#circuitpython-dev
devout jolt
Question: is there a mechanism allowing one to compile-in a code.py and libraries, kind of like Micropython's manifest files? (https://docs.micropython.org/en/latest/reference/manifest.html ) I know that CircuitPython's "/frozen" libraries are the equivalent, but I have two questions on that:
- Is there a way to add a frozen code.py?
- Can single-use frozen libraries / code.py live in the board directory instead of polluting "/frozen", with an appropriate define in "mpconfigboard.mk"?
tulip sleet
Question: is there a mechanism allowing one to compile-in a code.py and librarie...
yes, here are some boards with their own frozen code:
ports/nordic/boards/challenger_840/mpconfigboard.mk:11:FROZEN_MPY_DIRS += $(TOP)/ports/nordic/boards/challenger_840
ports/raspberrypi/boards/solderparty_rp2350_stamp_xl/mpconfigboard.mk:14:FROZEN_MPY_DIRS += $(TOP)/ports/raspberrypi/boards/solderparty_rp2040_stamp
ports/raspberrypi/boards/solderparty_rp2350_stamp/mpconfigboard.mk:14:FROZEN_MPY_DIRS += $(TOP)/ports/raspberrypi/boards/solderparty_rp2040_stamp
ports/raspberrypi/boards/solderparty_rp2040_stamp/mpconfigboard.mk:14:FROZEN_MPY_DIRS += $(TOP)/ports/raspberrypi/boards/solderparty_rp2040_stamp
You could change the default code.py print("Hello World") to instead import some frozen module
.
devout jolt
yes, here are some boards with their own frozen code:
```
ports/nordic/boards/ch...
Thanks!
brazen hatch
The sensor may be cooked. I had a 2640 do the same to me recently.
Try switching to another camera.
tulip sleet
The sensor may be cooked. I had a 2640 do the same to me recently.
Try switching...
It works fine on older builds, so it's not the sensor. This is mysterious. I am still narrowing down the problem. It's not the autofocus loading -- I removed that from both and it still works.
manic glacierBOT
wanton dirge
Hey @slender iron , I'm getting the PR in but seeing some CI fails because it looks like my local pre-commit installation wasn't catching format issues for some reason. I'll try to get the CI steps passing today.
slender iron
Hey <@252717193496756235> , I'm getting the PR in but seeing some CI fails becau...
Nice! I'm around to help if you need extra eyes
wanton dirge
Is there something I should add the name of my port to to get this CI scheduler run to pass?
modest hazel
Hi! I'm struggling to get a custom C module to show up in circuitpython. It was working ok in CP7.x, but now that I'm updating everything to CP9, it's not showing/registering, and I'm hoping someone can help me out. The module is _lightshow, and it's here: https://github.com/buildwithpiper/circuitpython/tree/piperCP9.0.0/shared-bindings/_lightshow.
I used MP_REGISTER_MODULE and cleaned the build, but it's not showing when I call help("modules")
devout jolt
Hi! I'm struggling to get a custom C module to show up in circuitpython. It was...
Are you seeing the files in shared-bindings/_lightshow being compiled? I suspect not. I think you also need to add _lightshow to /py/circuitpy_defns.mk
modest hazel
okay, I'll give that a try, thank you!
devout jolt
You could change the default code.py `print("Hello World")` to instead import so...
And it doesn't look like there' s a way to have a board-specific code.py that gets frozen in, beyond doing Makefile hacking like:
DUMMY_VAR=$(shell sed 's/\print(\\"Hello World!\\")/import myfrozenlib/' filesystem.c > filesystem.c.new && mv filesystem.c.new filesystem.c)
okay. I'm stuck again. I think I'm getting closer, but I'm getting this error now that it's picking up the files and trying to build them:
In file included from ../../shared-module/_lightshow/__init__.c:48:
../../shared-module/_lightshow/__init__.c: In function 'lightshow_tick':
../../py/mpstate.h:337:40: error: 'mp_state_vm_t' {aka 'struct _mp_state_vm_t'} has no member named 'lightshow_singleton'
337 | #define MP_STATE_VM(x) (mp_state_ctx.vm.x)
| ^
../../shared-module/_lightshow/__init__.c:112:34: note: in expansion of macro 'MP_STATE_VM'
112 | lightshow_obj_t *lightshow = MP_STATE_VM(lightshow_singleton);
| ^~~~~~~~~~~
I'm not sure where I should be defining lightshow_singleton beyond what I've already done - so I don't know why I'm getting this error.
slender iron
okay. I'm stuck again. I think I'm getting closer, but I'm getting this error ...
there is a struct that this macro expands out to maybe mp_state_vm?
modest hazel
Like this?
lightshow_obj_t *lightshow = MP_STATE_VM(lightshow_singleton);
if (!lightshow) {
lightshow = m_new_obj(lightshow_obj_t);
lightshow->base.type = &lightshow_type;
lightshow = gc_make_long_lived(lightshow);
MP_STATE_VM(lightshow_singleton) = lightshow;
}
Source: https://github.com/buildwithpiper/circuitpython/blob/piperCP9.0.0/shared-bindings/_lightshow/LIGHTSHOW.c
For background, this is for a 8x8RGB matrix that is driven using shift registers and closely modeled after the _pew module. Admittedly, I'm a little out of my depth being this deep into it, but it's a requirement for our base firmware.
slender iron
lemme find it
Looks like the new way to get it into there is with MP_REGISTER_ROOT_POINTER. Search for an example of that
modest hazel
okay, I added MP_REGISTER_ROOT_POINTER(mp_obj_t lightshow_singleton); to the end of shared-bindings/_lightshow/LIGHTSHOW.c, but that didn't seem to have any effect - should I be somewhere else?
modest hazel
😦 Same error unfortunately - I ran make clean at the top directory, then make -C mpy-cross and make clean BOARD=... in the port directory.
tulip sleet
And it doesn't look like there' s a way to have a board-specific `code.py` that ...
I think that's right
modest hazel
I think that's right
In my situation this doesn't seem to be an issue. I just can't seem to get it to register the singleton into MP_STATE_VM
At least I think that's what the issue is currently.
tulip sleet
sorry, I was responding to someone else!
modest hazel
oh, sorry!
tulip sleet
In my situation this doesn't seem to be an issue. I just can't seem to get it t...
There is a pre-compilation step that looks for all the `MP_REGISTER_ROOT_POINTER() definitions. If that source file is not getting scanned, then it won't go into the MP_STATE_VM
i don't remember where the list of files is specified, but the _lightshow directory needs to be listed somewhere to get that to work.
I don't think it's a special list, but if you haven't added it to the right .mk file yet, then it won't be preprocessed to find the registration
modest hazel
Okay - that's helpful - maybe the file needs to be listed explicitly in /py/circuitpy_defns.mk?
I think I added init.c there, but not LIGHTSHOW.c
tulip sleet
make it be set up like all the other shared-bindings stuff
right, it needs to be compiled. Look at some other things that use MP_REGISTER_ROOT_POINTER, like, say, keypad
modest hazel
Okay, I really appreciate the lead. I'll start trying to plug it in.
tulip sleet
in general don't use ALLCAPS for a filename, btw
tulip sleet
@slender iron https://forums.adafruit.com/viewtopic.php?t=213823 user installed PSRAM on RP2350 board and I2C started to fail. is PSRAM orientation correct? Do they need a new build?
devout jolt
<@252717193496756235> <https://forums.adafruit.com/viewtopic.php?t=213823> user ...
physical orientation looks correct, going by the Eagle board file and PSRAM datasheet
slender iron
@tulip sleet replied. not sure why it'd impact i2c
tulip sleet
<@329766224093249548> replied. not sure why it'd impact i2c
me neither. Is there built-in support or does blakebr have to make a new build?
slender iron
it should detect it automatically
onyx hinge
@tulip sleet I recently soldered a psram to my feather 2350 and it worked. Fwiw
Gpio8 (psram cs) is not shared with anything so it should be fine
manic glacierBOT
manic glacierBOT
slender iron
<@&356864093652516868> Our community meeting will be in 10 minutes in the CircuitPython voice channel. Talk with you soon!
manic glacierBOT
austere acorn
not sure how we'd silence cats anyway 😼
lone axle
Thanks for hosting Scott. Have a great week everyone.
onyx hinge
thanks Scott. Short & sweet
lone axle
Does anyone have experience or insight with Github rate limiting? Digging into this adabot tests issue I'm finding that the X-RateLimit-Remaining header is inexplicably not containing a value, which gets defaulted to 0 and leads to logic in adabot looking for the X-RateLimit-Reset header, but that one also has no value. The code is currently written to assume that it will have a value in this scenario so when it's missing a KeyError is raised.
