Maybe.
#circuitpython-dev
idle owl
lone axle
whats up?
idle owl
Can you pull down the BNO08x library, and fuss with this threshold: https://github.com/adafruit/Adafruit_CircuitPython_BNO08x/blob/ea0f05f7d3068246acafb65d65aece5cd8cfd992/.pylintrc#L256 and tell me how high it needs to be for Pylint to pass? https://github.com/adafruit/Adafruit_CircuitPython_BNO08x/runs/2127295355?check_suite_focus=true
ornate breach
The sky slaps knee
idle owl
@lone axle If I remember correctly, you have to change https://github.com/adafruit/Adafruit_CircuitPython_BNO08x/blob/ea0f05f7d3068246acafb65d65aece5cd8cfd992/.pylintrc#L26 jobs=1 for it to acknowledge the threshold change. Which I think is a bug, but I think they're aware of it.
Try it with jobs=2 first though to make sure I'm not misremembering.
lone axle
Can you pull down the BNO08x library, and fuss with this threshold: https://gith...
Yep, I can do that.
idle owl
@slender iron I remembered a point worth considering - you can't have the jobs increased in Pylint, and have it acknowledge the duplicate code threshold change. We ran into this with the two libs that we opted to increase the threshold on. It ignores changes to that threshold if jobs>1. It's a bug. So if we increase that threshold, we will not have Pylint running as quickly.
Though it seems like this is out of my hands, so I suppose it's more of an FYI than it is something to discuss.
idle owl
Yep, I can do that.
Thanks much. Let me know what you come up with.
slender iron
<@!252717193496756235> I remembered a point worth considering - you can't have t...
how did you change the threshold? In pylintrc?
idle owl
Yes.
I don't know of another way to do it.
It's not a normal check.
That you can disable.
idle owl
Ok.
I will have Tim come up with an increased threshold that works on that specific library, and we'll call that the new threshold. And have Dylan run an Adabot patch. CI doesn't trip the needs-release check anymore, so we won't have to do a sweep afterwards.
slender iron
👍
idle owl
@lone axle Slightly bigger favor. Can you check the other two libraries here as well and see what they would require? https://github.com/issues?q=is%3Aopen+is%3Aissue+archived%3Afalse+user%3Aadafruit+"duplicate+code+in+library+files"+
Threshold-wise.
lone axle
<@!382939733107408897> Slightly bigger favor. Can you check the other two librar...
I can, it will take me a little while to be able to make it to all of them though.
idle owl
I can, it will take me a little while to be able to make it to all of them thoug...
No worries.
lone axle
You are correct about the jobs=2 issue. It seems to ignore the min threshold when jobs is 2.
lone axle
@idle owl the magic number for bno08x is min-similarity-lines=12 with that set to 12 it starts passing pre-commit when I run it locally.
idle owl
Cheers.
slender iron
@onyx hinge are you looking at cascadetoml for the imx?
idle owl
<@!330227457296957440> the magic number for bno08x is `min-similarity-lines=12` ...
Let me know what you find with the other two when you get to them.
onyx hinge
@slender iron it's because of imx that I was poking at it
I'm not sure if it's going to be relevant though
slender iron
@onyx hinge we can work to get it going on 3.7
here is my WIP for rp2040 flash: https://github.com/tannewt/circuitpython/commit/431b52c5c73469c714bf29d531205bdc6387b004
onyx hinge
@slender iron I suspect the immediate term problem (supporting the GD chip) can be solved just by hand-writing the required config by hand, or by substituting in a "generic enough" one
"generic enough" is what we did on rp2040 feather right?
slender iron
yup yup
ya, for now
I see two options to fix the config for gd: 1) uncomment the single fast read version or 2) switch sequence 4 to using 0x51 (iirc)
which is writing the second status byte
onyx hinge
Way back when did you get very far in to understanding the imx flash config stuff?
slender iron
ya, I grokked it
onyx hinge
I see two options to fix the config for gd: 1) uncomment the single fast read ve...
That sounds like what I understood
slender iron
k, breakfast time for me and 🐱
simple pulsar
Hi. Is a non-beta release planned soon for Pi Pico?
tulip sleet
the next non-beta release will be for everything. The number of 6.2.0 issues is decreasing but is not zero yet
onyx hinge
there's also the "port status" .. will raspberrypi port remain designated "beta" when 6.2.0 goes out?
(I don't know that we've decided to mark it as "stable")
simple pulsar
https://circuitpython.readthedocs.io/en/latest/docs/index.html?highlight=beta#po...
@tulip sleet Thanks, I'm just writing at article on soil sensors on Instructables using Pi Pico and was wondering what version to recommend installing. I only need analogue inputs, digital outputs and i2c to work (and displayio).
tulip sleet
all those work, except busio.I2C is flaky on some sensors (not a CPy problem), and bitbangio.I2C would work
but really, it's in quite good shape, especially now the accidental overclocking of the flash chip has been fixed
simple pulsar
This is the 0 byte write thing? I get some occasional quirks on a nobrand SSD1306 screen but that also completely fails on me at times and then recovers so I've assumed so far it's the screen. (It has onboard 10k pullups, running it at 400kHz over 8 in. wires).
tulip sleet
that was fixed by beta.3. There are still some sensors (notably TCS34725, a color sensor), that don't seem to like something about the hardware I2C timing. The waveforms look fine on the Saleae
it's also possible to wedge some sensors into a bad state by doing probing, but nearly all seem to be fine. Also the PA1010D I2C GPS module likes bitbangio.I2C a lot better
onyx hinge
beta.3 had a serious problem for feather rp2040 but I don't think that one affected the pico
tulip sleet
i forgot the clocking issue was only on feather?
onyx hinge
(didn't we have the flash settings wrong? It's been long enough I don't quite remember)
tulip sleet
beta.4 will be out relatively soon, but we have no target day yet
simple pulsar
Unrelated to my plants, has there been any thoughts on the broken dma/wav playback? https://github.com/adafruit/circuitpython/issues/4208
tulip sleet
i haven't heard any further conjecture on that issue
lone axle
@idle owl here's what I found on the other two libraries to get them passing the dupe check from pylint locally:
LSM6DS:
min-similarity-lines=6
PyBadger:
min-similarity-lines=12
Convenient. Thank you!
We'll go with 12 apparently.
slender iron
@onyx hinge I released cascadetoml 0.3.0 on pip. hopefully it'll work on 3.7
idle owl
@onyx hinge Can you get me the still image of your heart circuit sculpture? Doesn't need to be high res, it's going into the newsletter. I can't really pull guide images from Learn, only images in pages, and I don't see that still anywhere else.
onyx hinge
@idle owl hold on...
idle owl
<@!330227457296957440> hold on...
