Hmm. Then have to add the README or not too....
#circuitpython-dev
idle owl
lone axle
the README inside of factory-reset will get created by default but deleted by that post_gen along with the directory if not needed.
lone axle
Your bash scripting is sharper than mine if so
idle owl
I doubt it
here's all I did
I added the factory-reset folder to the cookicutter.directory_name directory.
Then added this to the post_gen_project.sh
{% if cookiecutter.dev_board in ['n', 'no'] %}rm -rf factory-reset{% endif %}```It never occurred to me to bash script it. I used the cookiecutter ifs.
Is that wrong?
or if it works, it works?
lone axle
Ahhhhh yep that seems like it would work to me. And Doh! it didn't occur to me to use the template statements to write the if statement. I was trying to figure out how to make Ifs in bash.
idle owl
Ahhhhh yep that seems like it would work to me. And Doh! it didn't occur to me t...
We both went with what we knew ๐
Ok, great. now to test the other changes I added a few minutes ago
They work after a little tweaking! Turns out there was already a thing in there to automatically add the board files, but I had no idea what it was. It wasn't going to work the way it was in there, so I tweaked it a bit, and made it so you have to type in the name of the board file as it is in the repo because they're never consistent, they're what Limor types out when she designs them which can be anything.... ๐
Ok. I'll update the README, and then have Dylan test it without explaining anything.
@lone axle Thanks so much!
lone axle
Yep you're welcome
lone axle
```# Remove factory-reset if not needed
{% if cookiecutter.dev_board in ['n', 'n...
One thing that may be worth another small tweak. I'm not sure whether the cookie cutter variables and/or the in check in the if statement is case sensitive. If the default is uppercase "N" and it is case sensitive it's probably worth adding "N" and maybe "No" to the list that it checks.
turbid radish
?serverinfo
digital shoreBOT
adafruit
turbid radish
Only 8 more users needed!
idle owl
One thing that may be worth another small tweak. I'm not sure whether the cookie...
I made it n ๐
I caught that too
lone axle
nice.
lone axle
Huzzah!
idle owl
Thanks for your help ๐
lone axle
Before this I didn't even know there was a pcb cookie cutter. It's neat the way it does some of the git repo setup in the post gen. I may end up taking some inspiration from that for some other projects.
Yep you're welcome.
idle owl
Before this I didn't even know there was a pcb cookie cutter. It's neat the way ...
We only really use it intermally ๐
tulip sleet
I did not make a 7.x branch yet. It is incompatible with 7.x, right?
onyx hinge
at least not 7.1
onyx hinge
there are sure to be problems to shake out, as well as the rgbmatrix thing
so it's fine by me as long as branches are used 'somehow' to keep it from being in 7.1.0
which scott's plan sounded sensible, and just affects people who need to PR bugfixes into 7.1.0
tulip sleet
I will make a 7.1.x branch off main. That is what we did in the past for 6.1.x, 6.2.x, and 6.3.x.
(well maybe 6.2.x and 6.3.x were not off main)
slender iron
๐
slender iron
@tulip sleet want to merge #5615 after and also tag it as 7.2.0-alpha.0?
digital shoreBOT
adafruit
tulip sleet
!serverinfo
Keep an eye on joins in #bot-mods
tulip sleet
<@!329766224093249548> want to merge #5615 after and also tag it as 7.2.0-alpha....
If I tag that, I think it will eclipse the 7.1.0-beta.n releases in circuitpython.org
or you mean just tag?
tulip sleet
Do you want to review #5615? I have not looked at it at all.
slender iron
I had looked it over already. Just made it official
manic glacierBOT
orchid basinBOT
This note is visible on https://github.com/adafruit/circuitpython-org/pull/802, and is what provoked this PR:
turbid radish
?serverinfo
digital shoreBOT
adafruit
manic glacierBOT
I testedwith the above airlift setup all these cards, still using sdcardio. (see my full program listing above)
- kingston 32gb micro sdhc class 10
- no-name 128MB card (https://www.adafruit.com/product/5250)
- g.skill micro sd hc 32GB class 10
- sandisk edge 16GB micro sd HC class 10 (possibly https://www.adafruit.com/product/2820)
- 2GB micro sd "S...
manic glacierBOT
onyx hinge
@slender iron do you care whether a "nopsram" version of the esp32s3 devkit is added to CP?
crimson ferry
I was about to load up an S3 build, but realized there were no boards yet. @onyx hinge are you adding the three extant variations of S3 DevKit boards?
onyx hinge
@crimson ferry I'm close to making a PR for the devkitc (psram & no-psram versions), just doing final verifications...
they come up to CIRCUITPY and can flash their neopixel
I also have the "box", but it did not successfully boot into CIRCUITPY yet
idle owl
@onyx hinge Blinky NeoPixel? Who needs more!
lone sandalBOT
tulip sleet
?serverinfo
digital shoreBOT
adafruit
tulip sleet
@stuck elbow @analog bridge @onyx hinge @idle owl @slender iron Note that we branched 7.1.x a few hours ago so we could do some incompatible stuff inmain for 7.2 or later, so fixes for 7.1.0 are better PR'd against that branch instead of main, so we don't have to backport. It is a bit easier to merge from 7.1.x than cherry-pick from main. Thanks.
manic glacierBOT
onyx hinge
@crimson ferry hey if you see this can you tack the "box" onto your pull request with espressif? https://www.adafruit.com/product/5290
crimson ferry
@onyx hinge Done.
lone axle
Is it intended to pass a BOARD argument for make fetch-submodules?
I am getting this:
โฏ make fetch-submodules
Makefile:27: *** You must provide a BOARD parameter. Stop.
@lone axle try running in the top -level instead of in ports?
lone axle
Ah, yeah I do think that is the issue. In the top level I get this:
โฏ make fetch-submodules
make: *** No rule to make target 'fetch-submodules'. Stop.
I'm guessing I have not updated my local repo properly
blissful pollen
did you rebase your local to adafruit/main?
lone axle
yep. Was on wrong branch that wasn't tracking a remote.
blissful pollen
(I may have made that mistake myself ๐ )
lone axle
did you rebase your local to adafruit/main?
I didn't. I tend to stay away from rebase because I don't fully understand it yet and am a bit scared of rewriting things accidentally.
manic glacierBOT
blissful pollen
I'm far from a git expert but my usual is:
git checkout main
git rebase adafruit/main
git checkout -b newbranch
Extending Native Classes
onyx hinge
@analog bridge hey did you happen to test spi with the esp32-s3? It's not working for me (always raises an exception that says 'invalid pins', though I'm not sure that's accurate because there are a lot of reasons the function can return ESP_ERR_INVALID_ARG.
manic glacierBOT
onyx hinge
@analog bridge in fact ... a debug build sez E (27970) spi: spi_bus_initialize(762): invalid dma channel, chip only support spi dma channel auto-alloc
onyx hinge
the display of the esp32-s3-box still doesn't configure. oh well.
time to set this aside for a bit
dusk mauve
I got an S3-box but havenโt tried it out yet. Should have time on Monday if you want some testing/collaboration assistance @onyx hinge
Was looking at the schematic and saw some interesting devices on board that may need new drivers ๐ค
onyx hinge
@dusk mauve nice! I hope to have a pull request put in soon that will add basic support.
haven't gotten the display to work yet, and the audio features for sure need work to support.