Strangely the request imediately preceeding the one that fails does contain the X-RateLimit-Remaining header and it still has a large value 4952 remaining out of 5000 limit. But for some reason on the following request the headers exist in the response but don't seem to have a value. Link to actions output: https://github.com/adafruit/adabot/actions/runs/11112781607/job/30875604335#step:7:47
The request where they have no value gets a 200 status code and it's content looks like what is expected for the response, just the headers seem to be "messed up" in this way where the keys exist without values.
manic glacierBOT
lone axle
The docs here https://docs.github.com/en/rest/using-the-rest-api/troubleshooting-the-rest-api?apiVersion=2022-11-28#rate-limit-errors make mention of the retry-after header, which similarly has a key in the headers dict but no value. I don't see any information about what it means if the header exist but without a value.
onyx hinge
@lone axle I looked at the output of the actions and am scratching my head over it
@lone axle it looks like this line is only hit once you're already having an exception .. I can't find any evidence that the quiet= parameter ever existed to logger.error()... I would remove quiet=
E TypeError: _log() got an unexpected keyword argument 'quiet'
@lone axle tests/unit/test_common_funcs.py patches some get function or method, see mock_github_get inside test_is_new_or_updated.
If it's this get that's being used, the expected headers won't be there. You could try adding them (result.headers['key'] = 'value') and see if it fixes CI
scratching my head how this ever worked though
but I don't see this mock being used in the other tests so it still doesn't feel like a full explanation
(for anybody not familiar with the terminology, in this case a mock is a replacement for a function or method that makes it do something different during testing)
- remaining = int(response.headers.get("X-RateLimit-Remaining", 0))
+ remaining = int(response.headers.get("X-RateLimit-Remaining", 99))
```(in github_requests.py) you could also try defaulting the number of requests remaining to 99 and cross your fingers. then it wouldn't enter the part where it tries to get the time until rate limit reset
upper radish
(for anybody not familiar with the terminology, in this case a mock is a replace...
(I lurk for tidbits just like this 🙂)
lone axle
<@382939733107408897> it looks like *this* line is only hit once you're already ...
I ended up making a test branch to work on and I have removed the quiet argument in that branch https://github.com/adafruit/adabot/tree/testing_fixes. It is a mystery to me why it was there or how it ever worked in the past as well. I also could not find any documentation that made reference to quiet. I will try giving it a larger default. If nothing else I'm curious if subsequent requests start having numerical values for those headers again after the one where they are seemingly blank.
orchid basinBOT
lone axle
A little bit more info about the adabot / github rate limit issue discussed above: I actually mis-read a detail the headers print out yesterday. It's not the case that the headers contain keys without values as I stated before. What's actually the case is there is a Access-Control-Expose-Headers header which contains a long string of comma seperated strings which are a list of other Headers that are presumably meant to be present in the response. I was seeing X-RateLimit-Remaining and X-RateLimit-Reset in that list without a colon and a value and mis-interpreting it.
So those X-RateLimit headers seem to be not present in the response, but are listed in the. They don't have keys or values, which does make sense as the exception that bubbles up is a KeyError.
Another small difference I just noticed in the headers: For the pen-ultimate request (that succeeds and contains all expected headers) one of the headers present is this: 'X-Accepted-OAuth-Scopes': '', But in the following request which succeeds but seems to be missing those headers the value is different: 'X-Accepted-OAuth-Scopes': 'repo' I'm not sure if it's really related or not though. Perhaps it's just due to the different resources being requested.
Maybe this is a wider spread issue with github. Just found this discussion: https://github.com/orgs/community/discussions/140079 which appears to be about the same symtoms.
I am thinking one option to resolve it is to keep local variables storing the value of the relevant headers. Then if we find they are missing from the response just refer back to the ones we stored from the previous request and "manually" decrement the remaining value to account for the requests we send that don't contain the response header. Based on the discussion and which request it's failing for us on, it seems like requests to /releases endpoints are the ones with this issue currently.
tulip sleet
I am thinking one option to resolve it is to keep local variables storing the va...
You could open a ticket with GitHub about this. We do pay them.
base don the discussions link, it sounds like a change
lone axle
I'd be happy to do that, but I'm not sure where or how to do so. Although helpfully it does appear they've documented it here: https://docs.github.com/en/support/contacting-github-support/creating-a-support-ticket
Would it be worth trying to create a small reproducer before opening the ticket? It's buried rather deep within the Adabot code.
tulip sleet
I'd be happy to do that, but I'm not sure where or how to do so. Although helpfu...
If the problem can be gleaned from logs, that's often enough for them. But if you can show the problem with a URL you put in the browser or via curl or whatever, that would be better
lone axle
Okay, I was able to distill it down to this reproducer:
import os
import requests
no_token_resp = requests.get("https://api.github.com/repos/adafruit/Adafruit_NeoPixel/releases/latest")
print(no_token_resp.headers['X-RateLimit-Remaining'])
auth_headers = {'Authorization': 'token ' + os.getenv('GITHUB_TOKEN')}
with_token_resp = requests.get("https://api.github.com/repos/adafruit/Adafruit_NeoPixel/releases/latest", headers=auth_headers)
print(with_token_resp.headers['X-RateLimit-Remaining'])
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1-hJPXRQZXHs7SzZVYPoi8s2IM_5xo_gtMdRPCf_IZC0/edit?usp=sharing
slender iron
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is...
@lone axle are you working on fixing adabot's bundle update?
sinful spruce
How do you specify what state machine for the PIO's on the 2040?
lone axle
<@382939733107408897> are you working on fixing adabot's bundle update?
yes, I think https://github.com/adafruit/adabot/pull/378 will fix the issue we saw last week with the community bundle. However there seems to be something strange going on with some responses from the Github API now that are causing the tests to fail. It's unclear at the moment if it's a bug or intentional change from Github (albeit undocumented if so).
I have in mind a way to change the logic in Adabot to hopefully be able to work around the issue of the missing headers in the responses. I've have not attempted it yet, but do intend to.
Oh, actually I hadn't noticed the build job failed today too
slender iron
yes, I think https://github.com/adafruit/adabot/pull/378 will fix the issue we s...
Ok, I'll let you keep fixing it. Let me know if you need me to help
it's failed the last few days apparently
lone axle
KeyError: 'x-ratelimit-reset' this is the same root cause of the test failing though
So hopefully if/when the logic in there is changed it will resolve it for anywhere that this specific issue may pop-up.
slender iron
How do you specify what state machine for the PIO's on the 2040?
You don't. It determines it automatically
slender iron
So hopefully if/when the logic in there is changed it will resolve it for anywhe...
are you changing it or should I?
lone axle
are you changing it or should I?
Either way is fine with me. I am eating lunch and need to take care of a few other things afterward before I can work on it. If you are able and want to take a shot sooner you can.
slender iron
kk, will do shortly
tiny peak
@lone axle in case it might be relevant to your rate limit thing... I noticed that when I created a PR against adafruit/circuitpython a couple days ago, and when I pushed an additional commit to that branch, GitHub kicked off 500+ workflow actions that ran for about 45 minutes each time. Perhaps that has nothing to do with the problem you are discussing.
... but, it does seem like the kind of thing that might put stress on rate limit quotas.
slender iron
I think the error might be that it no longer includes that header
lone axle
<@382939733107408897> in case it might be relevant to your rate limit thing... I...
Thank you, it's something to look into. But my initial instinct is that they're unrelated. AFAIK the core repo actions don't make requests the Github API for anything, though they do a lot of other building and testing work. I don't think that adabot is actually hitting the rate limit because there are still 4900+ left in the last request that does show a value.
lone axle
I think the error might be that it no longer includes that header
I believe it's specific to /releases endpoints. Those are not returning any of the X-RateLimit headers. My idea to resolve was keep local variables with the values when we do receive them and then if they're missing just keep track with local variables. Then update them as needed for requests that we do recieve values on.
If there are tasks that need to request only /releases endpoints that won't work though because we won't ever receive the values. The tests which I have looked most closely into so far have a mix of lots of non /releases and a few that are /releases so I think that approach can work for them.
Thinking about the build process though I'm now questioning whether that really does ask for anything other than /releases endpoints.
tiny peak
FWIW, poking around in circuitpython/.github/{actions,workflows}, I see several instances of uses: actions/upload-artifact@v4, which might be hitting the github API? Perhaps multiple times per port * board build?
