
Hello and welcome! The Good First Issue label is definitely meant for folks who ...
Thanks! I am indeed familiar with Git and somewhat GitHub (I've mainly used Gerrit in my day job).
idle owl
Thanks! I am indeed familiar with Git and somewhat GitHub (I've mainly used Gerr...
I've never heard of Gerrit! GitHub is pretty straightforward, though, so I don't doubt it will make sense to you as you get into it. Given that familiarity, I would say you can probably pick up any of the smaller issues, if something that isn't labeled Good First Issue interests you. We often use that label for folks who are new to the whole process.
Another thing to keep in mind is that you can always tackle larger items in smaller chunks, and submit multiple contributions as you work through it.
As long as it doesn't break things, obviously. 😄
cunning orchid
I've never heard of Gerrit! GitHub is pretty straightforward, though, so I don't...
I think it was made by Google originally and then released to open-source. It's basically a Git server with a wrapper that pushes the workflow into code reviews (similar to GitHub's PR) and a UI that's clearly built by backend people with little thought for UI/UX flare (like myself) as well as handy integration capabilities for external CI/CD systems.
Nowhere near as visually pleasing as GitHub or GitLab but, seems to be a solid self-hosted option.
idle owl
I think it was made by Google originally and then released to open-source. It's ...
That's interesting! I had wondered whether there was such a thing, but never looked in to it.
cunning orchid
Another thing to keep in mind is that you can always tackle larger items in smal...
That's definitely something that I need to do in my day job as well. Might be good practice :).
manic glacierBOT
manic glacierBOT
small comet
Not a big rush to debug this, but just in case someone sees anything immediately obvious: I attempted to build an esp32c3 board from main - with this: make V=2 BOARD=beetle-esp32-c3 PORT=/dev/ttyACM0 flash monitor
The build works well enough and flashes, but it looks like the board goes into a boot loop. The repeating output from the monitor part looks like this:
Waiting for the device to reconnect.
ESP-ROM:esp32c3-api1-20210207
Build:Feb 7 2021
rst:0x15 (USB_UART_CHIP_RESET),boot:0xd (SPI_FAST_FLASH_BOOT)
Saved PC:0x40049a42
SPIWP:0xee
mode:QIO, clock div:1
load:0x3fcd5810,len:0x24
ets_loader.c 78
ESP-ROM:esp32c3-api1-20210207
Build:Feb 7 2021
rst:0x7 (TG0WDT_SYS_RST),boot:0xd (SPI_FAST_FLASH_BOOT)
Saved PC:0x40049a42
SPIWP:0xee
mode:QIO, clock div:1
load:0x3fcd5810,len:0x24
ets_loader.c 78
device reports readiness to read but returned no data (device disconnected or multiple access on port?)
Waiting for the device to reconnect.
ESP-ROM:esp32c3-api1-20210207
small comet
Ah, thanks @slender iron ! Well, I guess I have a board that can reproduce the issue, so also happy to give any fixes a try if needed. Oddly, I am able to install the version provided at https://circuitpython.org/board/beetle-esp32-c3/ (Adafruit CircuitPython 8.2.0 on 2023-07-05; DFRobot Beetle ESP32-C3 with ESP32-C3FN4). Also, I looks like it has 4M flash. (board specs here: https://www.dfrobot.com/product-2566.html)
small comet
@onyx hinge (or anyone familiar with asyncio really) is this code https://gist.github.com/jepler/e620cb8396a941866c95501b22b6c0ee supposed to work with 8.2.0 when using the "native" IOQueue with the ioctl protocol? When I run it on my esp32c3 board, only the logic task runs. I'm not able to trigger the pin events. Moreover, in the logic task I added something to print out the map in the global _io_queue instance, and surprisingly it is empty. (Also confirmed separately, if I don't add any "logic" tasks, the asyncio.run() terminates right away.
onyx hinge
@small comet I don't have any context in my brain about that, and I'm away from all cp hardware for a few days. it's possible that e.g., something about adafruit_circuitpython_asyncio pull #33 means this code doesn't work now. it would be nice to have working code that makes awaiting key presses "most performant", but that example code may well have bit-rotted because nobody's actually using it.
small comet
<@1130377638523977739> I don't have any context in my brain about that, and I'm ...
Ah, ok - no worries. Thanks for the additional context!
lone axle
Does anyone know if matrix portal S3 is shipping with UF2 bootloader? I don't seem to be able to get the double click reset working. I can hold boot and reset and it does seem to be in a bootloader mode, but no drive appearing, only /dev/ttyACM0
And potential followup, if it doesn't ship with that is it possible currently to install it? The page here: https://circuitpython.org/board/adafruit_matrixportal_s3/ has a 'Download bootloader.zip' button at the bottom but it leads to a "Not Found" page atm.