dusk mauve
What kind of display IC/interface does it use? I was saving that for last during my schematic dive but fell asleep before I got to it ๐
onyx hinge
@dusk mauve their BSP seems to have it as an ST7789 -- https://github.com/espressif/esp-box/blob/8fd70711a9f08d9803c807733ba46bf41af41e42/components/bsp/boards/esp32_s3_box.h#L31
#elif LCD_DISP_IC_ST
ESP_ERROR_CHECK(esp_lcd_new_panel_st7789(io_handle, &panel_config, &panel_handle));```2.4" 320x240 ST7789 should be well-supported by CircuitPython but I got no display (all white if I turn on backlight)
woven kite
I don't suppose you've had any luck with wifi on the esp32s3 devkitc boards @onyx hinge?
onyx hinge
@woven kite I didn't actually try it yet
did you try it and have it not work? If so, please file a github issue. we need to track things that are broken so they can get fixed.
hmmm ```Adafruit CircuitPython 7.2.0-alpha.0-5-g73ee59881-dirty on 2021-12-04; ESP32-S3-DevKitC-1 with ESP32S3
import wifi
[tio 09:23:28] Disconnected
Yeah same experience here
onyx hinge
I have some additional information from the debug port so if you'd like I can file the issue
woven kite
This is actually my first dip into circuitpython and hardware in general, so that would be great ๐
onyx hinge
@woven kite ^^ if you have a github account you can subscribe to the bug using the icon in the right hand column
I'm going to head out, happy I made some progress with the S3 boards but there's a lot more to do!
woven kite
Great stuff ๐ Thanks!
manic glacierBOT
lone sandalBOT
analog bridge
@onyx hinge I haven't done much testing beyond blinky leds and a working repl on the esp32s3.
lone axle
<@!351136009922019331> <@!747673663419187281> <@!426200547024961539> <@!33022745...
I think I may have found a fix for https://github.com/adafruit/circuitpython/issues/4867 . Do I understand you correctly that it would make sense for me to make my PR branch from 7.1.x instead of main. Or is that meant to be used in some other circumstance?
tulip sleet
I think I may have found a fix for https://github.com/adafruit/circuitpython/iss...
Yes, then the fix will immediately be in the 7.1.x branch and be there when we make the next beta. Also it's probably easy to merge from 7.1.x to main, because merging the other way will bring in things we don't want in 7.1.x, so we'd have to cherry-pick. Thanks.
manic glacierBOT
manic glacierBOT
lone axle
I suspect the workflow on this PR is essentially stuck forever: https://github.com/adafruit/Adafruit_CircuitPython_PyBadger/pull/54 should it be cancelled instead of waiting for github actions to notice and I assume stop it eventually?
part of the code that was added with that PR is a while loop to try_lock() the i2c bus here: https://github.com/FoamyGuy/Adafruit_CircuitPython_PyBadger/blob/346e8d35428cdbb07cc314c118cb622ffcf0410d/adafruit_pybadger/pybadge.py#L67-L68
while not i2c.try_lock():
pass
It seems like sphinx is trying to execute the code and getting stuck in an infinitely loop because try_lock() is never returning true. If I comment those lines out the sphinx build does run normally locally.
I took that code from the I2C scan example in the essentials guide. Is there a different way that libraries should be coded to attempt the i2c lock? Perhaps only try a few times instead of infinitely with the while loop? Or alternatively is there some way to tell sphinx that it needs to ignore those lines or something like that?
when running locally it gets stuck on "reading sources":
crimson ferry
that seems like a problematic thing for Sphinx to do since it has no idea what the referenced function does
lone axle
I think this post is describing similar root issue https://stackoverflow.com/questions/37818923/running-sphinx-html-make-gets-stuck-after-reading-sources
formal summit
Hello, Is this https://learn.adafruit.com/ulab-crunch-numbers-fast-with-circuitpython/a-simple-benchmark a good benchmark for testing different c libraries?
lone axle
Hello, Is this https://learn.adafruit.com/ulab-crunch-numbers-fast-with-circuitp...
This is the only example of bench-marking things in circuitpython that I am aware of so I'd say it is probably as good as anything at this time. To some extent it may depend a bit on what the c libraries are actually doing though. The benchmarking could probably be tailored to some extent for differen types of c libraries.
ornate breach
I'm curious what i'm missing?
make: *** No rule to make target `py/emitcommon.c', needed by `build-odt_cast_away_rp2040/genhdr/qstr.split'. Stop.
wait... i see what it is...
for some reason when I made a new branch it deleted it...
i have... no idea why the branch deleted a bunch of files.. so strange..
ornate breach
only 1000 more pull requests and we'll be at 6666
crimson ferry
Dan, can you look at this reply to our PID request and see what you think? I'm not exactly sure how to articulate the threshold for re-using vs. new. We have a build check for each unique board. https://github.com/espressif/usb-pids/pull/29#issuecomment-986378221
tulip sleet
Dan, can you look at this reply to our PID request and see what you think? I'm n...
I think CircuitPython cares less. Arduino does use the VID/PID to distinguish the boards for setting purposes.
I asked Limor about this but our internal Slack made it disappear: it was more than two weeks ago - grumble, so I can't remember her exact reply. Something like: "they are free"
but if Espressif is worried, we can scale back
crimson ferry
I've only asked for UF2 Bootloader and CircuitPython, no Arduino.
crimson ferry
@tulip sleet, do we need the build check for unique PID on each board build?
tulip sleet
<@329766224093249548>, do we need the build check for unique PID on each board b...
hmm, good point, that will catch this
spiral elk
mu I think used to use that for checking what board was connected but not so any more.
tulip sleet
yes, we changed that, because it was a lot of work to keep the list up to date
and we could ID the boards by the Interface name instead
crimson ferry
board_id?
tulip sleet
by the USB interface name presented to the host
"CircuitPython USB CDC"
we can discuss this tomorrow in the meeting
crimson ferry
OK, sounds good, thanks.
0x7*** still allows for 4096 PIDs. They use 0x8*** for 3rd-party boards. I'd assume the other 14 PID first-digits on their VID are RFU.
lone sandalBOT
digital shoreBOT
adafruit
onyx hinge
yay I figured out why the display on the esp32s3 box wasn't working, I'd forgotten to specify the display reset pin
analog bridge
@tulip sleet If I understand correctly... any bug fixes should go in 7.1.x and will later on get merged into main or separate PRs need to be made for the two branches?
tulip sleet
<@!329766224093249548> If I understand correctly... any bug fixes should go in `...
Yes, it's much easier to merge from 7.1.x to main. But if you are fixing things due to the ESP-IDF upgrade, then we will just do that on main. We won't backport the IDF upgrade to 7.1.x. We tagged 7.2.0-beta.0 (but will not release) in prep for the next series of betas after 7.1 final
Would you mind re-revewing #5670 and also reviewing #5672? There were some uncrustify changes worth including as well Thanks.
analog bridge
okay... regarding #5672, instead of merging it into main, can we include that fix with the 7.1.x merge since the uncrustify fix isn't causing pre-commit to fail
tulip sleet
sure, I will just close it
onyx hinge
this is odd .. I have in board.c: common_hal_displayio_display_construct(display, ..., /* backlight_pin /* &pin_GPIO45, ...); but CircuitPython doesn't activate the backlight. In the repl, I can activate it with PWM or as DigitalIO. Assigning board.DISPLAY.brightness does nothing. So after startup, doing this at the REPL turns the LCD backlight on and it operates until I hit ctrl-d: ```>>> microcontroller.pin.GPIO45
board.LCD_CTRL
p = pwmio.PWMOut(board.LCD_CTRL)
p.duty_cycle = 65535
this is the esp32s3-box
shouldn't constructing the display "take over" the passed in pin?
tulip sleet
is this related to the issue deshipu was having as well with backlight? Interesting it's on the ESP32-S3, I think deshipu's problem was on atmel-samd. Maybe it's a general problem with displayio
onyx hinge
I suppose it could be somehow related ...