I might be totally wrong about that though. This just caught my eye because one of the checks for a commit I pushed yesterday failed in a way that looked kinda mysterious. Like, maybe it had to do with some kind of rate limit or availability thing. IIRC, the check log said something about an error when attempting to upload a build to S3, so maybe it's unrelated.
slender iron
the S3 thing happens every so often. usually a retry fixes it
lone axle
Ahh, I just realized, when I tested with the reproducer code posted above I used my own private API token and saw the same behavior with the missing headers. Mine is definitely not used by anything with the core repo actions so I think the RateLimit header thing isn't tied to the core workflows.
willow totem
hey @lone axle , you got the adabot problem, with API rate limit headers missing. I've caught up on above (skimmed though), and I know the circuitpython library checking code works a bit differently...
Did anyone submit a support issue yet, I'm about to otherwise?
I'm just running it today for ardu libs, have patched github_requests.py to ignore if no rate info and continue on.
I play in this branch: https://github.com/tyeth/adabot/tree/arduino-ci-ver-bump
or this commit:
https://github.com/adafruit/adabot/commit/64e1b06d6090147c61c6d52ad29ea000d61be227
Also left a trail here: https://github.com/orgs/community/discussions/24760#discussioncomment-10812653
slender iron
hey <@382939733107408897> , you got the adabot problem, with API rate limit head...
I'm changing the code to assume we have requests left if the header is missing
maybe they removed the limit on that endpoint
lone axle
hey <@382939733107408897> , you got the adabot problem, with API rate limit head...
I did create a support ticket for it https://support.github.com/ticket/personal/0/3029134 (I'm not sure if it lets other org members see it?)
slender iron
(I can't see it)
willow totem
sorry sound on mute, I got distracted 🙂
lone axle
You might be right about opting that endpoint out of the limit. It doesn't appear to be counting a usage for that endpoint any more with a quick test locally. I'm getting response data and upon checking remaining count with another request elsewhere it isn't subtracting one for the /releases request.
willow totem
interesting
tulip sleet
<@329766224093249548> <https://github.com/adafruit/adabot/pull/379>
merged!
i am behind on email
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
turbid radish
Please post your Ask an Engineer questions beforehand in #1280611628081221785 channel.
onyx hinge
sigh, adafruit playground seems to have eaten a bunch of my work this morning 😕
I'd written a whole code-explaining section on https://adafruit-playground.com/u/jepler/pages/e-ink-countdown-in-circuitpython-with-custom-stand and .. poof!
I wonder if it's related to having had multiple tabs open on the editor; I thought this was correctly managed but maybe that's only on learn.
slender iron
sigh, adafruit playground seems to have eaten a bunch of my work this morning 😕
bummer! have you talked to internal folks about it?
onyx hinge
no, I haven't. I decided that it was a good moment to take a break.
slender iron
@heady bolt I can help you here since you are compiling circuitpython yourself
you can do a DEBUG=1 build of circuitpython and esp builds will log to UART all of IDF logging
heady bolt
you mean make BOARD=x DEBUG=1?
slender iron
correct
heady bolt
i wish i could figure out how to get into bootloader mode on the esp32 s3 consistently. i have a boot and reset button . ive tried pressing boot, pressing boot and reset at the same time, double clicking reset, unplugging holding boot and plugging in. none seem to work
heady bolt
ive gotten it in boot loader mode plenty of times, it just doesnt work consistently it seems. its a custom PCB
slender iron
holding boot while pressing reset should get you into the rom bootloader
heady bolt
just tried that, still nothing
slender iron
probably worth checking the board design then
heady bolt
holding boot while pressing reset does get me access to the drive
but not into where i can copy a new uf2 file
usually it creates a volume called BOOTESP32 or something to that effect
slender iron
holding boot while pressing reset does get me access to the drive
what pin is your boot button connected to? the rom bootloader shouldn't show as usb drive at all
merry hamlet
Hi, moving this question over from #help-with-circuitpython :
I’m attempting to build circuitpython from source because I have some minor modifications I want to test out. However, it takes around half an hour to even get to the point where the compiler complains about my silly syntax errors on my fairly old laptop. Is there a simple way to tell the build system to reduce the amount of code it attempts to compile? I’m already building only for the specific board I want to use for testing, and I’m building from the raspberrypi port directory.
I’ve been able to compile an unmodified version of the repo by the way. But that took just as long, or probably longer.
slender iron
Hi, moving this question over from <#537365702651150357> :
I’m attempting to bui...
It should be fast after the first run. What command are you running to build it?
what CPU is in your laptop?
tulip sleet
does your laptop have a spinning hard drive or a solid state drive?
merry hamlet
It should be fast after the first run. What command are you running to build it?
I've set up a task in vscode to build via wsl, like so:
{
"version": "2.0.0",
"tasks": [
{
"label": "Build",
"type": "shell",
"command": "c:\\Windows\\System32\\wsl.exe --cd \\\"/mnt/c/path/to/circuitpython/ports/raspberrypi/\\\" make -j12 BOARD=seeeduino_xiao_rp2040 V=1",
// "command": "make -j12",
"group": {
"kind": "build",
"isDefault": true
},
"presentation": {
"reveal": "silent"
},
"problemMatcher": "$gcc",
"options": {
"env": {
"TARGET": "${workspaceFolderBasename}"
}
}
}
]
}
I'm on a gen 8 i5, so I don't expect it to be fast, but this was a bit slower than I was hoping for.
merry hamlet
does your laptop have a spinning hard drive or a solid state drive?
I have an SSD (but it isn't new either), and both it and the CPU are at 100% resource utilization for a long time while building.
tulip sleet
WSL has slow disk access particularly older WSL
a gen 8 i5 should be fine. I do builds all the time on a gen 8 i7, and they take much less time than that (like 1/30)
merry hamlet
I reinstalled and set up a completely new WSL2 image specifically for this, so the version shouldn't be a problem. But the disk access speed could of course still be an issue.
tulip sleet
how much RAM is allocated for WSL?
merry hamlet
how much RAM is allocated for WSL?
Hmm, that's a good question. I'll try to figure it out
tulip sleet
I can't remember if you can specify how many cores and how much RAM
cat /proc/meminfo
merry hamlet
I can't remember if you can specify how many cores and how much RAM
It has access to all cores at least. The CPU is at 100% for a very long time
tulip sleet
also cat /proc/cpuinfo
merry hamlet
cat /proc/meminfo
MemTotal: 3948580 kB
MemFree: 2794812 kB
MemAvailable: 3213336 kB
Buffers: 44036 kB
Cached: 528624 kB
SwapCached: 0 kB
Active: 496388 kB
Inactive: 260628 kB
Active(anon): 2328 kB
Inactive(anon): 185152 kB
Active(file): 494060 kB
Inactive(file): 75476 kB
Unevictable: 0 kB
Mlocked: 0 kB
SwapTotal: 1048576 kB
SwapFree: 1048576 kB
Dirty: 7380 kB
Writeback: 0 kB
AnonPages: 182540 kB
Mapped: 137996 kB
Shmem: 3124 kB
KReclaimable: 102084 kB
Slab: 156684 kB
SReclaimable: 102084 kB
SUnreclaim: 54600 kB
KernelStack: 4004 kB
PageTables: 5012 kB
NFS_Unstable: 0 kB
Bounce: 0 kB
WritebackTmp: 0 kB
CommitLimit: 3022864 kB
Committed_AS: 759812 kB
VmallocTotal: 34359738367 kB
VmallocUsed: 25712 kB
VmallocChunk: 0 kB
Percpu: 3744 kB
AnonHugePages: 71680 kB
ShmemHugePages: 0 kB
ShmemPmdMapped: 0 kB
FileHugePages: 0 kB
FilePmdMapped: 0 kB
HugePages_Total: 0
HugePages_Free: 0
HugePages_Rsvd: 0
HugePages_Surp: 0
Hugepagesize: 2048 kB
Hugetlb: 0 kB
DirectMap4k: 117760 kB
DirectMap2M: 4009984 kB
DirectMap1G: 8388608 kB
on my 8th gen i7, to compile raspberry_pi_pico:
real 0m28.957s
user 1m51.553s
sys 0m39.243s
so meminfo says 4GB
lscpu | awk -F ":" '/Core/ { c=$2; }; /Socket/ { print c*$2 }' will tell you how many cores
I haven't tried WSL for a while, but it was more like 3-4x slower, not 30x slower
and it was due to slower disk access
merry hamlet
Any idea as to why this program only works for pin 5. The other pins 6 and 9 do not generate a square wave. I want to use mltiple state machines.
import time
import rp2pio
import board
import adafruit_pioasm
squarewave = """
.program squarewave
set pins 1 ; Drive pin high and then delay for one cycle
set pins 0 ; Drive pin low
"""
assembled = adafruit_pioasm.assemble(squarewave)
sm1 = rp2pio.StateMachine(
assembled,
frequency=1000 * 1 * 4000,
first_set_pin=board.D5,
)
sm2 = rp2pio.StateMachine(
assembled,
frequency=1000 * 1 * 24000,
first_set_pin=board.D6,
)
sm3 = rp2pio.StateMachine(
assembled,
frequency=1000 * 1 * 1000,
first_set_pin=board.D9,
)
print(dir(sm1))
print("real frequency", sm1.frequency)
print(assembled)
time.sleep(100000)
slender iron
Any idea as to why this program only works for pin 5. The other pins 6 and 9 do ...