No rush
onyx hinge
does https://cdn-learn.adafruit.com/guides/images/000/003/279/original/vlcsnap-2021-03-12-14h35m00s727.png?1615581396 work for you?
idle owl
Ah yes! Thank you. That works.
onyx hinge
@slender iron yay it's working here with py37. ```../cascadetoml/cascadetoml.py cascade filter sku="GD25Q64C"
{'sku': ['GD25Q64C']}
[[nvm]]
Data for path: flash/gigadevice/GD25Q64C.toml...```
onyx hinge
one thing I don't understand is, why do devices ship without quad enable .. enabled? what potentially breaks?
slender iron
the top two data pins are HOLD and write protect by default
onyx hinge
oh so you lose a function of those pins
slender iron
yup
onyx hinge
hadn't occurred to me
OK looks like some of the PWM setup is being done, but maybe the pinmux is not being set right...? (imx)
idle owl
@tulip sleet There was some library you and Limor were talking about and I'm supposed to add it to the contributor bundle. I never knew what library it was though.
idle owl
Ah thanks
tulip sleet
sure, its components map closely to their chosen blocks
idle owl
Added the lib to the bundle.
manic glacierBOT
slender iron
@tulip sleet my flash init changes are here: https://github.com/tannewt/circuitpython/tree/rp2040_flash
stage2.c.jinja gets templated and then compiled
onyx hinge
@slender iron I didn't dig into the compilation flags, but wondered whether making the functions in stage2.c static, or using -fwhole-program, or using -flto would get space back.
slender iron
ah, good point
onyx hinge
I assume you determined -fPIC was necessary, but .. why? don't you know the RAM address where you'll be loaded?
onyx hinge
yay!
slender iron
it gets loaded twice
onyx hinge
at different addresses? Ok.. I'll allow it
onyx hinge
🙎
slender iron
so close!
onyx hinge
yeah I'm getting about 270 here
slender iron
lto didn't help any more than static
so it does fit if I don't change the pad settings
onyx hinge
ssi_hw->dr0 = status_command;
ssi_hw->dr0 = status_command;
``` this is really supposed to be repeated?ya, the second could be zero
you need to send a second byte to trigger a read
may be able to use an eeprom mode to skip the extra writes and reads
onyx hinge
+ if ({{ clock_divider }} <= 2) {
ssi_hw->rx_sample_dly = 1;
+ }
``` based on the comment, can this assignment be made conditional? wait_and_read(1);
ssi_hw->dr0 = (uint8_t) CMD_WRITE_STATUS2;
ssi_hw->dr0 = {{ quad_enable_bit_mask }};
wait_and_read(2);
``` I don't suppose these can be merged, 3 `dr0` assigns + `wait_and_read(3)`?the wait causes chip select to de-assert
so you can't skip it
I think we might be able to always set the delay
@tidal kiln do you have the fan controller IC? There are a number of PRs for it: https://github.com/adafruit/Adafruit_CircuitPython_EMC2101/pulls
tidal kiln
nope. but on the way. ordered today.
slender iron
kk, cool
tidal kiln
with intent to take a look at those PRs
slender iron
thanks!
tidal kiln
np
tulip sleet
@slender iron I was trying to get a build going from your repo but failing. If you can send me an elf that includes stage2, I can look at the assembler. There doesn't look like a lot to shrink. Maybe the wait_and_read() routine could be squeezed in some way, since it only gets called with 1 or 2
rm -rf build-raspberry_pi_pico
Use make V=1, make V=2 or set BUILD_VERBOSE similarly in your environment to increase build verbosity.
{'sku': ['stage2.c.jinja']}
Traceback (most recent call last):
File "gen_stage2.py", line 11, in <module>
quad_enable_status_byte = flashes["nvm"][0]["quad_enable_status_byte"]
IndexError: list index out of range
@slender iron it's good to know why, thanks -- I don't see any obvious space savings, I tried tinkering with some stuff but it didn't make a difference
tulip sleet
is this in the regular elf?
slender iron
@tulip sleet did you init submodules?
onyx hinge
@tulip sleet there is a "4" watchdog_hw->scratch[1] = wait_and_read(4);
tulip sleet
yes, I did
slender iron
oh! do the feather, not the pico
tulip sleet
that shouldn't affect the assembly code
slender iron
the compiler can optimize it better then
tulip sleet
you could maybe just put the wait() part in a routine, and then exapnd 1, 2, or 4 reads into straightline code. that may or may not be an improvement
slender iron
it's probably more worth my time to just figure out how to use it from CP code
could even just link in a second copy
it all fits without the capacity read
slender iron
the benefit from doing in CP would be that we could do more complicated init if a board has very different flash chips
tulip sleet
but you were sad about losing some pico-sdk functionality?
you would have to rewrite?
slender iron
no, it'd be ok because the stage2 does full init
I just have to redo their internal functionality to stop things and then restart them
tulip sleet
it sounded like you could propose a PR to their current API to read more stuff
read more of the status register or whatever
slender iron
I'd really like a general command in their sdk
slender iron
not sure why they didn't just add that
tulip sleet
i think you could propose it and put it in our fork and it would eventually show up?
and then we could stop using our fork
tulip sleet
i don't remember; they are updating so fast that we should keep merging from upstream
if we are using our fork
slender iron
ya, they are releasing pretty slowly though
and develop hasn't moved that much either
anyway, I'll sort something out tomorrow
tulip sleet
i looked at the assembler (i fudged the size to 512 bytes in the .ld). this is all striaght line code, I don't see much to squeeze. I thought there were more conditionals and calls
tulip sleet
maybe wait_and_read could be shrunk by a few instructions, but not much, not enough to make a difference. it was 280 when I built it
slender iron
did you add the statics?
tulip sleet
no
slender iron
it dropped it by 10 I think when I did
tulip sleet
yaah 280 to 270
slender iron
only 18 off 😛
tulip sleet
i like your overall plan better, though
slender iron
ya, I'll just change the sdk
so I can use it to get capacity
time to get off the comp now though
good night!
tulip sleet
👋
earnest grove
I'm looking at building a macro keyboard (like so many others) and I was thinking of qwiik connectors. Adafruit has buttons with qwiik connectors, but they are not i2c. sparkfun has i2c qwiik buttons, but only mentions a Arduino library.
Anyone know if there is a Circuitpython library to use these in a project?
jaunty juniper
I'm looking at building a macro keyboard (like so many others) and I was thinkin...
there are drivers for some Sparkfun qwiic products in the community bundle
earnest grove
Thanks, I'll go look.
manic glacierBOT
unkempt crow
While chasing another problem I discovered a strange Magtag behavior.
Using an i2c device (in this case the tlv493d) and mqtt I was having to power off and on each time I was making a change, or I would get
ValueError: No I2C device at address: 5e
When I changed
i2c = busio.I2C(board.SCL, board.SDA)
to
i2c = busio.I2C(board.SCL, board.SDA, frequency = 400000)
the problem goes away.
tulip sleet
@unkempt crow that is interesting. The datasheet talks about 400 kHz as typical but doesn't say it's minimum. I will bring this to the attention of the right folks. If you have a non-ESP32S2 board to test the sensor on, that would be great. I will order one of these to test more thoroughly.
I am not sure this is ESP32-S2-specific.
@unkempt crow is this you? https://github.com/adafruit/Adafruit_CircuitPython_TLV493D/issues/9
tulip sleet
@unkempt crow could you try taking readings without using wifi to see if the lockup occurs then?
unkempt crow
<@642096241064542260> could you try taking readings without using wifi to see if...
`Using:
import board
import busio
import adafruit_tlv493d
print("starting up")
Hardware I2C setup:
i2c = busio.I2C(board.SCL, board.SDA, frequency = 400000)
#i2c = busio.I2C(board.SCL, board.SDA)
tlv = adafruit_tlv493d.TLV493D(i2c)`
I can turn the error on and off on the Magtag but not on a Clue
gloomy shuttle
Has there been given anymore thought to fixing the dir() issue that appears in the CPX library. It seems like there is a fundamental problem with the core and dir() calling properties, but could a temporary fix be implemented in the CPX library. https://github.com/adafruit/circuitpython/issues/4171
tulip sleet
Has there been given anymore thought to fixing the `dir()` issue that appears in...