tulip sleet
particularly the later comments
onyx hinge
If I initialize the display via python display = ST7789(bus, width=320, height=240, backlight_pin=board.LCD_CTRL) the backlight works including when I ctrl-d...!
tulip sleet
is that display long-lived
it disappears on ctrl-D, right? it's like some pin ownership problem
onyx hinge
no, if I initialize it from code.py it continues working including the backlight
.. after ctrl-c'ing to the repl or whatever
sooo if I put the display init in a different function and run it a bit later, it's OK. in main.c: ```c
// displays init after filesystem, since they could share the flash SPI
board_init();
// Reset everything and prep MicroPython to run boot.py.
reset_port();
// Port-independent devices, like CIRCUITPY_BLEIO_HCI.
reset_devices();
reset_board();
- board_display_init();
that positioning makes more sense. The reset_ lines do a lot of cleanup including resetting all the PWM stuff
onyx hinge
it would be a bit of a pain to change that for all boards though ยฏ_(ใ)_/ยฏ
tulip sleet
I was just looking at this yesterday when debugging #5665
onyx hinge
I wonder why it appears/appeared to work for others.
onyx hinge
I looked at pygamer, there's not an explicit 'never reset' for its backlight pin
tulip sleet
maybe it's worth putting in some print statements to track when the PWM is initialized, etc.
onyx hinge
the pin is still not properly 'claimed'. If I try to put PWM & digitalio on the same pin from the repl, it balks as expected:```python
p = pwmio.PWMOut(board.G11)
d = digitalio.Di
DigitalInOut Direction
d = digitalio.DigitalInOut(board.G11)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: MOSI in use
but I can 'take over' the display's backlight pin, which messes it up until hard-reset:python
board.DISPLAY.brightness = .25 # makes display dim
d = digitalio.DigitalInOut(board.LCD_CTRL) # makes display go black
d.switch_to_output(True) # display goes on again
there's clearly something wrong with managing the state of these things
tulip sleet
I'd say open a clear issue, and maybe you will feel inclined ๐
idle owl
@lone axle Do you need the cell anchor point PR tested? Or are you pretty confident about it. (I will test the PyBadger PR on an LC)
lone axle
<@!382939733107408897> Do you need the cell anchor point PR tested? Or are you p...
I am pretty confident with the cell anchor point one. Thank you
idle owl
Keen.
idle owl
@tidal kiln Opinion time. On the KB2040, would you include SCL/SDA as touch pins, even though they're only on the STEMMA QT connector? You could easily attach a cable to it to use the pins, and there's not a dedicated I2C device on the board.
I feel like yes, but you always have excellent counter opinions.
tidal kiln
hmm. let me look at schem.
idle owl
Ok
tidal kiln
they have pullups
tidal kiln
yah, i think so
onyx hinge
@tulip sleet do you understand how black changed? I thought pre-commit was supposed to ensure that the dependencies stayed at the same version.
or is it the version of black from requirements-dev.txt, which we didn't pin
tulip sleet
the pre-commit run is done by a script, not with black directly
yeah so from requirements-dev.txt, I think
onyx hinge
I see. makes sense, slightly unfortunate.
tulip sleet
unlike the libraries
tulip sleet
when I ran pre-commit locally, it also had uncrustify complaints. I don't understand why those are not showing up in the CI runs, though
idle owl
Ooh I think I can do all of the CP Essentials templates without soldering this KB2040. ๐
tulip sleet
@idle owl oops, just realized now I am doing the meeting. Not a problem, but was I maybe going to trade for later in the month, or is that separate?
idle owl
<@!330227457296957440> oops, just realized now I am doing the meeting. Not a pro...
Hah, been there. That was me last time I ran it. Completely forgot until a half hour before. Jeff asked for a trade, but I was already going to ask for a trade for that exact day, so I took it.
tulip sleet
I guess if there was any trading it would be with Scott for 12/27
and he is not here today anyway
idle owl
I would actually offer to trade you for the 20th, but today is apparently the quarterly fire drill for the building. Sometime between 9am and 5pm. Sigh.
So I don't want to have a fire alarm going in the background of the meeting.
tulip sleet
wouldn't you have to leave anyway, because it's a drill?
idle owl
Apparently not? that's what's weird.
They do it to make sure the system and the elevators work I guess.
tulip sleet
a "test", not a "drill", i guess
idle owl
Yeah they named it badly.
They do a lot badly here.
The elevators go to ground and stop working during an actual fire, apparently.
They test the alarm monthly.
But this is bigger.
But they did not say "evacuate when drill happens" so I'm not leaving.
tulip sleet
that makes sense, the firefighters have a special key so they can work the elevator during a fire
idle owl
Right.
@tulip sleet To that end, if it's going off during one of my things (i.e. Libraries section, my hug reports or status updates), I'll let you know to read it off. You don't have to remember my whole spiel for Libraries. You can simply read what's there. ๐
@tidal kiln Sigh.... thwarted.
I thought I fixed the script to report that.
Apparently not.
Oh, my original did. My updated version did not.
I was going to merge them and never did.
tulip sleet
LMK if you need another test!
worked! thanks
idle owl
I was like, why isn't this code working right? Oof...... Adafruit CircuitPython 5.3.0-1-gdd53e9a2b on 2020-07-10; Adafruit Circuit Playground Bluefruit with nRF52840
Maybe that's why, Kattni.
Yeah update fixed it.
ornate breach
This and others in Kattniโs coding adventures!
I have similar fun problems at work like forgetting to update a dependency
idle owl
Ooh I think I can do all of the CP Essentials templates without soldering this K...
In other news, nope. I have to add pulldowns to the touch pins, which means soldering it up. ๐ฆ
I mean I could fake it badly for testing, but I have to include an image in the guide so..... yeah. Kinda need it to look proper.
crimson ferry
@tulip sleet @onyx hinge I was going to make an In The Weeds topic for USB PID, but looks like we can handle it in the GitHub PR comments? (I also think there's a broader design discussion at a later time about the purposes and degrees of uniqueness for various board identifiers.... user-facing, build-facing, code-facing, host-facing)
tulip sleet
<@!329766224093249548> <@!426200547024961539> I was going to make an `In The W...
Sure
onyx hinge
@crimson ferry whatever way works best for you. fwiw I have one of the nopsram variant boards on the way, I am also interested in finding out whether we can dynamically figure out if psram is present and then reduce the number of builds..
I think with some small rework I just did it may "just work" but I don't have a way to test
crimson ferry
๐
lone axle
I am set up to do a backup recording
onyx hinge
regularly scheduled meetings coming up: December 13, 20, 27; January 3.
idle owl
regularly scheduled meetings coming up: December 13, 20, 27; January 3.
Putting it In the Weeds.
onyx hinge
To my knowledge, we support two "chips" with risc-v: fomu (which is an fpga with a soft risc-v cpu loaded on it); and, just recently the espressif esp32-c3 (boards by name: ai_thinker_esp32-c3s and microdev_micro_c3)
onyx hinge
belated hug report for jerryn testing alongside me with the sd cards
solar whale
๐ช
onyx hinge
create_and_get_feed ๐ค ๐ก
idle owl
create_and_get_feed ๐ค ๐ก
Might not be called that, but that's the idea.
onyx hinge
I think it's great! not questioning the name, either
idle owl
๐ Hooray for me finding the error. Broke my Feather, fixed it, and immediately verified I could break it again. ๐
idle owl
I think it's great! not questioning the name, either
Fair enough! Thanks. ๐
IS31FL3741 - LED driver.