Does it work if you use the same frequency on all three? This is better to ask in #help-with-circuitpython first.
sinful spruce
Does it work if you use the same frequency on all three? This is better to ask i...
Even when set at the same frequency it fails on the pins 6 and 9.
slender iron
Even when set at the same frequency it fails on the pins 6 and 9.
you don't get an error but it doesn't work? what board are you using? (moving to #help-with-circuitpython )
sinful spruce
~~you don't get an error but it doesn't work? what board are you using?~~ (movin...
Sorry my setup was wrong. It works as expected.
heady bolt
@slender iron im still getting sporadic resets on the board and i have deployed the debug version of the custom cirtcuitpython build. I will run the code a few times and then eventually it stops responding and i get this "Unable to connect to COM18: could not open port 'COM18': PermissionError(13, 'Access is denied.', None, 5)"
the only way to get it back is to reset the board, but this hasnt really happened before. just strange
slender iron
@heady bolt you'll want to monitor the debug uart output to see what is happening
heady bolt
where/how would i do that?
tulip sleet
do you have a USB/serial adapter?
heady bolt
im connected to the pcb by usb
manic glacierBOT
onyx hinge
@slender iron ah I think this feather (esp32s3 with 4MB flash + 2 MB PSRAM) doesn't have bleio enabled so I wouldn't be hitting the BLE / deep sleep bug with my project.
slender iron
<@252717193496756235> ah I think this feather (esp32s3 with 4MB flash + 2 MB PSR...
ah! that'd explain it
onyx hinge
Are there humans who understand the template language of circuitpython.org? I noticed the list of built in modules is showing wrong, with extra spaces and a trailing comma
this fixes it but .. what ???
diff --git a/_includes/download/board.html b/_includes/download/board.html
index a497f0d18..7a5ed7b74 100644
--- a/_includes/download/board.html
+++ b/_includes/download/board.html
@@ -146,7 +146,7 @@
{% if version.modules %}
<p>
Built-in modules available:
- <span class="download-modules {% if version.stable %}stable{% else %}unstable{% endif %}">{{ version.modules | join: ', ' }}</span>
+ <span class="download-modules {% if version.stable %}stable{% else %}unstable{% endif %}">{{ version.modules | join: " " | split: " " | join: ', ' }}</span>
</p>
{% endif %}
{% if version.frozen_libraries and version.frozen_libraries.size != 0 %}
scratch that .. the diff does nothing. my local version builds right without change while the current version is different
it's also back from 9.0.4 days and the ruby bundler hasn't been updated, so maybe it's version churn or something
yes it was a bug introduced somewhere along the line...
slender iron
is it jinja?
onyx hinge
it's liquid
i found the bug but I haven't figured out how to fix it yet. there are two problems, one is that the "don't add comma after last item" logic is wrong; the other is that the template language is inserting a whitespace that shouldn't be there.
manic glacierBOT
lone sandalBOT
heady bolt
@tulip sleet can you point me to how i can monitor the uart output to capture why this board keeps disconnecting?
slender iron
<@329766224093249548> can you point me to how i can monitor the uart output to c...
are you using circuitpython? what board?
haha, I asked you yesterday 🙂
it is custom
do a debug build and then monitor the TXD pin
for uart0
blissful pollen
@slender iron I think it makes sense to take effects out of draft now. I'm not sure how many extra effects I can get done now but those can be other PRs easily enough and gives a guideline to others. Does that make sense?
slender iron
<@252717193496756235> I think it makes sense to take effects out of draft now. I...
Sure! I was just reading the comments on that PR now
I'll do a review too
lone sandalBOT
lone axle
The X-RateLimit headers have re-appeared in the responses from /releases endpoints from Github API.
and they are subtracting API requests from limit now. I guess it was unintentional that they were opted out of that for a few days.
tulip sleet
@onyx hinge @slender iron for the remaining 9.2.0 bugs, could I assign some to you two? e.g picodvi and usb host bugs to Scott, MP3 bug to jeff, etc. I can take the BLE and Matrixportal bug.
slender iron
yes but I might punt them
tulip sleet
I would understand that, if they are not blockers
lone sandalBOT
blissful pollen
Can anyone point me to what may be wrong for this error when I am building the docs.
audiodelays/__init__.pyi:17: error: Name "BlockInput" is not defined [name-defined]
For an area like:
//| self,
//| max_delay_ms: int = 500,
//| delay_ms: BlockInput = 250.0,
...```
The stubs build locally for me, I know other modules (synthio) use BlockInput as a type. I cannot see what I need to include so it builds.
stuck elbow
Can anyone point me to what may be wrong for this error when I am building the d...
there should be an import for it at the top of the file, inside a try block?
blissful pollen
This is in the core though, that's what I cannot figure out.
Ah think I have it. wifi uses ipaddress and just has ipaddress.IPv4Address prefacing it with the module it is defined in.
manic glacierBOT
manic glacierBOT
@dhalbert Thank you for giving this issue attention!
I understand issue 1. However, I am still clueless how I can specify the different pinouts from the male plug of the MatrixPortal M4 board to the female socket of the WaveShare RGB matrix. I thought, the only way to do so is to pass the full list of address lines in the order of the pins.
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
onyx hinge
@tulip sleet I don't know why the doc build is failing in that PR. Whether it is more readthedocs churn? There are multiple things in the log that say "warning". But I think this one is new:
WARNING: Calling get_html_theme_path is deprecated. If you are calling it to define html_theme_path, you are safe to remove that code.
lone axle
Heh, I was just coming to mention that looks like it's going to be an issue across all the libraries. https://github.com/adafruit/Adafruit_CircuitPython_Display_Text/actions/runs/11217110767/job/31177912073#step:2:1002
onyx hinge
yeah https://pypi.org/project/sphinx-rtd-theme/ was updated a major version just a few hours ago
tulip sleet
<@329766224093249548> I don't know why the doc build is failing in that PR. Whet...
There is something that just changed in rtd, I think. I am afk but will find the email from rtd about it
I think they sent us a warning but it was not clear it was applicable
lone axle
I will look into it and see if I can find the fix unless one of you is already on it?
tulip sleet
yeah https://pypi.org/project/sphinx-rtd-theme/ was updated a major version just...
I just forwarded you an email
onyx hinge
that seems to be something else, not 100% sure though
here's the announcement on their blog, covers the same info: https://about.readthedocs.com/blog/2024/07/addons-by-default/
lone axle
Removing that line does appear to work in a library as well. I'll prepare a patch that adabot can run to get it fixed in the libraries.
manic glacierBOT
tulip sleet
<@&356864093652516868> The weekly CircuitPython meeting is in less than two hours: 2pm US ET / 11am US PT. Ttyl! Add your notes: https://docs.google.com/document/d/1-hJPXRQZXHs7SzZVYPoi8s2IM_5xo_gtMdRPCf_IZC0/edit?usp=sharing
austere acorn
Ohai, could someone please help me understanding why this fails? https://github.com/adafruit/Adafruit_CircuitPython_GPS/pull/113
Besides having the number 13 in it, I can't figure it out 😅
wraith crow
Ohai, could someone please help me understanding why this fails? https://github....