I did find a combined.bin that seems to work in the release here: https://github.com/adafruit/tinyuf2/releases/ looks like maybe just the URL for releases changed or something.
opal forge
hey folks! trying to get circuitpython 8.2 set up on a new MatrixPortal S3. the board comes with a simple sketch and doesn't seem to include the bootloader. i tried to do a full install using the installer (https://circuitpython.org/board/adafruit_matrixportal_s3/) but it is dying because it can't fetch the tinyuf2 bootloader.
it's trying to load https://adafruit-circuit-python.s3.amazonaws.com/bootloaders/esp32/tinyuf2-adafruit_matrixportal_s3-0.13.0.zip - which doesn't exist. the "download bootloader zip" for manually installing points at a different(?) non-existent file: https://github.com/adafruit/tinyuf2/releases/download/0.13.0/tinyuf2-adafruit_matrixportal_s3-0.13.0.zip.
as a member of the adafruit webdev team (👋🏻) i have the ability to peer into the actual bucket where we store these files and found this file instead: https://adafruit-circuit-python.s3.amazonaws.com/bootloaders/esp32/tinyuf2-adafruit_matrixportal_s3-0.15.0.zip
not sure how all these versions and builds are managed on circuitpython.org but it seems like this needs an update!
small comet
I think there is an issue filed for this
I think I might have an unrelated issue for the board I'm using (beetle-esp32-c3) - I was able to at least get a working system by switching to CIRCUITPY_ESP_FLASH_MODE=dio in boards/beetle-esp32-c3/mpconfigboard.mk. I'll avoid muddying the ongoing debugging on the issue, but fyi'ing here in case anyone runs into something similar.
opal forge
hey folks! trying to get circuitpython 8.2 set up on a new MatrixPortal S3. the ...
Bootloader worked once I found it. Was able to get CircuitPython 8.2.0 loaded and JP's matrix clock example running. There's an... issue tho' 😅 a pixel addressing issue?
lone axle
Bootloader worked once I found it. Was able to get CircuitPython 8.2.0 loaded an...
I think you need a newer version of circuitpython. Try the absolute newest links from S3.
I just set one of these up today and I was seeing similar artifacts on the display until I got a newer build than 8.2.0
I'm not entirely certain how the page on circuitpython.org gets built, but it might be worth considering supressing the top right panel with current stable release 8.2.0 for the Matrix Portal S3 since that build doesn't seem to be able to handle the matrix properly. There was a fix in protomatter that came afterwards I think that is necessary to get the display showing accurately.
opal forge
I think you need a newer version of circuitpython. Try the absolute newest links...
wow that did the trick! i used yesterday's (20230721) build and it worked right away.
lone axle
Nice, glad to hear it 🎉. I'll ask around a bit on Monday and try to see whether it's possible to have the device page link to a newer one, or at least remove the link to 8.2.0 for now.
opal forge
another quirk on circuitpython.org: the original is listed as Matrix Portal M4 (two words for "Matrix" and "Portal") and the new one is listed as MatrixPortal S3 (one word). the search feature on the downloads page turns up both if you search matrixportal but only turns up the old M4 board if you search matrix portal.
lone axle
😄 I noticed that as well. I think there may be a keywords list or something similar and I'm guessing the original contains the opposite "spelling" so it gets found with both. The newer one must not have the keywords loaded up yet.
if you have a moment, opening an issue on circuitpython-org repo would be good to get someone to take a look and try to make them work the same.
opal forge
if you have a moment, opening an issue on circuitpython-org repo would be good t...
done! thanks for the nudge on that.
lone axle
Thank you!
tulip sleet
<@426200547024961539> (or anyone familiar with `asyncio` really) is this code ht...
This code is inherited from MicroPython but is not necessarily expected to work. We haven't tested it at all, and the I/O code in CircuitPython has no knowledge of this.
inner orbit
Does someone have a link to the CP6 source?
spare jacinth
git clone https://https://github.com/adafruit/circuitpython -b 6.0.0 and/or https://github.com/adafruit/circuitpython/tree/6.0.0
6.x.y should be easy to find too
carmine girder
I got a Galactic Unicorn and would like to use it with CircuitPython. I'm mostly interested in using the RGBMatrix library, of course. While looking at the schematics in https://cdn.shopify.com/s/files/1/0174/1800/files/galactic_unicorn_schematic.pdf?v=1667910096 I am getting a little lost already, tho. Would you know what ports I should be providing to something like:
matrix = rgbmatrix.RGBMatrix(
width=32, bit_depth=4,
rgb_pins=[board.D6, board.D5, board.D9, board.D11, board.D10, board.D12],
addr_pins=[board.A5, board.A4, board.A3],
clock_pin=board.D13, latch_pin=board.D0, output_enable_pin=board.D1)
?
tulip sleet
Does someone have a link to the CP6 source?
clone the repo and then git checkout 6.3.0 to get the last release or 6.3.x to get the branch that 6.30 was made from. If you want to build it you will need to fetch the submodules, and also maybe choose gcc 10.something
tulip sleet
I got a Galactic Unicorn and would like to use it with CircuitPython. I'm mostly...