Yes
blissful pollen
yeah IS31FL3741 but I get tired of typing the full thing ๐
idle owl
yeah IS31FL3741 but I get tired of typing the full thing ๐
I wrote the guide. I agree entirely. ๐
errant grail
IS<something, something>41
blissful pollen
@idle owl Forgot to ping you as I got it going yesterday - the IS31 pixelbuf thing seems to work perfect with the LED animations as is!
idle owl
<@!330227457296957440> Forgot to ping you as I got it going yesterday - the IS31...
That's good to hear!
thorny jay
onyx hinge
I verified that the aarch64 toolchain from arm.com [https://developer.arm.com/tools-and-software/open-source-software/developer-tools/gnu-toolchain/gnu-a/downloads] uses python2 in its gdb. I wonder if scott is using the one from his Linux distro, which is based on python3 [https://aur.archlinux.org/packages/aarch64-none-elf-gdb/]
@idle owl my calendar's not showing that we were swapping the 13th & 20th .. were you waiting for me to change it, or was I confused? I think you had agreed to take the 13th from me
idle owl
<@!330227457296957440> my calendar's not showing that we were swapping the 13th ...
I just hadn't gotten to it. Yeah we swapped.
onyx hinge
I can take the 27th from Scott if it's a problem
onyx hinge
thanks
idle owl
Fixed!
errant grail
Thanks!
onyx hinge
I'm heading out for now. ๐
idle owl
I need to head out as well. Later folks!
ember iris
Thanks all!
manic glacierBOT
idle owl
@lone axle Hey! Do you have a few minutes to chat? If not, no rush. Please ping me if/when you do.
lone axle
<@!382939733107408897> Hey! Do you have a few minutes to chat? If not, no rush. ...
Yep I do.
idle owl
Great. So here's my idea. Part of the issue with the existing PRs across the libraries is someone to actually test them. A TON of them are super old, and will need a lot of work. Some of them simply need to be closed. I'm wondering if you're interested in going through the list of library PRs on circuitpython.org/libraries and seeing what's what. Jose David went through it not that long ago, but he doesn't have the permissions you do to deal with fixing things where needed or closing PRs if it makes sense. You don't have to make that decision on your own, obviously, we can make those decisions together. But! You are now in a position to get the hardware where necessary, and can make changes to PRs if that makes sense.
You also don't need to do this all at once.
A one or two a week even would be a good start.
@lone axle What do you think? You're welcome to decline as well. ๐
I even have a list from Jose David of where things were at when he created the list. And his thoughts on what was needed. There have been changes since, but it might be worth a look still. https://github.com/jposada202020/PR_CP/blob/main/list.md
Have had that open in a tab since he sent it.
lone axle
@idle owl Sounds good to me, I am up for doing that. I was actually just looking through some the other day and found some in need of pre-commit run. On ones like that, if the original user allowed maintainers to push to their branch. And they haven't responded in a little while since being asked about it, should I go ahead and run pre-commit and add a commit to the PR branch so that it can move along to getting a full review.
idle owl
<@!330227457296957440> Sounds good to me, I am up for doing that. I was actually...
Yes, that's a great idea. That's exactly what I'm thinking. I feel like you can push to a PR regardless of that allow maintainers thing. Or you can at least pull changes into a new PR, and the commits are still attributed to the original user (Tekktrik has done this multiple times).
You can let me know when you find one you think should be considered for closure, or if you have any questions along the way.
lone axle
Yep I know how to do that as well so I can do that on any if they do prevent me from pushing.
idle owl
Excellent.
@lone axle Thanks so much! This should be good. I would like to see us catch up. 68 isn't a ton, but there are some that are ancient, and I hope we can get caught up with that and then try to stay on top of things better.
Whatever cadence works for you is fine with me.
lone axle
Yep you're welcome. Happy to help out where I can to catch up!
onyx hinge
crumbs, updating my kaluga has left me with the camera not working. this was with the idf4.4 version, trying going back to 7.1.x..
it hardfaults during initialization of the camera
the 'watchdog reset' that doesn't make it enter safe mode
not strictly speaking a hardfault
tulip sleet
Here is the notes document for Mondayโs 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 youโll be attending the meeting - itโs super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and weโll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1Bwnt-pbFcrfo1gn_tH4L0HXkMXZbcqG5eoAOfPK1xco/edit
tidal kiln
@idle owl hey. i think it's been many months since i've done a CP lib release. anything new to it? just wanna/merge release this:
https://github.com/adafruit/Adafruit_CircuitPython_LTR390/pull/13
idle owl
<@!330227457296957440> hey. i think it's been many months since i've done a CP l...
Nothing new. Merge PR. Click on releases. Increment the tag, put the boilerplate in there, release. Done.
tidal kiln
cool. thanks.
onyx hinge
crumbs, updating my kaluga has left me with the camera not working. this was wit...
no, more likely it's the good old "enabled jtag, so turning on the camera [which shares pins with jtag] scrambles the microcontroller state but good"
which I forget every time I do it to myself
idle owl
@tulip sleet I usually turn off public editing, and leave editing available to the four of us on the old notes doc, BTW. I guess it really doesn't matter.
tulip sleet
<@!329766224093249548> I usually turn off public editing, and leave editing avai...
Twice now I tried to turn off "edit with the link", and it turned off my access too. So I couldn't change it back, since it's owned by someone else.
idle owl
Twice now I tried to turn off "edit with the link", and it turned off my access ...
OK, I'll get a screenshot next week if I remember of where to change it.
You're simply changing it in a different spot than you should be.
tulip sleet
You're simply changing it in a different spot than you should be.
I hit "Change" here:
and changed it to "View"
tulip sleet
(this is from the Dec 13 one, I didn't change it, of course)
tulip sleet
maybe something changed
I think it is because I was not listed explicitly on the template, so whenever the template was copied by someone other than me, I did not get person access, just link access. I just added myself as an editor on the template.
so that should fix it
idle owl
I think it is because I was not listed explicitly on the template, so whenever t...
Fair enough. We'll find out!
ornate breach
Thanks Scott! ๐
slender iron
np ๐
manic glacierBOT
mortal mica
Getting some odd build failures on the PR I made. Is this a github issue or an action issue or something else?
slender iron
I'm looking into it now
it's a weird bug that happens if you edit a file on github
idle owl
@solar whale Eliminates all of the feed setup lines, and the send_io_data helper I wrote. py io.send_data(io.create_and_get_feed("test-feed")["key"], temperature)
Creates the feed if it doesn't exist, sends to it once does, or sends if it already exists.
solar whale
<@!343570430042308609> Eliminates all of the feed setup lines, and the `send_io_...
Excellent! Thank you!
idle owl
Excellent! Thank you!
I'll PR it tomorrow and tag CPLibians. I'd appreciate having it tested. ๐
solar whale
Sure. Be happy to test it tomorrow.
idle owl
Thank you!
lunar gull
By the way, I am trying to track down the elusive requests issue with the MagTag Google Calendar now.
idle owl
@lone axle While I'm remembering it, the mag_tag.py file in PyBadger should have been named magtag.py. I'm not sure how much it will break to rename it but it's worth considering. I didn't see it go by to begin with. Sorry about that.
Heading out for the night. Have a good one, all!
lone axle
<@!382939733107408897> While I'm remembering it, the `mag_tag.py` file in PyBadg...
I can work on changing it and finding usages in examples and learn. I don't think are too many.
idle owl
I can work on changing it and finding usages in examples and learn. I don't thin...