I believe you've run into an issue that is being worked on #circuitpython-dev message
austere acorn
Ah, I see ... I guess I should have scrolled up in the channel, lolz
So there is nothing I have to do about it? It's getting fixed ?
wraith crow
You may be able to simply remove this line "html_theme_path = [sphinx_rtd_theme.get_html_theme_path(), "."]" from docs/conf.py as I think that's what foamyguy is planning on doing across all the libraries but I'm sure it will be solved pretty soon anyway.
austere acorn
Got it, thank you!
lone axle
There is a patch PR in to adabot now with the fix for it: https://github.com/adafruit/adabot/pull/381
For the moment it's fine that PRs are failing th actions due to that change in sphinx-rtd-theme. It should get resolved once this patch is run.
onyx hinge
and in fact news of its release just came out on the Python discord
onyx hinge
and thanks to Foamyguy for taking on that same readthedocs problem across the libraries!
lone axle
Thank you! I can run the patch this afternoon.
lone axle
Thanks for hosting Dan, have a great week everyone.
tulip sleet
yw, and thanks everyone for attending!
onyx hinge
@slender iron did you ever run into this? in a pico-sdk project I am generating video on core1. Something in there is doing a modulo, which calls into __aeabi_uidivmod through veneer, which I think means it goes into the slow flash .... ? anyway, I am getting video tearing because of that, I think.
what's weird is it was working until I made some changes in my CMakefile
slender iron
I have CP setup to crash if the flash is accessed. That'll tell you exactly where it is happening
this is why I want the compiler to be smarter
the veneer would be used if your main code is in ram
onyx hinge
well I guess my question is, do you know how to move __aeabi_udivmod into RAM?
the code is just doing a % modulo operation but I guess that is a function call
slender iron
I don't know if pico-sdk has a flag for it. you can check the map file to see where it is placed
it might be in ROM instead of flash
tulip sleet
maybe this is an accidental cache-line-boundary change?
onyx hinge
@slender iron ```#include "src/rp2040/hardware_structs/include/hardware/structs/mpu.h"
#include "src/rp2_common/cmsis/stub/CMSIS/Device/RP2040/Include/RP2040.h"
OK got it 🙂
lone axle
Some of the libraries have a different chunk in their docs/conf.py file that configures the theme: https://github.com/adafruit/Adafruit_CircuitPython_AMG88xx/blob/main/docs/conf.py#L152-L162 it will need a separate patch to target those ones due to the surrounding code being different enough to not apply.
Test run of the patch finished with 343 successful, 9 skipped for that different conf.py file, and 8 failed seemingly they're missing the docs conf file.
I'm going to start the real run in a moment and I'll check into the remaining failures, I think some may be newer minimal repo's without their cookiecutter files yet and no releases / not in bundle.
A handful of the ones that successfully commited then had this very generic actions failure.
I'm going to try re-running to see if maybe it was just a transient issue caused by having committed to so many repo's in a relatively short time window.
slender iron
<@252717193496756235> ```#include "src/rp2040/hardware_structs/include/hardware/...
Look at the other framebuf impl for rp2350
maybe I omitted it there
lone axle
This patch should cover the 9 with the alternative config files. https://github.com/adafruit/adabot/pull/382 though I think the tests take a while to complete so it won't be green for a bit.
light scroll
Thanks for the pointer to ../contributing. I As a retired engineer who spent far too much time in unfocused meetings at big companies, your focused & efficient meeting is a breath of fresh air. (even though the baby-project statu should have been first item. 😄 ❤️
(I hope to get around to contributing in a couple months)
slender iron
Thanks for the pointer to ../contributing. I As a retired engineer who spent far...
Glad you appreciated it! This is the place to get help when diving in.
lone axle
This library seems to have some issue with the github release action. https://github.com/adafruit/Adafruit_CircuitPython_CST8XX Both of it's releases failed that action. The error for the recent release is:
Error: Error: unexpected status code: 403
{"message":"Resource not accessible by integration","request_id":"9940:38FE6F:12BA518:145FF8A:6704567A","documentation_url":"https://docs.github.com/rest"}
It seems like maybe there is something different in the permissions or setup of the token that gets used inside of actions perhaps?
lone axle
This library has the same issue where the gh releases have all failed. https://github.com/adafruit/Adafruit_CircuitPython_PCA9554
The rest of the first group of 343 have all been released after the patched docs config.
crimson ferry
@lone axle the VEML7700 library has some kind of sphinx issue on a PR,... is it the same thing?
lone axle
<@382939733107408897> the VEML7700 library has some kind of sphinx issue on a PR...
Yes, I think if you merge main to your PR branch and then push, it should make it pass the actions.
crimson ferry
OK, I'll see if I can figure out how to do that, this was an in-place patch edit on github.com (I thought it would be easy. Narrator: It's never easy.)
crimson ferry
That worked.
manic glacierBOT
manic glacierBOT
orchid basinBOT
tulip sleet
@slender iron I have draft release notes for 9.2.0-beta.1 and can wait for any imminent PR's.
@onyx hinge 👆 too - I see some PR's by or of interest to you.
slender iron
I won't be able to review PRs until this afternoon. I wouldn't wait
we can always do another release
I intend on doing the picodvi bugs this afternoon too
tulip sleet
I intend on doing the picodvi bugs this afternoon too
i was just wondering whether you think the picodvi bugs will be easy. I could wait until tonight for the release.
but sure, I could do another beta soon anyway
orchid basinBOT
onyx hinge
@tulip sleet did you have other PRs in mind as ones I might be interested in?
tulip sleet
<@329766224093249548> did you have other PRs in mind as ones I might be interest...
nope, I think I mistakenly thought one was open when it was already closed
tulip sleet
<@329766224093249548> did you have other PRs in mind as ones I might be interest...
... ok, I realize I meant issues instead of PR's
9644 and 9662
onyx hinge
9644 I need to look at for sure
9662, no idea what to do. I don't think it's a code problem since my setup ran fine for days.
tulip sleet
i'm not sure either if you can't reproduce
tulip sleet
"needs retest" 🙂
onyx hinge
thanks
slender iron
i was just wondering whether you think the picodvi bugs will be easy. I could wa...
I think they will be easy but haven't done it yet so I don't know
tulip sleet
I think they will be easy but haven't done it yet so I don't know
I can just wait.
slender iron
👍
lone axle
I got a reply today on the Github ticket about the X-RateLimit headers, confirming the change we saw to the headers and the resolution a few days later:
Sorry for the delayed response—we're currently handling a high volume of support inquiries, which has impacted our response times.
I wanted to follow up on your concern. There was indeed a change in the behavior of the Releases endpoint around the time you mentioned, and it was resolved on October 3rd.
If you have any more questions or need further assistance, feel free to reach out. Happy coding! 🚀👨💻
Cheers,
Rocky
tulip sleet
I got a reply today on the Github ticket about the X-RateLimit headers, confirmi...
does "it was resolved" mean they reverted it?
does it work the old way now?
lone axle
That is my understanding of what they mean, and yes it's back to behaving the old way.
slender iron
@wraith crow just pinged you on github
@tulip sleet just sent out a PR for the MDNS fix. picodvi is all looked at too
tulip sleet
Yup, looking good. I think I will go ahead and make a beta release.
thanks for the quick fixes and diagnoses
slender iron
np! I want to get 9.2 stable
wraith crow
Scanned the code quickly, looks good to me, I'll give it some light testing tonight just for fun 😁
tulip sleet
@slender iron ok sure -- i'll merge and make a beta
tulip sleet
easier to test then too
I am marking this as an "incompatible change" because the hostnames can be somewhat different. Or is that an unusual case? I'm not sure.
slender iron
ya, it'll change the mdns hostnames
should we wait for 10? I assume these names are usually discovered instead of hard coded
tulip sleet
I think it's ok. The removal of all those LWIP board-specific names is also an incompatible change.
maybe a more surprising one.
wraith crow
Running out of flash on the c3s?
slender iron
mdns I presume
tulip sleet
these have dualbank turned on, does that make sense in the long run? Also ps2io is on, but they dont' have full USB, so maybe not very useful
slender iron
no, I think we'll want dualbank off for these with 10 too
I'm changing some mdns config to see if it helps size
tulip sleet
i have a guitar class in a few minutes, until 6pm PT, so not chomping at the bit for a release.
slender iron
kk, I'm about to leave too
wraith crow
I'm sorta glad the C3's didn't fit, I'll be much happier with working hostname/MAC addresses in the beta release
crimson ferry
the good news is... I haven't seen any mdns hostname mangling 🙂
(well, other than the -2, -3, etc. b/c the hostnames are the same - and, oh my, the hostnames and instance names increment differently, e.g., Adafruit QT Py ESP32S2-2 (cpy-t_qtpy_esp32s2-3ffe2e90-4))
wraith crow
I assume the goal of the unique MAC addresses in the host name is to avoid all the mangling issues 🤞 I built Scott's PR with fixed MAC address conversion:
cpy-bpi_picow_s3-7cdfa1f23d14 MAC Address:7c:df:a1:f2:3d:14
cpy-8r8_hacktablet-7cdfa1e15694 MAC Address:7c:df:a1:e1:56:94
crimson ferry
last three octets should be sufficient
there will be a lot of overlap in the first three (espressif owns 208 OUI ranges, raspberry pi only 7 ...out of 16+ million)
manic glacierBOT
wraith crow
I was thinking that also, but figured I'd let Scott decide, if he drops three of the octets he also has the option to add three more characters to the board name so your t_ would turn into ruit_ 😁 The practice of using the last three octets is what allowed my coding mistake to slip through, I converted part of the hex encoded mac address into an unsigned integer and then back to a hex value which just happened to turn out to be 6 didgets (hexets?) long ☹️
manic glacierBOT
manic glacierBOT
onyx hinge
@lone axle soooo it looks like adabot added some files it shouldn't have, due to testing before pushing. 🤯 ``` build_deps/mpy-cross-8.x | Bin 0 -> 1427128 bytes
build_deps/mpy-cross-9.x | Bin 0 -> 354520 bytes
and that made the build in the actual bundle repo fail 😠
the overnight release does actually have files so that's good I guess, but there's more work to be done.