I worry about a hack fix that makes it appear that dir() works in general.
@unkempt crow could you open an issue on the library with the info? Thanks a lot, and we will investigate further
unkempt crow
<@642096241064542260> could you open an issue on the library with the info? Than...
The busio library?
unkempt crow
are you lubarb on GitHub?
Will do. And yes, I'm probably lubarb on GitHub
gloomy shuttle
I worry about a hack fix that makes it appear that `dir()` works in general.
I understand. I am teaching a workshop using the CPX and have a lesson coming up on CircuitPython. The refactor of the CPX library has completely broken the ability to use dir(), so it will be a lot harder to walk them through using the REPL to see whats available.
tulip sleet
if you just do cp.<tab>, does that break also?
gloomy shuttle
if you just do `cp.<tab>`, does that break also?
Perfect, thank you! That is an excellent workaround.
tidal kiln
@unkempt crow responded on github. let's move to #help-with-circuitpython and can you post a photo of your setup there?
manic glacierBOT
I've been prodding this library as an alternative to some other code I have, and it's super-impressive. One question: what's the most appropriate way to change the auto-generated device name?
In my MIDI Monitor tool, I get this:
and I've had good success with other boards on changing the detected devicename - which I appreciate might be a USB setting rather than a MIDI-specific...
slender iron
@still zephyr I invited you to CircuitPython librarians so you can do PR reviews on libraries
still zephyr
Thanks @slender iron
slender iron
np, it just lets you help me 🙂
still zephyr
🙂 ok no problem always like to help
slender iron
thanks! it's appreciated
slender iron
@tulip sleet are you hoping to do an rc next instead of a beta?
tulip sleet
@slender iron - maaaybee an RC next, but we would want to do a feature freeze and do a 6.2.x branch at that point, since we have so many new things coming in. Looking at the 6.2.0 issues https://github.com/adafruit/circuitpython/milestones/6.2.0 I think 4431 is worth trying to fix. As for 4042, other boards, we don't have PR's for a few not-yet-on-sale ones, so we could skip those. 4041 is up to you, and I am looking at 4010 right now to decide if it's worth trying to fix
tulip sleet
we could start with 7.0.0. betas right away, as soon as we do the incompatible changes: new ulab API, remove the transitional compatibility stuff
slender iron
I think I'd do one more beta since beta.3 is a bit rough
tulip sleet
i feel that's good too
slender iron
I will want to do further 6.2 stable releases for rp2040 boards too
tulip sleet
i thought maybe you were pushing for an rc 🙂
slender iron
no, seems like you are 🙂
tulip sleet
no 🙂 I just want to get usb_cdc turned off, etc., and the Feather RP2040 fixes in
I could make a beta any time: the release notes are up to date. I could do it this afternoon
slender iron
ya, that'd be my vote
tulip sleet
i think it's a good time -- there are plenty of fixes piled up
ok, I will start on that, I just need to do the download counts and contributors
slender iron
jeff and your minds didn't explode with my flash setup for rp2040 so I can probably get that in next week
tulip sleet
righto -- i think contributing to pico-sdk is the right tack
slender iron
👍 that's my task today
it'll be nice to read capacity from the flash
I just ordered ~30 different flash chips from digikey....
tulip sleet
ok, I will start on the beta right now. Mostly need to wait for zillions of download logs files (I have all of Feb but no March)
slender iron
I'm pretty happy with the nvm.toml stuff
thanks! we should publish daily jsons of counts on public s3
that way we know where to pick up with the next release
tulip sleet
i know how to do it with some AWS features or other, I just haven't gotten started on that, but it sounds like something I might have time for. I was hoping I could have the server logs be confined to HTTP GET's but that doesn't seem to be a setting choice
slender iron
👍
tulip sleet
so I can troll through them with a server-side script; I think I wwas going to use Lambda or something like that
ok, regular beta, and then maybe some infrastructure work for the next release
slender iron
👍 thanks!
manic glacierBOT
tulip sleet
@onyx hinge I am about to make a release. Is #4429 good to merge?
onyx hinge
@tulip sleet if it's not too late, I think it is but I didn't fully look into the dirty area issue that kmatch raised. there may be something about what he's saying that I don't understand
tulip sleet
yeah, that's why I didn't merge it, so i could wait until beta.5
i know you did try it, but i didn't follow the discussion
crimson ferry
CI pylint is busting me for things it didn't bust me for a few days ago in library examples, like importing non-existent secrets, and import order of said non-existent file
and missing docstring
tulip sleet
@onyx hinge I think I will wait on the merge of #4429 since there is some mild question
crimson ferry
ok, weird, the pylint errors are referring to last week's file, not the present one, but last week's file succeeded last week
onyx hinge
@tulip sleet that is fine, either way the effect is mild
tidal kiln
@gilded cradle hey. do you think this is something blinka should be taking care of? relates to this:
https://github.com/adafruit/Adafruit_CircuitPython_DHT/issues/63
but here's how blinka ties in:
https://forums.adafruit.com/viewtopic.php?f=60&t=175209#p859042
i guess the other option is to make sure the user code exits nicely?
gilded cradle
@tidal kiln possibly, but I'm not sure how simple getting it to work in Blinka would be since maybe libgpiod_pulsein is holding onto it.
tidal kiln
yep. that's exactly what's happening. just not sure who/where/how has the responsibility of killing the thread.
gilded cradle
It may be an issue in the libgpiod_pulsein script, which is written in c. Some deeper exploration would be needed as I'm mostly guessing at this point. 🙂
gilded cradle
Ah, ok. So it may be totally doable with a try/except KeyError block.
tidal kiln
yah. something like that try/except....just some mechanism that would make sure that blinka's deinit stuff gets called....
gilded cradle
right
tidal kiln
there may be some pythony solution to this i'm not aware of
like registering a callback with SIGINT?
gilded cradle
🤷♀️ not sure. I've only minimally played around with subprocess.
tidal kiln
yah. same here. just some basic launching etc.
you OK if i open an issue just so we have something for reference?
gilded cradle
Sure
tulip sleet
@onyx hinge or @slender iron could you approve https://github.com/adafruit/circuitpython-org/pull/652? For reasons having to do with previous accidental beta.4, I submitted this from my own repo instead of adafruit blinka. But everything else is OK.