Ah right. Sounds good. Thanks!
slender iron
can't printf debug when printf is broken
slender iron
(I spotted the issue in the code. Now printf works)
orchid basinBOT
tulip sleet
it's a weird bug that happens if you edit a file on github
Blobless clones look very interesting. Maybe that is the solution for the big broadcom repo??
slender iron
yup, I'm thinking that for CP too potentially
might be able to combine it with shallow too: https://github.com/adafruit/circuitpython/pull/5653/commits/d7822879ea0f89843491f22a105c37fa253cf131
tulip sleet
yes, that too. Is it an option in the checkout action, I wonder?
slender iron
clone and fetch I think
tulip sleet
the article seems to recommend blobless over treeless for what we need, since we may need to find a certain tag?
also there's that submodule fetch warning, though it sounds like we just need to not recurse; fetch the submodules separately
slender iron
I bugged the blog author on twitter about adding actions/checkout support
slender iron
@solar whale I fixed the pi4 by adding the .dtb files to the sd card I think
solar whale
<@!343570430042308609> I fixed the pi4 by adding the .dtb files to the sd card I...
Sounds great. I donโt know exactly what that entails but it sounds encouraging!
slender iron
in the firmware directory there are a bunch of .dtb files
lunar gull
<@&356864093652516868> I figured out what the issue is with the MagTag examples. It's the combination of Google Auth having its own request object with sessions that create sockets that will be "locked" to that session, and then MagTag/PortalBase trying to clean sessions but failing because there's still sessions in the pool. I am not sure if I can submit a PR for this, but maybe Melissa sees this?
slender iron
they need to be on the sd card too
though I did fix a build thing too...
maybe that was needed too
hoary moat
.dtb files are kind of important ๐
solar whale
Ah. Ok. Iโll be patient and wait for a PR
slender iron
it's pretty simple
CHIP_VARIANT = "bcm2711" in mpconfigboard.mk
(adding quotes)
though zero2 doesn't have them
zero2 style may be the default
solar whale
Ok. Iโll try that tomorrow. So use โ for now
slender iron
ok, np. thanks for testing it!
the boot process is different between the cm4 and the pi4
solar whale
I started reading up. It is complexโฆ
slender iron
this works for my pi4
solar whale
Great. Thanks. I can give it a try in a little while.
slender iron
no rush. will have a PR with this fix and full uart support before I call it a night
figured I should look at this before finishing the learn guide ๐
solar whale
Thanks for pressing ahead with it!
slender iron
๐
solar whale
this works for my pi4
Woohoo! It works!!
no fb
Adafruit CircuitPython 7.1.0-beta.1-27-g19211390d-dirty on 2021-12-06; Raspberry Pi 4B with bcm2711
Board ID:raspberrypi_pi4b
Auto-reload is on. Simply save files over USB to run them or enter REPL to disable.
code.py output:
Code done running.
Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 7.1.0-beta.1-27-g19211390d-dirty on 2021-12-06; Raspberry Pi 4B with bcm2711
>
> >>>
>
> ```Thanks to @jerryneedell for testing. The logic looks good to me, but I did not test.
Good grief, 13k file changes everytime they change something?!
lunar gull
and then GitHub dips
manic glacierBOT
orchid basinBOT
mortal mica
@gilded cradle Are there docs on how to use the new aliasing feature? I have several hardware boards that use the exact same CP build and would rather alias them to a single build with docs on each board compared to having a common download page that documents all permutations. (keyboards with 18, 44, 61 and 87 keys - all mapped out to Pykey60)
analog bridge
The ESP32-C3 datasheet lists GPIO8=0 and GPIO9=0 combination as invalid but I consistently get the following result with this boot combination:
ESP-ROM:esp32c3-api1-20210207
Build:Feb 7 2021
rst:0x1 (POWERON),boot:0x0 (USB_BOOT)
wait usb download
If anyone has an ESP32-C3, can you check for this behavior.
proven garnet
I know I'm late for the holiday, but I've got a menorah that uses ESP32 to get time as well as use an API to figure out candle lighting times. More pertinently, @tulip sleet I'm going to use asyncio to have it flash the lights while connecting to the network. It's still WIP but didn't know if you were looking for examples of asyncio implementations. Excited to test it out once I wire everything on the breadboard!
Debating using asyncio for making the lights flicker at varying rates but may just stick to on/off
tulip sleet
I know I'm late for the holiday, but I've got a menorah that uses ESP32 to get t...
Very intereresting. We don't have a non-synchrous version of the network fetching, so it may stall when fetching the data. But if it's only on startup, it doesn't matter too much. In the long run we want to provide something like aiohttp. There is a low-level stream impl for networking in the asyncio library but I haven't tested it.
I am not sure you really need asyncio here, because the fetch does not need to be interleaved with the lights.
orchid basinBOT
proven garnet
Got it. I have it set for multitasking flashing the lights while it connects to wifi and then connects to an NTP server so I'm curious if either of those will cause it to stall.
The asyncio is for doing a loading-like animation with the lights, kinda like a loading bar, if that makes sense? So I'd like to manipulate lights while other things are happening, but let me know if that's not intended use.
I can share the repo if you'd like to take a look
tulip sleet
Got it. I have it set for multitasking flashing the lights while it connects to...
The asyncio is for doing a loading-like animation with the lights, kinda like a loading bar, if that makes sense? So I'd like to manipulate lights while other things are happening, but let me know if that's not intended use.
That sounds like a fine application for that.
proven garnet
Sweet! Thanks!
orchid basinBOT
orchid basinBOT
gilded cradle
<@!478616919721771019> Are there docs on how to use the new aliasing feature? I...
See https://github.com/adafruit/circuitpython-org/pull/805/files#diff-e1b6f641479fad99434e07f34f60e7ab692ab7078f4d5c1b9e710cc0f5c4c257R4. I just created a line with board_alias that links to the board_id of the cp build I'm trying to link use.
mortal mica
See https://github.com/adafruit/circuitpython-org/pull/805/files#diff-e1b6f64147...
Thanks That's an easy way of doing it!
gilded cradle
Thanks That's an easy way of doing it!
๐ That's what I was going for.
jaunty juniper
it does mean that the downloaded file will not have as specific a name as for other boards (if I understand the PR correctly), but that shouldn't be an issue, and it beats tinkering with that one list lost somewhere in the core
onyx hinge
darn, autodetecting SPI RAM size doesn't mean the build will also work on a no-spiram module ๐ I had hoped. let me try a few more things...
idle owl
@solar whale I tagged you for review as well, but here's a link to the Adafruit IO PR: https://github.com/adafruit/Adafruit_CircuitPython_AdafruitIO/pull/81
solar whale
<@!343570430042308609> I tagged you for review as well, but here's a link to the...
Thanks -- I won't be able to try it until later this afternoon.
idle owl
Thanks -- I won't be able to try it until later this afternoon.
No rush. Brent needs to take a look at it and I don't think he's even around today.
solar whale
It looks good to me!
idle owl
It looks good to me!
Thanks, Jerry!
tulip sleet
@slender iron did one of your CI changes fix this kind of error:
Run git describe --dirty --tags || git log --parents HEAD~4..
fatal: No tags can describe '4cdbc0d86067c706d9569786b60c27ff4f17d588'.