@lone axle I tagged you for review on some PRs to fix this. Thank you for working on the adabot, everybody understands it's hard to get right on the first try!
lone axle
<@382939733107408897> I tagged you for review on some PRs to fix this. Thank you...
Thanks for the fix. Review is in. I do not have rights to merge in adabot the repo.
onyx hinge
@lone axle I can click merge on that PR in a bit. can you do https://github.com/adafruit/Adafruit_CircuitPython_Bundle/pull/476 too?
lone axle
Yep, done.
icy saffron
hi! I've been trying to get CircuitPython building for the ESP32-S3 (specifically for the Memento board) with ustack enabled so I can check how heavy my resource usage is, but enabling it in py/circuitpy_mpconfig.h causes linking to fail. is it not supported on Espressif platforms? the documentation doesn't include a supported-boards list for either it or uheap
onyx hinge
hi! I've been trying to get CircuitPython building for the ESP32-S3 (specificall...
It looks like uheap is definitely broken due to incompatible changes in the core. However, I can successfully compile for adafruit_magtag_2.9_grayscale with ustack enabled, so you may have done the procedure incorrectly.
CircuitPython almost always enables modules by editing a .mk file, not a .h file. It's necessary to manually make BOARD=... clean after changing mk files.
So I added the line CIRCUITPY_USTACK = 1 at the end of ports/espressif/mpconfigport.mk, then did make BOARD=adafruit_magtag_2.9_grayscale clean and finally make BOARD=adafruit_magtag_2.9_grayscale.
I'll load it on my board in just a second and see if it imports/works.
I'll investigate uheap a little more too and file an issue if it is for sure not working. It is likely due to the "split heap" code that @slender iron switched us over to in the last months...
icy saffron
It looks like `uheap` is definitely broken due to incompatible changes in the co...
interesting, thank you! is this information in the docs, perchance? :p
I had to guess how to enable it since I'm not much of a C developer
onyx hinge
When it comes to how-to documents they are often on the Adafruit Learn system. There's this (though it doesn't cover the need to make clean so it could be improved)
https://learn.adafruit.com/building-circuitpython/customizing-included-modules
You may want to include a particular module that is not included by default for the board for your board. You can customize which modules to include in your CircuitPython build by adding or changing settings in the mpconfigboard.mk file for your board.
icy saffron
alright
onyx hinge
>>> import ustack
>>> ustack.stack_size()
15360
>>> ustack.stack_usage()
948
```ustack does work on my board so hopefully you can get it going toolet me see if it'll work
where would I find a list of the allowed options for port makefiles?
onyx hinge
the closest to that is the source file py/circpy_mpconfig.mk which provides defaults and often small comment blocks about what can be enabled/disabled
CIRCUITPY_UHEAP ?= 0
CFLAGS += -DCIRCUITPY_UHEAP=$(CIRCUITPY_UHEAP)
```for instance from this plus my experience, I inferred that I could enable the `uheap` module by setting `CIRCUITPY_UHEAP` in the right `.mk` file. but without experience you wouldn't just guess it.I have to step away for now. If you're a github user you can track the uheap build problems on that issue. You can always ask more questions here or in #help-with-circuitpython and hopefully someone else can step in to answer.
icy saffron
mhm, thank you for your help!
onyx hinge
👋
icy saffron
it works, hooray
although the reported values don't actually seem to... change? weird
and max_stack_usage() just doesn't exist
tulip sleet
and `max_stack_usage()` just doesn't exist
MICROPY_MAX_STACK_USAGE is a separate flag
icy saffron
calling it os feels like a bit of a misnomer really, since most of what it actually does is filesystem stuff; it's not even under the Standard Libraries section with the rest of the CPython analogues
tulip sleet
and `max_stack_usage()` just doesn't exist
note that PYSTACK is enabled, so python stack frames are on a separate stack
icy saffron
what's the distinction? the reason I'm interested at all is because I had to increase the max size of pystack, but I don't really get what it does
tulip sleet
what's the distinction? the reason I'm interested at all is because I had to inc...
THe python stack frames could be on the C stack, but with PYSTACK enabled, they are on a separate stack.
manic glacierBOT
It turns out this diagram has several errors:
I am going to get that fixed!
The schematic is correct, of course:
The pin-out of the Waveshare panel is the same as the Adafruit panels. Pinouts of our panels are discussed here: https://learn.adafruit.com/32x16-32x32-rgb-led-matrix/new-wiring. Note you hav...
slender iron
THe python stack frames could be on the C stack, but with PYSTACK enabled, they ...
without pystack, large python frames get allocated to the heap instead of the c stack. so pystack prevents them from using the heap
takes the cats to the vet
tulip sleet
@slender iron in the circuitmatter README, if I do python -m circuitmatter test_data/recorded_packets.jsonl, I get
FileNotFoundError: [Errno 2] No such file or directory: '../esp-matter/connectedhomeip/connectedhomeip/credentials/test/attestation/Chip-Test-DAC-FFF1-8000-0000-Key.der'
it looks like I need to clone esp-matter next to circuitmatter?
I haven't yet done anything with esp-matter
slender iron
yup! sorry about that
tulip sleet
np, just wanted to make sure you knew
slender iron
it is documented under the chip-tool section
please update the readme as you find these things
tulip sleet
@slender iron did you have any trouble with cffi during the esp-matter install.sh? It's complaining it wants it to build wheels but it's not installed. from pip's point of view it is installed
manic glacierBOT
manic glacierBOT
orchid basinBOT
stuck elbow
doesn anybody know where common_hal_displayio_bitmap_load_row is implemeted? it's defined in shared-bindings/displayio/Bitmap.h but I can't find the implementation
onyx hinge
@stuck elbow looks like there's a declaration in Bitmap.h but no definition & no uses.
the implementation was removed a very long time ago in 2019
icy saffron
would there happen to be a way to interrogate a CircuitPython microcontroller about its capabilities that isn't "send Python over the raw REPL?" I'm considering remaking the vscode extension because it's pretty broken right now
a machine-readable list of USB vendor/device IDs would also be cool
tulip sleet
would there happen to be a way to interrogate a CircuitPython microcontroller ab...
If the CIRCUITPY drive is available, you can look at boot_out.txt to identify the device.
icy saffron
right
I suspect the problem will be finding the serial port corresponding with a USB drive
tulip sleet
See https://github.com/adafruit/Adafruit_Board_Toolkit and https://github.com/neradoc/discotool for identifying serial ports
the serial ports have certain names. If you have multiple devices connected, that is a bit more of a problem, but for one device, it's easy
using that library or that tool (they use the same techniques)
icy saffron
cool, thanks
icy saffron
See <https://github.com/adafruit/Adafruit_Board_Toolkit> and <https://github.com...
both of these use platform APIs, unfortunately; I'm not sure if I can easily do that from JS
tulip sleet
looking for something...
tulip sleet
both of these use platform APIs, unfortunately; I'm not sure if I can easily do ...
https://github.com/circuitpython/web-editor is what runs https://code.circuitpython.org, which asks the user to choose the serial port, using webserial on chrome-based browsers. Is that good enough/
tulip sleet
that code is all js, running in the browser
icy saffron
Node isn't quite the same as Chrome, though :p
let me see if I can find my code
icy saffron
the current one uses https://serialport.io/ I think
icy saffron
okay, I can get a vendor and product id from that
now I just need to find a node USB library that isn't broken
manic glacierBOT
slender iron
@tulip sleet I didn't have cffi trouble but esp-matter is a pain.
tulip sleet
i will check their issues, etc.
empty salmon
If I were to make a "very similar board" to an existing CircuitPython board but use 4MB flash in place of the original board's 2MB flash, will CircuitPython still load and work? or is CircuitPython expecting an exact match for the flash chip?
slender iron
i will check their issues, etc.