RGBMatrix drives HUB75-style matrices. It is not clear to me at all that the Unicorn is a HUB75-style setup
btw, #help-with-circuitpython is probably better for this, and we ask that don't ping folks unless you are already in a convo with them
midnight ember
Someone is having issue updating their NRF52840 on linux using the new bootloader update learn guide. #help-with-circuitpython message
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/1Zx4UF48QN6YoRq-qVAJgdsfr-5KW401aJPTYwjxMQns/edit -- I look forward to everyone's updates!
tulip sleet
@slender iron I think I will cancel some of the simplest backport merge jobs so we don't have to wait 30 hours for them to finish.
next time I will merge more of the cherry-picks into a single PR
slender iron
fine with me
onyx hinge
Missing the meeting today. Group hug plus one for danh for all the back port fixes
brazen hatch
I am pretty sure the solution to this is just a module revision that has a capacitor on 3v3 and a board revision that has a better regulator.
As a note, with the module, regardless of if it was powered with 5V or 3V the issue occured.
turbid radish
We will likely keep it out on Mondays
midnight ember
Welcome back Anne ❤️
midnight ember
Thanks to everyone who worked on the newsletter over the past weeks, they did a great job.
midnight ember
Woot! Can't wait to read the guide. Thank you for the backporting devs too.
Sounds like a fun time, nice one Deshipu.
idle owl
@stuck elbow For your In the Weeds topic, are you referring to the core, the libraries, or both?
stuck elbow
<@351136009922019331> For your In the Weeds topic, are you referring to the core...
both
idle owl
Got it, thanks.
midnight ember
You can get your own health data out of a fitbit api, you can't do that with an apple watch. So this is a semi-open source way to work with your own personal watch health data.
ember iris
with the amount github changed the authentication process, that guide update must have taken ages! But goodness it'll be a huge help to have it working with the access token setup
idle owl
with the amount github changed the authentication process, that guide update mus...
That page was the least difficult, to be honest. 😄
The 100+ screenshots that went into the entire update on the other hand......
midnight ember
Nice job on all the projects including the arc weld and grind. 🧑🏭
ember iris
oh wow yeah that would take ages to process and organize
brazen hatch
The access tokens are such a pain to get right on linux.
I just have them in my dotfiles and copy the .gitconfig and .gitcredentials.
Making the files on your own is easier than doing it with commands.
idle owl
You're not the first person to say/do that.
brazen hatch
Yea I also have my own gitea instance, so it would be double the pain to do manually.
idle owl
Oof yeah.
tulip sleet
https://learn.adafruit.com/adafruit-matrixportal-s3 guide is now live
midnight ember
I'll actually agree. As a beginner the type annotations were labeled as a good first issue but I really struggled with it during hacktoberfest. You must have previous python experience to really tackle those.
yeah the ones I tried doing had like 50 types annotations to do per file
because the ones with only a few were taken out on like day 1
lone axle
There are still a few that have a limited number of types required. But it would definitely be good to remove 'good first issue' tag from any that are the more gnarly ones.
https://github.com/adafruit/Adafruit_CircuitPython_MPRLS/issues/16 and https://github.com/adafruit/Adafruit_CircuitPython_MPL3115A2/issues/18 have fewer than 10 instances each. I think there are a few more. I'll try to work through that list this week or next and remove labels from the bigger ones.
There are also a few that have PRs with requested changes but have been dormant for a while since the review, those would be good candidates to remove the label from as well in most cases.
midnight ember
It wasn't an NRF52840 🤦♂️ It was a SAMD51 Metro M4 with airlift.
turbid radish
Thanks all
gilded cradle
Thanks everyone
lone axle
Thanks for hosting Liz, Have a great week everyone!
midnight ember
Thank you Liz, great job.
ember iris
Thanks all, hope everyone has a great week!
candid sun
thanks folks!
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/1RMdYD7ABzChlYjE7HK4RsJOmBiQUU2Vo0iU61Q5Lus0/edit
spare jacinth
i just found a little bug, perhaps already reported, trying to combine 2 f-strings raises a SyntaxError: py print( f"Hello" f"World" ) funny enough, all other combinations (f + no-f, no-f + no-f and no-f + f) work
slender iron
f-strings are a bit of a hack because they get transformed to a string format call
could be related to that
idle owl
i just found a little bug, perhaps already reported, trying to combine 2 f-strin...
I ran into that recently. I think @tulip sleet had some idea why, but perhaps I'm misremembering.
tulip sleet
tulip sleet
@onyx hinge I don't know how "back" you are. Do you have a comment about https://github.com/adafruit/circuitpython/pull/8189 ? We could put it in 8.2.1 if you think it's fine for all Espressif. Or do you think it needs testing on other ones? Can wait until tomw if you'd rather not think about it now.