Try --always, or create some tags.
fatal: ambiguous argument 'HEAD~4..': unknown revision or path not in the working tree.
i can push an empty commit on that failed PR (maybe there are others) if it will fix this.
onyx hinge
curious. the RGB LED (neopixel) on esp32-s3-devkitm-1 (no psram) doesn't work reliably, it flickers random colors.
the first yellow blinks are right but then it just starts showing random colors
slender iron
@tulip sleet I was experimenting in my pending PR for issue label notifications
I haven't figured out a good fix
I kinda think it's a github bug
I feel like we want to move away for actions/checkout
in favor of a partial checkout
tulip sleet
I kinda think it's a github bug
this says it is due to a shallow clone: https://stackoverflow.com/questions/6445148/git-fatalno-tags-can-describe-sha1-number
you were trying this only on main, so maybe we should push one of your changes to 7.1.x, which is where the PR is
slender iron
ya, I know that. the python script is supposed to deepen it back to july 2021 though
it does deepen it when the change is pushed from a computer
I believe the issue is only after a github.com edit
tulip sleet
I pushed an empty commit to the PR. Maybe I should push an empty commit to 7.1.x?
slender iron
the branch is having trouble? not a pr?
tulip sleet
the pr is against 7.1.x
slender iron
what pr?
slender iron
hrm, you did all of those edits from your computer right?
tulip sleet
yes
tulip sleet
Failure is here: https://github.com/adafruit/circuitpython/actions/runs/1547752964
more precisely
that was last night's push
slender iron
this bit should fetch the branch history: https://github.com/adafruit/circuitpython/runs/4439291431?check_suite_focus=true#step:5:208
tulip sleet
yeah, it's failing in the next step
slender iron
right, because the branch history hasn't been fetched correctly
tulip sleet
so the --shallow-since="2021-07-01" may be buggy, Iguess
slender iron
ya, that's my thinking
I think what we really want is a clone with commit history
*partial clone
tulip sleet
we need the tree, but not all the blobs, like the blog post you suggested this morning. Then it will fetch the blobs it needs on demand.
slender iron
right. actions/checkout can only do shallow though
tulip sleet
I am trying to see which version of git is being used
slender iron
as-is
tulip sleet
Getting Git version info
Working directory is '/home/runner/work/circuitpython/circuitpython'
/usr/bin/git version
git version 2.34.1
that's the latest
hunh interesting: https://github.com/actions/checkout/issues/646
slender iron
sounds similar
onyx hinge
The ESP32S3-DevKitM-1 and ESP32S3-DevKitC-1 have very similar schematics, with the neopixel connected to the microcontroller via a zero-ohm resistor. no pull resistor is shown. On the DevKitC, the neopixel works fine. On the DevKitM, the neopixel flashes random colors, especially while the CPU is running. If I create a DigitalInOut of the pin, even if its direction is 'input', the flashing stops. Deinitting the pin makes the neopixel flicker again. So at the low level switching it between InputEn=0 and InputEn=1 makes a difference .. but why? And what should CircuitPython do about it?
slender iron
the github editing theory feels like a race condition to me
onyx hinge
(and why wouldn't the DevKitC be affected by .. whatever)
slender iron
it's sounds like floating flickering
tulip sleet
does neopixel_write turn the pin back to input when it's done?
slender iron
I think it's actually more power efficient for us to pull pins a particular direction when the input buffer is always on
onyx hinge
@tulip sleet no, neopixel_write doesn't touch the pin direction. but the supervisor deinitting it sets it to InputIn=0 OutputEn=0 etc
onyx hinge
I (395503) gpio: GPIO[48]| InputEn: 1| OutputEn: 0| OpenDrain: 0| Pullup: 0| Pulldown: 0| Intr:0
``` these sorts of messages are logged on the debug console by esp-idf when changing pin modes. If it's in the first state, the LED will flicker; if it's in the second state, the last set color stays.the led flickers because of floating line I'd assume. The input buffer may change the capacitance and therefore the floating
onyx hinge
and any attempt to probe it will also change the conditions under test ๐
slender iron
It may be best to pull the pin when not in use. I'm off for a walk now.
onyx hinge
a lot of adafruit parts don't seem to have pulls on the built in neopixel, indicating it's not typically a problem
tulip sleet
I'm also off for a walk
orchid basinBOT
On the Blinka devices there is a link near the bottom of their MD files:
The URL for that link appears to have the master branch coded into it. So that would need to get updated to main in each md file I think. i.e.
There are a few places in...
crimson ferry
@onyx hinge is the module difference irrelevant from a build perspective between the
ESP32-S3-DevKitM-1-N8 - ESP32-S3-MINI-1 Dev Board - 8 MB Flash (5311) and the
ESP32-S3-DevKitC-1-N8 - ESP32-S3-WROOM-1 Dev Board - 8MB Flash (5312)?
onyx hinge
@crimson ferry I don't know for sure! I had intended to get the other "C".
crimson ferry
I didn't see anything different in the datasheets
other than I think flash being inside the chip or can or not maybe, don't know if that affects flash speed or anything
solar whale
Is there a known issue with the adafruit feather esp32s2 and the latest tip of main -- Mine keeps crashing into HardFault -- OK with 7.1 beta
crimson ferry
@solar whale I'll D/L from S3 & try it out
solar whale
thanks -- I'll create an issue after I'v done more testing.
crimson ferry
I can't get it into bootloader with stacking headers without two hands and a tool, my finger is too big to squish the BOOT button between the Stemma QT and the stacking header o_O
20211207-45caaa8.uf2 crashes, seems maybe while scanning I2C (or something earlier in my code corrupts something accessed later)
solar whale
Consistent with what I am seeing
crimson ferry
boot.py seemed to run fine though, and code.py got a few hundred lines in before it manifested
solar whale
same
I am finishing up another test -- I can create an issue after I'm done-- or feel free to go ahead and create one.
crimson ferry
From minimal code in REPL, it seems to not like the onboard battery gauge (or I2C in general). I'll can file.
ember iris
`20211207-45caaa8.uf2` crashes, seems maybe while scanning I2C (or something ea...
I might be hitting that error! I've been trying to narrow it down for a while, thank you!
idle owl
@solar whale Thanks for testing!
I'll leave it for Brent to take a look at since he knows that lib better than I do.
solar whale
No problem -- glad to help.
idle owl
I don't think it could have been done any other way, but he said he'd take a look, and I appreciate that offer. ๐
ember iris
From minimal code in REPL, it seems to not like the onboard battery gauge (or I2...
I have extreme difficulty getting it to crash when reloading code from the computer. I hit the problem frequently from no power to power on, and it crashes when it gets to the i2c portion of the code, is this similar behavior? I am not powering by battery at any point
crimson ferry
I had to power cycle after each safe mode to get it to happen again, but it crashes even if I reload from REPL (after interrupting it before it hits the I2C code)
there could be some complicating factor
solar whale
It is crashing when I try to initialize an I2C sensor, scan was ok, but init crashed
Seems to be I2C in general
ember iris
I think mine crashes on this line: i2c = busio.I2C(board.SCL, board.SDA), I don't scan at all
crimson ferry
yeah, scan seems fine
crimson ferry
I think mine crashes on this line: `i2c = busio.I2C(board.SCL, board.SDA)`, I do...
oh, hmmm
solar whale
I was able to i2c = board.I2C()
crimson ferry
and ```py
import busio
i2c = busio.I2C(board.SCL, board.SDA)
I think*, grain of salt, otherwise I'd be commenting in the github issue. I made an outdoor weather station and fought to get the parts fit into the enclosure, so I just assumed I messed something up physically. Give me a bit and I'll bring it in and edit the code to test
Ok I might be hitting a different error:
i2c = busio.I2C(board.SCL, board.SDA)
Traceback (most recent call last):
File "code.py", line 389, in <module>
RuntimeError: No pull up found on SDA or SCL; check your wiring
Code done running.
Auto-reload is on. Simply save files over USB to run them or enter REPL to disable.
this hangs ```Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 7.2.0-alpha.0-43-g45caaa8e2 on 2021-12-07; Adafruit Feather ESP32S2 with ESP32S2
import board
import adafruit_si7021
si = adafruit_si7021.SI7021(board.I2C())
Apparently wrapping mine up in
# Start i2c and UART bus and ADC and connect to sensors
for i in range(5):
try:
i2c = busio.I2C(board.SCL, board.SDA)
break
except RuntimeError as e:
print(e)
solves my issue
solar whale
oops -- I forgot to turn on the I2C Power in when I got the hang -- may be misleading -- I did turn it on for the first crash when trying ot init the SI7021
solar whale
wierd -- bme280 works but si7021 does not ....
ember iris
Hmm. this is a weird set of issues. Does just trying again get around the issue, or is it a hard enough fault you can't event catch it with an except?
manic glacierBOT
Thanks to @jerryneedell for testing. The logic looks good to me, but I did not test.