You shouldn't need to pip install everything to run the replay
slender iron
If I were to make a "very similar board" to an existing CircuitPython board but ...
with what micro? some ports expect exact match
empty salmon
Ah. Sorry about that.
@tulip sleet - I am working with the SAMD21 and used a 4MB flash rather than the 2MB. I did try to use the same vendor/manufacturer for the flash and stay in the same family of chip. (aka I used a W25Q32JVSS vs the W25Q16JVxQ for the itsy bitsy M0).
I wasn't sure if my problem is the flash size or if it's not close enough to the same model
empty salmon
Yeah. I was trying to cheat and reuse the itsy bitsy M0 since there will only ever be about 100 of this board in the wild.
I'll go ahead and build a new board.
tulip sleet
@slender iron @onyx hinge if you can approve https://github.com/adafruit/circuitpython/pull/9709, I will do the 9.1.0-beta.1 release
tulip sleet
thanks!
slender iron
caught me right before I left 🙂
tulip sleet
caught me right before I left 🙂
whew 🙂
onyx hinge
thank you! so excited to be at "beta" 💜
orchid basinBOT
atomic summit
Howdy folks 🙂
I've had reports from 2 different people in the last week or so stating similar to this...
Is there a known bug with wifi on feathers2 with circuitpython 9.1.x? I've been trying to get some of my older code working on 9.1.4 and realized that as soon as 'import wifi' runs things lock up. Can easily replicate it in the REPL, just import wifi. I tracked it down to the 9.1.0 release. Previous version 9.0.5 works, 9.1.0 locks up. To make things worse after power cycling the flash comes up empty and possibly corrupted.
Have there been any other reports of anything like this since 9.1x shipped?
I'm, flashing 9.1.3 on all of my boards on my test jigs, and I am not experiencing any wifi issues or flash corruption issues, but that's all in a controlled environment.
atomic summit
I can repo this on a FeatherS2 running 9.1.3... simply go to the repl and import wifi
and it hangs... doesn't come back. If I reset it, I don't get flash corruption, but I've only done this twice.
tulip sleet
I can repo this on a FeatherS2 running 9.1.3... simply go to the repl and `impo...
There was something that was fixed https://github.com/adafruit/circuitpython/issues/9486 rather more specific, but was provoked by import wifi. Please try 9.1.4, and also I am in the process of releasing 9.2.0-beta.1. Try that also.
atomic summit
There was something that was fixed <https://github.com/adafruit/circuitpython/is...
Oh, ok, I'll try that out and report back, thanks!
tulip sleet
Oh, ok, I'll try that out and report back, thanks!
OK, the 9.2.0-beta.1 links are up on circuitpython.org now
empty salmon
What (or where) is the process for requesting a USB PID for a new SAMD21 (M0) board?
empty salmon
@stuck elbow - all the hardware will be published. I use the MIT license for my stuff.
When I did a search on "SAMD21", I found references to PID requests within CircuitPython but they were from a few years back. Hence, my question.
Thanks.
tulip sleet
https://blog.adafruit.com/2024/10/10/circuitpython-9-2-0-beta-1-released/
Highlights of this release
- New
busio.I2C.probe()andbitbangio.I2C.probe()methods to check for a single device address. - Use new ESP-IDF I2C driver.
- ESP32-S3: implement
sdioio. - Incompatible change: Change default hostname for all Espressif boards to the ESP-IDF default, which is
espressif. Previous board-specific names were not applied consistently. Usewifi.radio.hostnameto set a custom hostname. - Incompatible change: Use default hostname for mDNS. Fix mDNS collision mangling.
stuck elbow
<@351136009922019331> - all the hardware will be published. I use the MIT licens...
then the website I linked should work for you
lone axle
For the Spirit Board that I am working on I created a helper class AnchoredTileGrid that extends TileGrid and adds the anchor_point and anchored_position behavior from Display_Text and a few other widget libraries.
I think it has the potential to be useful more generally than just for the Spirit Board. I'm curious if there is any interest in having those properties added directly to TileGrid within the core? Or creating a library to hold the AnchoredTileGrid class so that it's easy to re-use outside of this specific project.
manic glacierBOT
lone sandalBOT
tiny peak
For the Spirit Board that I am working on I created a helper class `AnchoredTile...
Do you have a sense of how that might impact runtime performance, memory allocations, and flash size? I've been experimenting recently with using TileGrid for relatively high frame rate animations, and I've been impressed with how well that works. It's nice to have the implementation be pretty fast.
lone axle
Do you have a sense of how that might impact runtime performance, memory allocat...
If it were a library it would be no impact unless you choose to use it instead of the standard TileGrid. If you did choose to use it I'm not sure of the impact it would have, I wouldn't expect it to be drastic but I haven't done any measurements.
If it were implemented in the core directly on TileGrid I would expect it to increase flash size, and memory usage slightly due to the code to support the new properties, and the values those properties hold. But the code needed won't be that much I think, the python implementation is here: https://github.com/FoamyGuy/Adafruit_Learning_System_Guides/blob/tft_spirit_board/TFT_Spirit_Board/shared/lib/anchored_tilegrid.py In the core it would probably expand a bit further due to handling of the tuples, but still shouldn't be that big.
I would not expect it to impact the rendering framerate of cycling through tiles in the TileGrid if the TileGrid is not moving also. If the TileGrid were moving, and you choose to move it with the new properties I would expect it to be ever so slightly slower than moving it with the x and y properties directly because it has to do a bit of math to calculate the position and then set the x and y properties.
tiny peak
If it were a library it would be no impact unless you choose to use it instead o...
interesting... for the Pumpkins vs. Skeletons game I just finished, I animated the x, y coordinate of the flying pumpkins. That let me use an 8x8 tile for the sprite, but, because I didn't have to stay tied to the 8x8 grid, the trajectory could look very smooth.
The frame rate I managed to get on an ESP32-S3 was mostly plenty fast enough, but with the occasional glitch once in a while. I don't really have a sense of how much of that was due to garbage collection pauses, time spent on data transfers over the SPI bus, or whatever else. But, my impression was that I was pushing the system close to the limit of what it could animate smoothly.
lone axle
The frame rate I managed to get on an ESP32-S3 was mostly plenty fast enough, bu...
Is your code published anywhere? I could try tweaking it to use the python version of AnchoredTileGrid and see if it makes a noticable difference. I don't have a core implementation to test, but am willing to make it if there is interest.
tiny peak
Is your code published anywhere? I could try tweaking it to use the python versi...
https://github.com/samblenny/pumpkin-toss-game
also, guide at https://adafruit-playground.com/u/SamBlenny/pages/pumpkins-vs-skeletons-game-for-circuitpython describes the hardware setup which includes a USB gamepad
I'd be very curious to hear how it goes if you try it.
lone axle
I'd be very curious to hear how it goes if you try it.
I don't think I have a USB gamepad to try it with. I'll poke around and see if one comes up though. It looks like the way you've got the code factored out should make it easy enough to mock the controller inputs if I don't find one. Although it perhaps lessens the meaningfulness of the test outcome for your use-case since I'd be removing the USB Host operations from the equation.
tiny peak
you might be able to make do with a USB numpad or keyboard. like, just read 8 bytes or whatever from endpoint 0x81 and look for anything that's not 0
anyhow... hope the spirit board thing goes well. I'm headed afk.
lone axle
@tiny peak is the YT video of your game here: https://www.youtube.com/watch?v=QStcSDSYKOE a realtime representation of how it looks when you play it? Or is that slowed down any?
I was able to get a modified version of the game working using a joy featherwing (connected via I2C) for input instead of a USB gamepad device.
I have the host featherwing connected and initialized and I have a device plugged in, but I think that device is presenting as a keyboard, not a gamepad and I don't think it would be sending much data.
In mine the pumpkin flying seems to render a fair bit faster. I'm guessing due to not really having to share the SPI bus between the gamepad and the display, but I'm not certain.
@tiny peak I've done a bit more testing now with pythom implementation of AnchoredTileGrid swapped in for the pumpkin tilegrid in your project.
I do not see any appreciable different in the rendering framerate when using AnchoredTileGrid vs. normal TileGrid. I did not measure scientifically though, just by eye.
I also tried changing the code that moves the pumpkin to use anchored_position instead of x and y To my eye, I still do not really perceive a difference in framerate rendering it looks visually the same as the normal TileGrid version to me.