I cannot merge it myself because I am the PR author
onyx hinge
@tulip sleet hmmmm I think there's a problem...:
the languages list seems to be everything, but we withdrew several translation since the last release
build_release_files has language_allow_list = [ "ID", ...
tulip sleet
it looked like it was a different order
onyx hinge
but maybe it didn't get used in all the right places .. en_GB is on the list in your new files.json but it shouldn't be
tulip sleet
i will clean out the bad languages by hand; i think we didn't have a release since that change?
onyx hinge
the allow list needs to be moved so that generate_download_info() uses it too. I'll enter an issue.
tulip sleet
@onyx hinge where is the allow list?
never mind I was looking in the wrong repo
Removing "hi",
"en_GB",
"cs",
"ko",
"el"
onyx hinge
I think that's right?
tulip sleet
i checked it systematically, so yes
onyx hinge
Note: Not building languages {'en_GB', 'ko', 'hi', 'cs', 'el'}
I wonder if we can/could json.dumps(..., sort_keys=True)
onyx hinge
thank you!
tulip sleet
well it was a happy accident I had to do the PR by hand and you had a keen eye
thank you
signal pasture
https://github.com/adafruit/circuitpython/pull/4384 😦 what are the alternatives, besides the repl?
tulip sleet
you just need to re-enable CIRCUITPY_USB_CDC and you'll have the separate CDC channel again
signal pasture
do i have to do that in c/c++? or in python?
tulip sleet
you have to do a custom build of CircuitPython
signal pasture
i would probably screw it somehow
tulip sleet
it's a very useful thing to know how to build. Have you done Linux command line before? Do you have a Linux box? I would recommend that over Mac or Windows.
signal pasture
would it be possible to, somehow, have something to be able to control enabling that 2nd usb_cdc? or it's part of the bootloader?
i can spin a vm, and have some devices with linux
tulip sleet
it will be controllable in boot.py. I have to do the work. It's my next major thing to do.
because we want to make it easy to use again
you could stick with beta.3 for now if you don't need to move forward
signal pasture
oh, if it will be able to be enabled in boot.py, then yeah, i will wait for it
i dont see anything i need in there on my rp2040, so, i will hold on to beta3
thank you!
tulip sleet
if you are desperate I can spin you one, but it's not that hard. https://learn.adafruit.com/building-circuitpython/build-circuitpython just follow the directions
signal pasture
nah, im not desperate at all
tulip sleet
like, for instance, you may want to turn off CIRCUITPY, etc., in the future once you get a device you want to leave plugged in permanently
signal pasture
i actually want to make a device to be permanently plugged in, but i kinda want to have flash access too, so c# can open/read/save a file into there
but i do have some doubts about boot.py, so, i will ask in #help-with-circuitpython
raven canopy
@tidal kiln @gilded cradle maybe look into using atexit, to force the deinit when its not used in a context (with ...)? https://docs.python.org/3/library/atexit.html#module-atexit
gilded cradle
Oh yeah, I think you're right. I think we used that for neopixels.
tidal kiln
@raven canopy neat. thanks! will check it out.
crimson ferry
@ionic elk do you have an esp32-s2 CircuitPython UDP socket server test .py? I haven't been able to get it to work with clients yet (https://gist.github.com/anecdata/3d5be8bd12db74f9301af6b348272278)
solved: it needs to bind the actual IP address, not just 127.0.0.1 or localhost
slender iron
@tulip sleet thanks for doing the release!
tulip sleet
yw - it was a lot less work this time except for the language thing
the ongoing release notes is maybe more time altogether, but a lot less painful
orchid basinBOT
orchid basinBOT
lone sandalBOT
ionic elk
@crimson ferry UDP servers? I actually don't know what's involved with that. I just wrote sendto and recvfrom_into. Shouldn't any server structure regarding datagrams happen on top of those?
There's no connectivity state with datagrams so the concept of a UDP server isn't built into the API in any obvious way I know of
crimson ferry
I got the server part working (I think), it needed to bind to the explicit IP address rather than "127.0.0.1" or "localhost", though I have yet to be able to read the replies from the server (can read replies from CPython servers)
ionic elk
It's the "bind" part I don't get
crimson ferry
I guess bind was pre-exisiting for TCP
ionic elk
Yes, I guess you have to bind but I'm not sure I proofed that for UDP
I can try poking at some test sketches for that
With "bind" you need to actually grab the real IP address from the ESP32-S2 via the debug console or wifi module, have you done that?
signal pasture
can i ask something weird about what you're talking about?
crimson ferry
ESP32-S2 UDP Client works fine, as you know. Server, when bound to explicit IP, will receive and show data sent back. I haven't yet gotten CPython, ESP32-S2, or ESP32SPI UDP clients to be able to read that returned data. May have to get serious and watch the packets on the network.
signal pasture
(i know it's kinda rude to stick myself in the middle of a chat, but, i may be able to help)
crimson ferry
(I wasn't expecting the explicit IP required, since CPython is pretty casual about it)
@signal pasture totally fine, go ahead
ionic elk
@crimson ferry so an ESP32-S2 socket will receive UDP data fine with recvfrom_into, when bound to an address? But when it sends a reply back with sendto, it fails?
crimson ferry
@ionic elk from the server perspective, it shows data sent back to the client. But none of my clients can rad that (yet?)
ionic elk
I mean, presumably both ends require an address bound to them, if you're using ESP32-S2s on both ends
@crimson ferry can you be more specific about what you mean by "shows data sent back to the client"
What actual calls are you using
@signal pasture we're talking about the ESP32-S2 implementation of socketpool.Socket. We don't have TCPServer on Circuitpython
or at least not on the module level
maybe up in the libs or something I dunno, I'm the C monkey
crimson ferry
UDP server: bind, recvfrom_into, and sendto. Client: sendto, recvfrom_into. (sendto includes the IP/Port tuple)
"shows data sent back to the client": size = s.sendto(buf[:size], addr) print("Sent", buf[:size], size, "bytes to", addr)
ionic elk
So on your UDP server, you have valid data that seems to be going out with sendto, and it isn't working
so either the server sendto, or the client recvfrom, is failing on the reply. Gotcha.
signal pasture
but is anything listening on the other side?
ionic elk
I can check it out, thanks for clarifying
crimson ferry
yes, the client is waiting with a recvfrom_into
ionic elk
@signal pasture presumably the client listens with recvfrom_into (a blocking call) after it sends the original request.
signal pasture
yes, but is the server listening?
crimson ferry
TCP Client and Server are functional at a low level on ESP32-S2; UDP Client is all good; UDP Server seems to work though I haven't yet gotten replies back at the clienthttps://gist.github.com/anecdata/3d5be8bd12db74f9301af6b348272278
crimson ferry
@signal pasture In my scenario, both client and server are repeatedly sending and listening
ionic elk
@signal pasture we are talking about datagrams, so listen() is not used, if that's what you're thinking of
None of this uses TCP calls
ionic elk
@crimson ferry can you post an issue for possible problems with either sendo/recvfrom in a reply context from a UDP server, with whatever code you can post?
I'll track it from there.
signal pasture
by "listen" i meant "some device on the other side that is ready to receive anything"
crimson ferry
yeah, I want to play around more to make sure I'm not missing something dumb
ionic elk
@signal pasture yes, that would be the client, using the call recvfrom_into, which will block until it receives something.
crimson ferry
my goal is to get TCP and UDP sockets fully interoperable across CPython, ESP32-S2, and ESP2SPI (though there are no servers in the latter)
ionic elk
@crimson ferry sure. And so just to recap, I've never tried UDP from the server context you're describing, so it's possible I missed something about the interaction between bind and sendto, or something
crimson ferry
<@542699284810694660> yes, that would be the client, using the call `recvfrom_in...
or timeout
signal pasture
makes sense then
ionic elk
@crimson ferry it actually also might be a timeout problem, when I think about it. I might need to go in and make sure that sendto has appropriate timeout handling
crimson ferry
sendto is fine from the client side
ionic elk
UDP didn't get the huge timeout/blocking pass I did on TCP
crimson ferry
sendto appears to succeed from the server side
ionic elk
well, give it a bit more work and then tag me in an issue whenever you'd like me to look
thorny jay
Is there a better way to do "profiling" and find where time is spend?