Good grief, 13k file changes everytime they change something?!
solar whale
Hmm. this is a weird set of issues. Does just trying again get around the issue,...
I tried ```import board
import digitalio
import adafruit_si7021
ip = digitalio.DigitalInOut(board.I2C_POWER_INVERTED)
ip.switch_to_output()
ip.value = False
try:
si = adafruit_si7021.SI7021(board.I2C())
except Exception as e:
print(e)
print(si.temperature)
Hmm than unfortunately I'm probably not hitting the same issue. I might just have stressed my hardware/a wire too much when building it's enclosure
solar whale
What I2C sensor are you using?
ember iris
a bme280, pm2.5, and sgp40
ember iris
I tried ```import board
import digitalio
import adafruit_si7021
ip = digitalio....
For the power on the pin, I doubt this will be different but this is how I setup the i2c pin level:
x = DigitalInOut( board.D7)
x.direction = digitalio.Direction.OUTPUT
x.value = False
it should be functionally identical, but it's the only difference I see. And since the bme280 works for both of us I doubt this'll change anything
manic glacierBOT
solar whale
I have to go offline -- enough fun for today -- thanks @crimson ferry and @ember iris
crimson ferry
it was consistent earlier, but now I can't get it to crash, and I2C works even if the control pin is high
ember iris
I wonder if we hit three different i2c issues at about the same time
crimson ferry
there could be multiple issues. it does seem to require a read now, not just init
this is the point in debugging where my mind wanders to things like: is the board heating a factor?
ember iris
this is the point in debugging where my mind wanders to things like: is the boar...
haha Oh yes, I imagine there's a root cause, or a family of root causes. It was just a unique moment of "We have this problem and this is what it does" followed by "Huh mine is different than yours" that I really enjoyed
crimson ferry
yeah, what all seemed clear upfront became... murky
ember iris
Honestly that murkiness is one of the reasons this kind of work/project is enjoyable. Even if I have a mountain of other area's to address first
idle owl
@gilded cradle Am I missing something, or can you not do wav playback on MagTag? I feel like I'm missing something.
gilded cradle
<@!478616919721771019> Am I missing something, or can you not do wav playback on...
As far as I know you can't.
idle owl
As far as I know you can't.
Oh bummer. Ok. I need to reply to a forum issue then. Thanks
gilded cradle
Oh bummer. Ok. I need to reply to a forum issue then. Thanks
Unless Jeff added it and I didn't hear.
idle owl
@onyx hinge ^^ ?
onyx hinge
audiobusio should be on magtag https://circuitpython.readthedocs.io/en/latest/shared-bindings/audiobusio/index.html
idle owl
audiobusio should be on magtag https://circuitpython.readthedocs.io/en/latest/sh...
How do you do it with audiobusio? I can't figure it out...
onyx hinge
but that needs an i2s thingy
onyx hinge
maybe it's not there
idle owl
Ok
onyx hinge
no audiopwmio, no audioio
winter mortar
I'm experimenting with asyncio, and keep getting a traceback ending at core.py line 283, a call to sys.print_exception, which doesn't exist (in 7.1.0-beta.0 on pybadge). Should I update to beta.1, file an issue, something else...?
orchid basinBOT
idle owl
no audiopwmio, no audioio
Yeah that's what I ran into.
idle owl
Alright, thanks for the info. Reporting back to the forums.
stuck elbow
at least it's consistent
onyx hinge
at least it's consistent
yeah very sus
stuck elbow
fix it in software ;-)
have a lookup table of where the glitches happen and how much to correct fot them ;-)
idle owl
Hah!
stuck elbow
(please don't actually do that)
idle owl
Oh, now you tell me. I already started! ๐
orchid basinBOT
winter mortar
But how does it sound? You might be able to start a new genre
idle owl
glitchwav
stuck elbow
I can actually imagine how it sounds from that graph, and it makes my teeth hurt
orchid basinBOT
orchid basinBOT
onyx hinge
but if someone did want to look at it [audioout on esp32s2] here was my branch from the time: https://github.com/jepler/circuitpython/tree/esp32s2-dac-dma
tulip sleet
okay! does it need to be done in the 7.1.x branch too?
orchid basinBOT
lime trellis
has anyone reported sdcardio issues for 7.1 or 7.2 SAMD builds?
(on my pycubed boards) I'll have a board that happily mounts and operates an SD card for hours. However, if the card is ever written to from a PC (properly removed and reinserted from the unpowered board,) the board will **forever ** fail to find the SD card until 7.0 is flashed. At which point it can successfully mount the SD -> UF2 update to 7.2 -> can successfully mount SD again. I don't have any SD-capable feathers to test on
tulip sleet
has anyone reported `sdcardio` issues for 7.1 or 7.2 SAMD builds?
(on my pycu...
this is interesting. There have been SD card issues mentioned here which have been hard to reproduce: https://github.com/adafruit/circuitpython/issues/5600 and https://github.com/adafruit/circuitpython/issues/5644
lime trellis
this is interesting. There have been SD card issues mentioned here which have be...
#5600 could be related to what I'm seeing. My current theory is something to do with removing power to a 7.2 board (regardless if it's been ejected properly). I'll investigate more and let you know
lime trellis
this is interesting. There have been SD card issues mentioned here which have be...
following up: board with CP7.2 can successfully operate an SD card immediately after flashing until power is removed from the board (unplug usb) for the first time. Any attempt thereafter (tried 20 various unplugs/replugs/waiting) will be unsuccessfully until 7.0 is flashed. Then it works immediately and consistently between unplugs. If 7.2 is flashed, the behavior cycle is repeated (it'll find the SD until unplug, etc...)
Any thoughts?? tried multiple sd cards
sharp jacinth
Question about development of the circuitpython audio library. It doesn't seem that there had been much development on it recently. Does it still not work?
manic glacierBOT
crimson ferry
ninja: error: unknown target 'esp-idf/esp_system/__ldgen_output_sections.ld' ...is this something to do with moving to idf 4.4?
analog bridge
`ninja: error: unknown target 'esp-idf/esp_system/__ldgen_output_sections.ld'` ....
that is related to idf 4.4
is your esp-idf submodule updated?
crimson ferry
not sure, I did make without -j and and got a different error: supervisor/port.c:81:10: fatal error: esp32/spiram.h: No such file or directory
I re-installed: esp-idf/install.sh
analog bridge
the submodule should be at ddc44956bf commit, what do you get with git status in the esp-idf folder
crimson ferry
HEAD detached at e493a4c30e
analog bridge
it is not updated, you need to re-init it
git submodule deinit -f esp-idf
git submodule sync && git submodule update --init esp-idf
run the above in circuitpython/ports/espressif forlder
crimson ferry
lots of downloading, but this looks promising... thanks!
crimson ferry
it's got other problems now, but it's after 1am, so... another time.
lone sandalBOT
manic glacierBOT
tulip sleet
Question about development of the circuitpython audio library. It doesn't seem t...
We fixed a bunch of audio problems on various ports: I'm not sure what you mean.
onyx hinge
I just don't see any relevant sdcard changes, for this problem that exists in "7.2" (I assume tip of main) but not in "7.0"... behavior different after power cycling points to problems getting the SD card to enter SPI mode, which sticks until power is cycled, but 7.0 and current tip of main both do exactly the same thing, code hasn't been touched...
and as usual I can't reproduce this; the role of a computer writing to the SD card is also puzzling
tulip sleet
do you think this has anything to do with the space-saving done in oofatfs? Was that post 7.0.0?
onyx hinge
no, for a couple of reasons ..