All of the ones I've gotten running do seem faster to me than the one depicted in your YT video. If you're interested in seeing the difference let me know and I can get you some timestamped links to my stream this morning to where I ran the different versions.
tiny peak
<@1220648290941669377> is the YT video of your game here: https://www.youtube.co...
the video is playing at normal realtime speed
tiny peak
<@1220648290941669377> I've done a bit more testing now with pythom implementati...
interesting... I'm watching through the start of your stream now
lone axle
I struggled for a bit on a USB solution that didn't pan out. The back half contains all of the bits relavent to displayio and tilegrid.
tiny peak
yeah, the USB protocol for Xbox 360 gamepads is a bit of a head-scratcher until you get used to it
lone axle
Running it on a USB Battery instead of plugged into the Hub / PC actually seem sto make the pumpkin fly a bit slower strangely. It looks closer to the speed of your video this way, but still maybe a tad bit faster.
tiny peak
I think USB IO has a big impact on overall system timing
lone axle
Ahh, actually I was wrong. Changed too many variables at once.
I had unplugged the gamepad that I tried to use. (I left it plugged in to so that the game would detect it and let me start). It turns out if I unplug that after the game lets me start the pumpkins fly a bit slower than if I have it plugged in.
I guess unplugged it's spending a tad more time "looking" for that device or trying to read from it than doing the same when it remains plugged in (even though I'm not actually using it for the game)
battery doesn't make any difference, only whether or not I leave the gamepad plugged in.
onyx hinge
<@&356864093652516868> A gentle reminder that there is no meeting today, as Adafruit observes Indigenous Peoples' Day. The meeting is 24 hours later than usual, at 2PM ET / 11AM PT. See you tomorrow!
lone axle
I created https://github.com/FoamyGuy/Adafruit_CircuitPython_Anchored_TileGrid to hold the class utilized by my spirit board project. If anyone has a moment can you create a repo by the same name under adafruit org. I can move the code over to there and then get it added to bundle.
tulip sleet
I created https://github.com/FoamyGuy/Adafruit_CircuitPython_Anchored_TileGrid t...
I think you can request a transfer of the repo.
lone axle
I think you can request a transfer of the repo.
It disables the choice for me. I also tried "specify by organization name" instead of the dropdown with "adafruit" but it returned a similarly worded error that I am not allowed to request it.
manic glacierBOT
manic glacierBOT
nocturne sequoia
Hello everyone
I'm just starting out in circuitpython
I am currently making this project : here is the link:
https://learn.adafruit.com/matrix-portal-scoreboard/code-the-scoreboard?embeds=allow
I installed everything, I'm at: Adafruit IO Setup
But I don't understand where to install or copy the secrets.py file
Thank you for the help sorry for the translation I am French
Bonjour a tous
Je debute juste dans circuitpython
je suis en train de faire ce proget : voici le lien:
https://learn.adafruit.com/matrix-portal-scoreboard/code-the-scoreboard?embeds=allow
J'ai tout installer j'en suis a: Adafruit IO Setup
Mais je ne comprend pas ou il faut installer ou copier le fichier secrets.py
Merci pour l'aide desoler pour la traduction je suis francais
random junco
Hello everyone
I'm just starting out in circuitpython
I am currently making this...
Hi! Welcome to CircuitPython - can you please ask for help in the #help-with-circuitpython channel? That would be the correct channel as this channel is for the development of CircuitPython itself. You'll probably get help faster there. 🙂
nocturne sequoia
sorry thank you I will do it
desoler merci je vais le faire
blissful pollen
@onyx hinge to merge main into my PR is it just fetch adafruit/main and then merge adafruit/main? I have a less then stellar track record with not breaking my PRs with merges 🙂 If you don't have it off the top of your head don't worry I'll look into it more today.
onyx hinge
<@426200547024961539> to merge main into my PR is it just fetch adafruit/main an...
If you're not comfortable with merging then probably just don't worry about it
lone axle
It seems that RTD is unhappy with ubuntu-latest as the value for build.os in our readthedocs config file now.
Error
Config validation error in build.os. Expected one of (ubuntu-20.04, ubuntu-22.04, ubuntu-24.04, ubuntu-lts-latest), got type str (ubuntu-latest). Double check the type of the value. A string may be required (e.g. "3.10" instead of 3.10)
I assume it had worked in the past but it looks like they've changed to ubuntu-lts-latest I'm going to try it out on one and if it builds with no issues I could work on a patch for it this week. Though I think some of the libraries are still set to older hardcoded versions as well so it may take a few passes to get them all updated and back into alignment.
onyx hinge
@lone axle do you think this is a recent change on their part?
lone axle
I would guess that it's somewhat recent. our cookiecutter and newest libraries that I could find are using the value ubuntu-latest
Hmm a few months at least. I tried to search their changelog for related keywords. I didn't find the specific change that added "lts" but I did quickly find this PR from june where it looks like the lts was already in place at that time https://github.com/readthedocs/readthedocs.org/pull/11421/files
Welp another moment of poking into the git history and I'm fairly confused about how it was working in the past. It looks to me like the last change to the "latest" build version was 2/5/24 which appears to be when it was added as an option, however at that time it already seems to have the value ubuntu-lts-latest. So maybe ubuntu-latest never worked. I'm not sure how widespread that value is in our libraries. I've only checked a few but have seen mostly ubuntu-20.04.
candid sun
<@&356864093652516868> here is the notes doc for today's meeting: https://docs.google.com/document/d/1UY8EJC4OuO3upz7Wcdu8zkrxFtqC0utJWgaZ2GLybkA/edit
We'll have our weekly meeting in about 2 hours 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 -- I look forward to everyone's updates!
lone axle
I think it's most likely that the cookie cutter was updated erroneously to ubuntu-latest I made that change back in july, I assumed RTD version constants were the same as github actions one, but they differ in this case. Scanning the bundle it appears there are only a few libraries that have the incorrect value. The vast majority have ubuntu-20.04 so they could stand to be updated still.
slender iron
@fleet hollow if you have cycles to help someone on the forum https://forums.adafruit.com/viewtopic.php?p=1032165#p1032165
tulip sleet
maybe because all the libraries were updated
onyx hinge
- Total Library Stats
- 1304306 PyPI downloads over 334 libraries
no that's what it says in the report
lone axle
Ahh, yeah that could be. The actions install some of them I think.
tulip sleet
there must be many mirrors
lone axle
Thank you 👻
onyx hinge
thanks liz!
random junco
Thanks Liz! Have a great week everyone
lone axle
Thanks for hosting Liz, have a great week everyone!
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/1sFQK8Cn37B9dAOMinQ33eyYkPvlUz7M3em01pG8sxYI/edit
slender iron
@tulip sleet have time later to figure out my next step for circuitmatter?
(after I get lunch)
tulip sleet
sure
slender iron
kk, will ping you when I'm ready
slender iron
@tulip sleet I'm back now and free whenever
slender iron
@tulip sleet https://github.com/adafruit/circuitmatter/issues
tulip sleet
<@329766224093249548> <https://github.com/adafruit/circuitmatter/issues>
Thanks! I read them in my emails.
manic glacierBOT
manic glacierBOT
lone axle
Driver code style question: For a device like this: https://www.adafruit.com/product/6062 with 3 channels that it can read from. I know generally we prefer properties for accessing data in the drivers. Would it be preferrable to have 3 properties (one for each channel), or to have 1 function which accepts the channel as an argument?
slender iron
Driver code style question: For a device like this: https://www.adafruit.com/pro...
Generally we do one object per channel
each object looks like one sensor
lone axle
Generally we do one object per channel
Okay I think I see, so the final code will end up something like:
reading = ina.channel1.bus_voltage
ina.channel1.shunt_resistence = 0.1
other_reading = ina.channel2.shunt_voltage
I'd do channel[0] probably
or you can do ina[0]
both of those would allow you to iterate over the channels
manic glacierBOT
manic glacierBOT
manic glacierBOT
tulip sleet
@slender iron I have a BLE MTU/NUS change I would like to bounce off you, when you have time.
slender iron
<@252717193496756235> I have a BLE MTU/NUS change I would like to bounce off you...
now works for me
slender iron
@tulip sleet this may be a good overview of matter https://www.youtube.com/live/zn9wyo6woU8?si=mbs2N63W7yy08l9j&t=2214
tulip sleet
tnx!
tiny peak
has anybody done any previous work on building gamma correction into displayio? I'm particularly interested in ST7789, but also generally interested in anything that goes from 24-bit rgb displayio.Palette colors to RGB565 using busdisplay.BusDisplay under the hood. I see lots of stuff in old guides about gamma correction for arduino with NeoPixels, but haven't found that much related to CircuitPython.
slender iron
has anybody done any previous work on building gamma correction into displayio? ...
None that I know of. Probably something colorconverter can do