start2 = time.monotonic() <put the code here>
stop2 = time.monotonic()
print("Time copy: %0.3f s" % (stop2-start2))
simple pulsar
Hi. I was just looking at some of the PIO code in CircuitPython and this line looks wrong. The tests for tx and rx on this line look superfluous and would do something wierd if transfer was called with two NULL data pointers (probably not possible with current code). https://github.com/adafruit/circuitpython/blob/main/ports/raspberrypi/common-hal/rp2pio/StateMachine.c#L620
orchid basinBOT
turbid radish
?serverinfo
digital shoreBOT
adafruit
manic glacierBOT
orchid basinBOT
ionic elk
@onyx hinge hey Jeff, is AudioPWMIO normally supposed to be interruptible by the REPL or no?
onyx hinge
If you're interactive, .play() will return to the REPL while the sound continues playing
if you exit the repl sound will stop
ionic elk
so a ctrl-c should exit
onyx hinge
.play() returns immediately in other words
so normally it's followed by a while True: or while speaker.playing: loop which will be interruptible with ctrl-c
is that answering your question?
cedar moth
Just a heads-up that I posted an issue under the Adafruit_CircuitPython_DisplayIO_SH1107 library on GitHub regarding the SH1107 128x64 OLED FeatherWing Display and the Feather RP2040 over I2C. Since the problem seems to occur only with the RP2040 and this display (also tested with M0 and STM32 boards and a different display with no issue), I wasn't sure where best to post but wanted to bring this to the attention of those of you working on the RP2040 in case it is relevant.
https://github.com/adafruit/Adafruit_CircuitPython_DisplayIO_SH1107/issues/4
blissful pollen
@slender iron (or really anyone else) looking for something else to work on in the core if you have something that's not high priority but needs to be done. I have a couple weeks vacation and well can't go anywhere so may as well do something!
slender iron
@blissful pollen there is a quick PIO ASM bug you could knock out
will think about stuff for the core
blissful pollen
cool is there an issue for it? I can go look for it
slender iron
yup, only issue on the PIO_ASM repo I think
orchid basinBOT
simple pulsar
Hi. Has anyone managed to get an implementation of PWM on PIO on RP2040 to work like the ones floating around for MicroPython?
manic glacierBOT
There seems to be an issue with the way that the progress_bar library is working. Running the progress_bar simpletest I am getting this mostly filled bar with lines in it:
The function responsible for updating the progress bars bitmap is here: https://github.com/adafru...
jaunty juniper
you gotta love accidentally sending a form before filling it in 🙄
vague latch
Hey all how does circuitpython work under the hood. What i am trying to do is add functionality to my code to provide the deque module from the collections library in python. The library is in c and so i was thinking it should'nt be hard to add it to the circuitpython base. Am i missing something here. Is there a guide or instructions on how to do this sort of thing
solar whale
@vague latch are you familiar with building CircuitPython https://learn.adafruit.com/building-circuitpython once you can build it, you can modify it....
vague latch
so is there anything stopping me adding a standard python library
solar whale
It is written in C, not python -- if you want to ad a python library, you can do that without rebuilding CP.
vague latch
yeah, so actual cpython library
would it be worthwhile adding in the whole collections library then.
is it a balancing act between functionality and size for the main build
solar whale
Right -- If you think it is something that needs to be added, you can open an issue for discussion or add it and then create a PR to see if it is something that should be added for all.
vague latch
ok. thanks for that. ill do some testing.
mental nexus
I usually start by writing a Python library to try things out. That’s usually much quicker to get something working and try out all the logic. (Unless it needs access to something that not available to Python then that’s a different story.) Then if it needs to be faster that’s really the reason to put it into the core C code. But if you want to learn more about how CircuitPython is built then by all means go for it! And ask around here, the guides are a starting point but there are many ways to get stuck. Ive learned a lot by asking here when I’m stuck.
vague latch
so i really want to get up and running an async type code base.
i watched david beazleys discussion on asyncronous and concurrent programming and his code utilises heapq and deque
i was looking to add these functionalities to my code and noticed neither of them are implemented.
or if they are i couldnt see them
mental nexus
Have you tried the tasko options? I’ve not used them. It I know some support of async is available.
vague latch
it wasnt running at the speed i was expecting so i thought i will try and write one myself.
mental nexus
Got it, I see a similar issue noted here: https://github.com/WarriorOfWire/tasko/issues/1. Maybe that’s your issue 🙂
vague latch
maybe. I will look into that. I cnat seem to build circuit python forthe feathers2 at the moment anyway
does the circuitpython build require a certain version of python
jaunty juniper
what issues are you getting ?
vague latch
I dont remember what the issue was i t basically never looped through
or the issue with the build
that one was fnmatch.py Nonetype is not iterable
lone axle
Feather S2 is an ESP32-S2 which does require some extra set up to build.
Scroll down to the "readme" section of this page: https://github.com/adafruit/circuitpython/tree/main/ports/esp32s2
if you have not done the install and export steps for the esp-idf you'll need to do those.
jaunty juniper
esp tools want python3 but call python
lone axle
That version should be okay. I am able to build and I use python 3.8 as well
jaunty juniper
on an ubuntu VM I installed python-is-python3 for that
vague latch
ok
lone axle
mine is set up that way (python3 is actually called python in my environment)
vague latch
ok my error is in ```circuitpython/ports/esp32s2/esp-idf/tools/ldgen/generation.py", line 84, in do_section_expansion
expansions = fnmatch.filter(sections_in_obj, section)
File "/home/tim/miniconda3/lib/python3.8/fnmatch.py", line 55, in filter
for name in names:
TypeError: 'NoneType' object is not iterable
ninja: build stopped: subcommand failed.
Makefile:310: recipe for target 'esp-idf-stamp' failed
make: *** [esp-idf-stamp] Error 1
did you run these commands in the repo also?
git submodule sync --quiet --recursive
git submodule update --init
yep
lone axle
Those and the install / export of the esp-idf are the main gotchas that I am aware of. If you've taken care of those I'm not sure what else could be giving you trouble.
I don't recognize that error specifically. But as other folks come online someone may know what it's about.
vague latch
hmm thanks for your help anyway
ill have a fiddle with it and look at the update of tasko
lone axle
I honestly don't know much about anaconda, but I noticed your path has miniconda in it. I have seen other folks have issues with anaconda environment before. I'm not certain if it's your issue. But if you have a way to try a "vanilla" python environment it might be worth a shot.
onyx hinge
hmmm I think there's a (probably preexisting) problem with dirty areas for TileGrids that transpose_xy when updating grid elements
I sorta suspect the coordinates "dirtied" are not accounting for transposition alone
mental nexus
I sorta suspect the coordinates "dirtied" are not accounting for transposition a...
I’ve never looked into how the refresh areas works with rotations. I suspect you’ve probably found a weakness where dirty_area isn’t transposed.
Looks like it is trying to handle it here https://github.com/adafruit/circuitpython/blob/e084a92671b01220c795d114bee9d5058dbfa680/shared-module/displayio/TileGrid.c#L586
crimson ferry
http://wifitest.adafruit.com 403 Forbidden 😦 Is it look for certain headers or blocking some kinds of IPs? --> n/m, just needs the full URL http://wifitest.adafruit.com/testwifi/index.html
onyx hinge
I'm not able to follow what that code is doing right now, maybe some other time
@crimson ferry hmm it looks like you're supposed to use a particular URL within that host according to the examples. Do you think it's "new" that the front page doesn't say something? ```$ curl http://wifitest.adafruit.com/testwifi/index.html
This is a test of Adafruit WiFi!
If you can read this, its working :)
No, just me remembering wrong
onyx hinge
https://web.archive.org/web/20190507081127/http://wifitest.adafruit.com/ In 2019 it already said 403 to the front page
tulip sleet
haha: https://wifitest.adafruit.com/wifitest/ (note reversed words) says This is a test of the CC3000 module! If you can read this, its working 🙂
onyx hinge
interesting
@crimson ferry ah I see you'd already updated yourself when I said that. glad you figured it out
crimson ferry
interesting, wasn't aware of the alternate
onyx hinge
Seems like it's not unreasonable for the "/" page to say something besides 403, so I sent an internal message to our web folks. Dunno whether they'll do anything, but the issue has been raised.
they often have bigger fish to fry
crimson ferry
I can understand not having anything at the root, keeps out the riff-raff. I do that with some of my domains.
crimson ferry
Is anyone else still seeing Adafruit CircuitPython 6.2.0-beta.4 on 2021-03-18; Adafruit Feather RP2040 with rp2040 disappearing after reset?
...and is there a better procedure to get it back than the Raspberry Pi org PICO flash nuke UF2?
manic glacierBOT
crimson ferry
...I have a very strange Requests bug that only manifests on the Feather RP2040 (some primitive behaves different? timing? other?) But I want to make sure the flash is stable to rule that out as an issue.
ornate breach
hey @tulip sleet can the bin generated for circuitpython be loaded into memory through DAP and be fine or is it better to just drop the uf2 via usb?
i'm developing my workflow for burning bootloaders and loading circuit python onto a lot of boards in a pretty fast go, and want to know if I can just DAP it into memory or if I need to individually drop circuitpython on them via USB
tulip sleet
hey <@!329766224093249548> can the bin generated for circuitpython be loaded int...
You could load the .bin with DAP; we do that kind of thing at the factory, though we often don't load CircuitPython on the boards, but instead some Arduino test program. There's nothing special about using the .UF2 bootloader.
ornate breach
oh gotcha, i'm wondering if it's worth while to do or just let the end user do it.
I don't really have plans to add arduino support for my boards because it's just cumbersome
one day maybe
tulip sleet
the tester loads a test program; the human checks the test output, and then that is it
ornate breach
i'm going to eventually need to make a more robust test jig
manic glacierBOT
still zephyr
@lone axle let me know if you are around, want to discuss about the readinto issue
lone axle
<@!382939733107408897> let me know if you are around, want to discuss about the ...
Yep, I am and sure! we can do that.
still zephyr
So I am trying to test your solution, what imageload version did you use?
lone axle
I think we have two different issues there 1) the unexpected argument with the current released version of CircuitPython. and 2) Certain images get loaded improperly and then look different than their raw image file.
still zephyr
1.4?
Yes at the beginning I was not sue, and I was not having the same error as you, so I guess is the imageload version, as I am using an old one
lone axle
Yeah, version 0.14.0 is where the issue was introduced I think. That was the one I was using when I noticed the unexpected argument error.
still zephyr
ah ok, gotya, I am testing this right now, I will put my review in the comments. I was not sure about the version of the imageload, I guess with old version works ok
lone axle
Yeah, that is true. The old versions did not try to use the new readinto() function so they do not have these issues.
still zephyr
@lone axle review is done, do I need to do anything else?
lone axle
That is perfect. Thank you! I'll give it a little longer today to see if anyone else has any thoughts or ideas. But if we don't get anything else I'll go ahead and merge this tonight so that we hopefully avoid too many folks ending up with the combination of versions that leads to the crash.
onyx hinge
drat, there's no UART pin left on the rp2040 feather once you use an rgbmatrix featherwing
(rx only)
I guess I can try implementing the PIO UART
ornate breach
😬
manic glacierBOT
onyx hinge
rx-only uart with pio: https://gist.github.com/c172ba263b6596ded7ead4e8cd4da805
manic glacierBOT
still zephyr
@mental nexus Are you planning to make a Cartesian plane widget, if not I would like to work on it
mental nexus
<@!366059101203202052> Are you planning to make a Cartesian plane widget, if not...
By Cartesian plane, that means like an x,y graph? I don’t have plans but it would be a great addition. Go for it! I saw some plotting code recently in these chats, maybe Arduino. If I run across it I’ll try to point you to the link if it might be helpful. But go for it!
still zephyr
By Cartesian plane, that means like an x,y graph? I don’t have plans but it wou...
Yes, I was looking at Microplot, even do a PR, just checking with you, do not want that we work in the same thing 🙂 Thanks
manic glacierBOT
fossil gorge
Crud. TIL that the PyGame library doesn't support Apple's M1 processor yet.
ornate breach
😮
nocturne monolith
YES! Just made the Feather M4 feather light up when I push a button. I'm proud of myself ^_^ This it my first CircuitPython experience!
ornate breach
nice job! 🙂
blinky type projects are greatly undervalued for the impact they have 🙂
nocturne monolith
Time to hook up a speaker for a tone!
fossil gorge
YES! Just made the Feather M4 feather light up when I push a button. I'm proud o...
Great stuff @nocturne monolith! Today the LED, tomorrow - the world! 🙂
nocturne monolith
MUAHAHAHAHA
fossil gorge
Now I'm concerned just what tone you're planning on emitting! lol
pulsar bloom
+1 on changing PulseIn frequency.
nocturne monolith
Actually, I hooked up the button to a keyboard. Small steps!
manic glacierBOT
When the MX25R1635F flash IC is connected to nRF52840 over quad-SPI bus - the CircuitPython does not boot.
Primary reason for that is incorrect 'quad_enable_bit_mask' definition used in current 'devices.h'
This is an excerpt from the datasheet:
For compatibility reasons with all the family of the MX25R1635F IC's and with all the modes of operation it is wort...
manic glacierBOT
@jepler Thank you for the review and merge actions!
800 us of 'start_up_time_us' is good enough. It is specified as a minimum value in the datasheet:
idle owl
@lone axle You up for pulling down a PR and seeing what the min-similarity-lines threshold would need to be to get it passing?
lone axle
<@!382939733107408897> You up for pulling down a PR and seeing what the `min-sim...
Yep I can do that.
idle owl
Yep I can do that.
@lone axle Then if you could PR the change to that PR, I'd really appreciate it.
If it will let you anyway.
Or push directly to the PR, whatever works.
If neither works, you can make a separate PR. Rerunning the Actions job on PRs seems to pick up changes to .pylintrc.
onyx hinge
squeee .. I PR'd into pico-sdk and they accepted it
manic glacierBOT
I added the line
CIRCUITPY_USB_MSC = 0
to the bottom of:
https://github.com/adafruit/circuitpython/blob/1be5ca7881a8aa8d32a783bc6a94994886463c30/ports/raspberrypi/boards/pimoroni_tiny2040/mpconfigboard.mkThis is what my mpconfigboard.mk board looks like
I get the same error when adding CIRCUITPY_USB_MSC = 0 in mpconfigboard.mk for Raspberry Pi Pi...
ionic elk
Does anyone have experience using SWO as a logging output in Jlink?
manic glacierBOT
modern wing
Good afternoon all you wonderful folks!
idle owl
<@&356864093652516868> CircuitPython Weekly meeting in a few minutes here. If you're participating with Hug Reports and Status Updates, please add yourself to the notes doc. https://docs.google.com/document/d/11uksHbpF2KT8sBC9Bov9XLtSgM62HXAHUKVxKhjr5cE/edit
We're finishing up an internal meeting - we'll be there soon!
@slender iron Reminder that you're on notes for the meeting today. I am still unsure of my power status. Seem stable? But 🤷♀️
slender iron
yup, on it
idle owl
Thanks!
idle owl
BRB
silver tapir
lurking
modern wing
chonk++;
slender iron
👍
slender iron
🥚
slender iron
committee? maybe
lone axle
Committee I think. A few of the weblate contributors were talking with each other in the chat during my stream over the past weekend.
cobalt grail
A big hug to the weblate GANG.
thorny jay
😊
still zephyr
Sorry for my "unreadable" comments ....
slender iron
(on desk of ladyada last night)
it's a flat flex cable pinout for easily connecting displays
slender iron
@still zephyr have you tried reading the I2C registers directly?
I wouldn't wait for the pico-sdk
solar whale
1 cord = 3.6 cubic meters
thorny jay
And the Blue LED will be what?
still zephyr
<@!607375730610798634> have you tried reading the I2C registers directly?
Not yet, I was not aware I will look/dig into that this week,
still zephyr
https://github.com/raspberrypi/pico-sdk/blob/master/src/rp2040/hardware_structs/...
oohhh wow 😮 🙂 Thanks
slender iron
🙂 it's usually my preference
since you can use the datasheet as documentation for it
blissful pollen
text only for now! sorry forgot to add it to the status area too
slender iron
got ya
gilded cradle
slender iron
@blissful pollen audomixer for rp2040 would be good
it uses assembly that the cortex m0 doesn't have so it'd need a generic C implementation as an option
blissful pollen
Sure i'll take a look at that
slender iron
solar whale
Sorry -- I have to leave early -- have a great week 👋
slender iron
@blissful pollen picking up and finishing pending PRs would be awesome too
still zephyr
Thanks @lone axle 🙂
blissful pollen
<@!228322543507865601> picking up and finishing pending PRs would be awesome too
You mean ones that are now pretty old/potentially abandoned? (though I have to poke a couple library ones I have)
slender iron
yup, the old core ones
I think the one to switch to different python files to run is pretty close
slender iron
@thorny jay are you text only still?
thorny jay
I'll try my mic.
slender iron
kk
still zephyr
@idle owl Yes it helps, thank you very much.
@timber mango Thank you for contributing and be in your first CP meeting 🙂
idle owl
<@!330227457296957440> Yes it helps, thank you very much.
If you have any further questions, feel free to ping me. But I think you're on the right track, and any issues can be handled in review.
lone axle
One of the ways we've attempted to alleviate this issue with pin names in some of the libraries to allow keyword arguments for uses to add to the initialize if they want to customize the pins that are used (which is needed in the case if pin names are different). A more general "board-level" solution would be really cool though.
still zephyr
If you have any further questions, feel free to ping me. But I think you're on t...
Thanks again I will
idle owl
You're entirely welcome!
timber mango
@still zephyr it was a pleasure to join the meeting.
@idle owl thanks for the clarification.
If you guys want to discuss further the PR just ping me.
idle owl
<@!607375730610798634> it was a pleasure to join the meeting.
<@!33022745729695...
You're welcome! Thank you for the contribution!
crimson ferry
https://learn.adafruit.com/adafruit-feather/feather-specification doesn't seem to mention pin naming. Perhaps it should. But FeatherS2 has used otherwise-standard Feather "D"-names on different pins.
FeatherS2, e.g., microcontroller.pin.GPIO9 board.D11 board.IO9 board.SCL
(not the same physical pin location as D11 on other Feathers)
blissful pollen
@slender iron quick question, since the M4 is pretty port specific worth splitting audio mixer into common-hal?
mental nexus
Thanks all!
slender iron
@blissful pollen I don't think so because it'll work on any cortex with those instructions
@onyx hinge has some #if statements in the mp3 lib that you can use
blissful pollen
okay that answers my next question if you knew where they'd be 🙂
slender iron
idle owl
<@!330227457296957440> https://github.com/adafruit/circuitpython-weekly-newslett...
Thank you!
slender iron
np
manic glacierBOT
thorny jay
Ok, maybe I start to understand the FeatherS2 issue: https://gist.github.com/dglaude/495f6faacc674222d9797fb87e08fcfe
So I believe those 5 Dx are wrong, they should match M4 things.
slender iron
yup
thorny jay
D5 = org_board.IO33
D6 = org_board.IO38
D9 = org_board.IO1
D10 = org_board.IO3
D11 = org_board.IO7
So this should be changed (or at least checked in the code) and @atomic summit should fix that graphic too.
atomic summit
So this should be changed (or at least checked in the code) and <@!1957651760358...
They are not wrong as there is no rule that states they need to match. Discussing this where you stared talking about it in my FeatherS2 thread now 🙂
Fixing the graphic is not an option as I ship printed cards with my boards and have shipped thousands.
slender iron
where are we at with the RTC on rp2040? https://forums.adafruit.com/viewtopic.php?f=60&t=177035
simple pulsar
On the wav playback problems on RP2040 / Pi Pico, I can reliably hang CircuitPython with some code I cited in: https://github.com/adafruit/circuitpython/issues/4208#issuecomment-804296887
onyx hinge
Here is the notes document for 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/1CXqf-nL0NEFt2aXH2ouDlDym6o2VWNz8JRLPNmWUnQQ/edit?usp=sharing
manic glacierBOT
slender iron
@tulip sleet are you game to review https://github.com/adafruit/circuitpython/pull/4388?
slender iron
this might be a record for me:
orchid basinBOT
hoary moat
Has anybody else seen Feather RP2040 stop running CP after ~36 hours on time? Neopixel is off and heartbeat LED is off. Running Friday’s S3 uf2
onyx hinge
@mental nexus thanks for that PR!
mental nexus
<@366059101203202052> thanks for that PR!
I was glad to finally catch this bug. The end behavior was totally weird since it depended on how fast it was drawn, since the refresh timing seemed to matter on the final display result. It lets me know I eventually need to learn how refreshes get triggered, but that’s yet another rabbit hole.
fossil gorge
Sorry I missed the meeting this week. <feelsbad.gif>
It's been... a day. Add to it the first day of kids physically back in school in 1+ year. Will catch up on the YT replay
ornate breach
I’m hoping to be settled enough with my work routine to really start at least lurking during the meetings.
manic glacierBOT
I've not had any particular i2c issues but looking at #4082 the clock speed looked wrong. Here's a view from Ikalogic SQ25 of a (Maker) Pi Pico talking to a generic SSD1306-driven screen and using the change contrast command to get something to capture (at 25MHz).
This has the 10k pullups on SSD1306 and whatever the RP2040 might be doing.
I too...
manic glacierBOT
I tried some 4 byte writes as I'm pretty sure I've read a discussion (perhaps with @Gadgetoid) about the special case bitbang code kicking in at <= X where X may equal 2.
The trace looks very different for 4 byte write. Here's first part. It's sped up a bit to 265kHz but it's still not the requested 400kHz.
manic glacierBOT
manic glacierBOT
jolly roost
Hi, we would like to add a new board (this one: https://github.com/watterott/senseBox-MCU) to circuitpython (our status: https://github.com/felixerdy/circuitpython/tree/main/ports/atmel-samd/boards/sensebox_mcu). The board is samd21 based and we are already able to run easy programms. The Board has 2 XBee Slots which can equipped with wifi, sd or lora-bees. One problem we are facing at the moment is, that all pins are not powered by default. The Board has the option to power each port individually (see: https://github.com/watterott/senseBox-MCU/blob/master/arduino/samd/variants/sensebox_mcu/variant.cpp#L199 and https://github.com/watterott/senseBox-MCU/blob/master/arduino/samd/libraries/senseBoxIO/src/senseBoxIO.h how it's done with arduino) When we try to power the xbee 2 ports with ```
xb2_pwr = DigitalInOut(board.XB2_PWR)
xb2_pwr.direction = Direction.OUTPUT
xb2_pwr.value = True
stuck elbow
@jolly roost wait, you are trying to power a radio module from a gpio pin?
@jolly roost you are aware that those pins can supply like 4mA of current only?
jolly roost
<@!570235973988777985> you are aware that those pins can supply like 4mA of curr...
the modules are powered with 3.3V directly from the regulator not from the gpio pins. But its possible to deactivate the whole power circuit for each interface (UART, I2C, XBee with SPI etc.) individually. In arduino it works easy and its also possible to power up the i2c power circuit on the board an read for example data from an bme680. But when I try the same with the XBee 2 PWR the board crashes
stuck elbow
what does it give as the reason for entering the safe mode?
jolly roost
You are in safe mode: something unanticipated happened. The microcontroller's power dipped. Make sure your power supply provides enough power for the whole circuit and press reset (after ejecting CIRCUITPY).
slender iron
@jolly roost learn.adafruit.com/how-to-add-a-new-board-to-circuitpython that safe mode is due to the brown out detector firing
you may have in-rush current when enabling a portion of the board that dips the main supply
stuck elbow
bigger caps \o/
onyx hinge
the microcontroller may have different brownout settings, but those aren't exposed in CircuitPython. A custom build apepars able to change them. ```#ifndef SAMD21_BOD33_LEVEL
// Set brownout detection to ~2.7V. Default from factory is 1.7V,
// which is too low for proper operation of external SPI flash chips
// (they are 2.7-3.6V).
#define SAMD21_BOD33_LEVEL (39)
and incidentally there you have our motivation for the value we use
@jolly roost
if your board has no flash chip then it may be safe to change the brown-out level see 37.10.3 Brown-Out Detectors Characteristics of the datasheet https://community.atmel.com/sites/default/files/forum_attachments/SAM-D21-Family-Datasheet-DS40001882B.pdf
jolly roost
thank you for your help. I'll try to change the brown out detection level and try again
onyx hinge
@slender iron as far as I could tell you also got all issues @tulip sleet raised resolved, but I was too cowardly to merge it with my approval.
slender iron
🙂 I'm working on imx now
onyx hinge
squee, much appreciate
slender iron
kinda daunted about testing all of the boards though....
tulip sleet
i was waylaid by a veterinary event; all done now
slender iron
everything ok?
onyx hinge
I have 1010/1020/1050/1060 evks, I can go through and test 'em all. @slender iron are you doing it in tinyuf2 or cp or both?
tulip sleet
Gryphon had a abcess on her lip that opened up; we didn't know there was an abcess, and she was bleeding noticeably. She may or may not need dental work later. We though it was more serious.
she has been in rough shape for a couple of weeks
slender iron
that's good news overall I think
tulip sleet
yah
slender iron
I remember you mentioning that
tulip sleet
she is quite old
slender iron
I have 1010/1020/1050/1060 evks, I can go through and test 'em all. <@!2527171...
just cp. tinyuf2 isn't on my radar
I'd love to hand it off to you 🙂
I kinda feel like I should do the other external flash settings too. That way we'd be on nvm.toml completely
onyx hinge
okay -- well, I am pretty sure that when you boot tinyuf2->cp, that it's tinyuf2's flash_config that is used even after it hands off
but I could be wrong, etc
slender iron
@onyx hinge ya, that sounds right actually
onyx hinge
but to keep in mind when testing!
slender iron
because it's the bootrom that sets everything up
should we remove the flash settings from CP completely?
slender iron
I guess that'd mess up teensy
I wonder if we want a separate repo for imx flash
then we pull it into cp and tinyuf2
onyx hinge
could be?
it doesn't belong "in" nvm.toml, does it?
"nvm.jinja"
heck maybe somebody wants your rp2040 generator for their pico-sdk project
slender iron
ya, that's a good question
nvm.toml could have the python script or a couple for different techniques
least common denomiter settings vs detect and change settings
I think I need to look at https://github.com/adafruit/circuitpython/issues/4431
before I switch away from rp2040
I need to think on the imx some more
lavish steppe
i just rebuilt circuitpython and had to manually install cascadetoml with pip btw dont know if i should make it an issue or what but never had to do that before
tulip sleet
i just rebuilt circuitpython and had to manually install cascadetoml with pip bt...
the build process now needs that. We should probably put a check in a makefile for it.
lavish steppe
the build process now needs that. We should probably put a check in a makefile f...
yah maybe update this page saying you need to install cascadetoml with pip https://learn.adafruit.com/building-circuitpython/build-circuitpython idk tho
oh btw the digitzer works great on windows but acts super weird on mac, dont have linux to test but im guessing will be weird there too. it works but sometimes
tulip sleet
it's been very difficult or impossible to make an HID definition for certain devices that works across the three major platforms. Same has been true for Gamepad
lavish steppe
that must be super annoying 😦
tulip sleet
yah, frustrating ... OK, I have updated the Learn Guide to tell you to do pip3 install -r requirements-dev.txt.
lavish steppe
you added it to before the builder has git cloned the repo isn't requirements-dev.txt in the repo?
tulip sleet
oops, yes :/
lavish steppe
haha no problem glad i could help and not just ask questions 🙂
orchid basinBOT
mental nexus
It seems like the imx boards are going to be the next exciting area for CP. Is there a suggested demo board to use to get started on the RT1011 while we wait for the Metro M7 that’s coming soon? Or if coming soon is really soon, soon, then is it best to wait a bit for that?
ionic elk
@tulip sleet can I chat with you real quick about a sleep thing?
tulip sleet
sure, in a minute
ionic elk
np, whenever you have a bit
tulip sleet
@ionic elk ok, in Amelia
slender iron
It seems like the imx boards are going to be the next exciting area for CP. Is t...
Ya, we're working to polish it up for release. the 1010 EVK is good because it has swd: https://www.digikey.com/short/pz34z0v5
onyx hinge
huh, that's weird. I guess the system doesn't really have any way of noticing duplicate entries. ``` { MP_OBJ_NEW_QSTR(MP_QSTR_D0), MP_ROM_PTR(&pin_GPIO_09) },
{ MP_OBJ_NEW_QSTR(MP_QSTR_RX), MP_ROM_PTR(&pin_GPIO_09) },
...
// UART
{ MP_OBJ_NEW_QSTR(MP_QSTR_RX), MP_ROM_PTR(&pin_GPIO_09) },
@gilded cradle re Justin's question ^^: circuitpython-build-tools will need a new release for your changes to be pushed to PyPI, and subsequently used when the bundle release happens tonight.