I am assuming that the error is "no SD card", which is way before FAT comes into play
with this most recent report I don't think the person actually said
but second, the space savings was only in mkfs (format) so we don't even use that code on SD cards
tulip sleet
that's what I thought. I was wondering if it had to do with checking the dirty bit or where it was formatted
onyx hinge
the very first thing the SDCard constructor does is try to put the card into SPI mode and get a response that indicates it's entered SPI mode. If this fails, the exception says "no SD card"
tulip sleet
there was a PR about fixing SPI mode never reset or something. If only you could reproduce in any way, I would say do a bisect
manic glacierBOT
onyx hinge
the change that there has been in SD cards is the 'CMD25' stuff but that only comes into play after you've gotten to the point of writing a card
orchid basinBOT
onyx hinge
I wonder if a card reset mid-CMD25 could be in a weird state .. but that's not what users are reporting
tulip sleet
@onyx hinge I have not followed this all that closely. What is the simplest setup and test program to try to reproduce? I have lotsa cards.
onyx hinge
@tulip sleet pygamer is probably one of the easiest pieces of HW
tulip sleet
i have one
manic glacierBOT
onyx hinge
import board
import busio
import sdcardio
import storage
spi = busio.SPI(board.SD_SCK, MOSI=board.SD_MOSI, MISO=board.SD_MISO)
cs = board.SD_CS
sdcard = sdcardio.SDCard(spi, cs)
vfs = storage.VfsFat(sdcard)
storage.mount(vfs, "/sd")```leaving the airlift case aside, what users seem to be encountering is that the SDCard constructor throws 'no SD card'
I've seen 'no SD card' when attempting to hot plug SD cards; in that case it goes away when I hard reset. with a card inserted at power on, the above program can initialize the sd card as often as desired over soft resets
tulip sleet
i will try on GC first because it has full SWD connector
onyx hinge
I appreciate you taking a look!
tulip sleet
at what version did these problems seem to start?
onyx hinge
https://github.com/adafruit/circuitpython/issues/5600 says after 7.0.0
reported against CircuitPython 7.1.0-beta.0 and later says "Given that it worked fine with identical hardware with 7.0 I don't think it was an issue with the SDCard. Downgrading made it work again."
the airlift problem is reported as affecting 6.3 as well, so it's not a new problem whatever is going on there
onyx hinge
has anyone reported `sdcardio` issues for 7.1 or 7.2 SAMD builds?
(on my pycu...
this reports that the problem is between 7.0 and "7.2".
airlift problem is https://github.com/adafruit/circuitpython/issues/5644
tulip sleet
I will test with tip of main and if I can reproduce, will try bisect
(i did a bisect yesterday between some 7.0.0 alphas. That was a bit painful due to the MPy merges. Lots of git bisect skip.
onyx hinge
hmm indeed
tulip sleet
@onyx hinge OK, I have a bunch of MicroCenter cards: I have 32GB's that aren't found; one 16GB that is found and one that isn't. I also have a SanDisk 16GB that works.
So I will try 7.0.0 and see what happens, and then bisect if there's a difference
yes 7.0.0 works; great, will start a bisect
idle owl
@tidal kiln Please ping when you're around, I'd like you to proof something for me. (At least I think you'd be good to proof it. I could be wrong.)
tidal kiln
@idle owl yo
idle owl
I wrote a template for setting up the IDE for RP2040 Arduino. Do you mind checking it? I'll DM you the URL since it's not a real preview URL.
tidal kiln
sure
idle owl
Template previews look a bit odd.
Also just noticed the "SOME OTHER TEXT HERE" I forgot about at the beginning ๐
idle owl
@tidal kiln Fixed the intro.
I feel like the plan is for a second template showing how to load Blink. But Limor is busy, and I forgot to ask until today, so that will wait.
tulip sleet
@onyx hinge bisect done: It was [f22299d2b374a8150ec63d9338175aa51b593dd6] sdcardio: Use CMD25 across multiple writeblocks() calls. Sometimes power-cycling the board (as opposed to just resetting) made a difference, so I tested with power-cycling.
onyx hinge
<@!426200547024961539> bisect done: It was `[f22299d2b374a8150ec63d9338175aa51b5...
@tulip sleet that change can result in "no SD card" exception?
tulip sleet
yes, I used your test program above. It either threw an exception or didn't
(I added a print("mounted") at the end.)
I think I saw cases where I didn't power-cycle the board and it worked, so I made sure to power-cycle
onyx hinge
that's wild, that code shouldn't even execute until well after "no SD card" can occur. obviously I'm overlooking something. and have no cards here that repro the problem.
onyx hinge
we can revert CMD25 in the short term, if it helps 7.1.0 be releaseable. We do want the speed sometime soon though, it's a big speed increase when writing jpegs or gifs from camera
init_card calls cmd which can now return early / fail for some new reasons
if exit_cmd25 fails, or if wait_for_ready fails. the first shouldn't happen (can't be in_cmd25) but the second could, in which case it doesn't actually write the needed command...
tulip sleet
yes, confirmed that's the commit
can I turn on some logging for you to see when it fails?
tulip sleet
I will turn on the DEBUG_PRINT for now
onyx hinge
here's a somewhat speculative fix change ```patch
diff --git a/shared-module/sdcardio/SDCard.c b/shared-module/sdcardio/SDCard.c
index 9eb32d926..649858fcb 100644
--- a/shared-module/sdcardio/SDCard.c
+++ b/shared-module/sdcardio/SDCard.c
@@ -230,11 +230,18 @@ STATIC const compressed_string_t *init_card(sdcardio_sdcard_obj_t *self) {
common_hal_digitalio_digitalinout_set_value(&self->cs, false);
- assert(!self->in_cmd25);
- self->in_cmd25 = false; // should be false already
- // CMD0: init card: should return _R1_IDLE_STATE (allow 5 attempts)
{
bool reached_idle_state = false;
for (int i = 0; i < 5; i++) {
-
if (cmd(self, 0, 0, NULL, 0, true, true) == R1_IDLE_STATE) {
-
// do not call cmd with wait=true, because that will return -
// prematurely if the idle state is not reached. we can't depend on -
// this when the card is not yet in SPI mode -
(void) wait_for_ready(self); -
if (cmd(self, 0, 0, NULL, 0, true, false) == R1_IDLE_STATE) { reached_idle_state = true; break; }
timber mango
Was having some issues with GPS timekeeping, and thought it was an issue in the adafruit_gps module.. Turns out it was related to int() rounding. Can someone else confirm that this is an issue, or is it just a float inaccuracy?:
Adafruit CircuitPython 7.1.0-beta.0-75-g932131b4f on 2021-11-18; Adafruit Feather Bluefruit Sense with nRF52840
int(184105.000)
184104
tulip sleet
Was having some issues with GPS timekeeping, and thought it was an issue in the ...
it is a float issue: there are only about 5 1/2 digits of precision
onyx hinge
do you want round() instead of int() ?
timber mango
Well the adafruit_gps module does int()
tulip sleet
is that really an HHMMSS or something
timber mango
with .000 numbers
Yeah, the GPS featherwing returns time as HHMMSS.000
so the 3 .000 seems to put off the seconds
tulip sleet
it should be parsed as a string, not as a number
onyx hinge
or int("184105.000".split(".", 1)[0]) awfulness
slender iron
or just slices to pick out the digits
timber mango
Actually through it would strip off the irrelevant decimals on its own, but int(184105.000) and int(184105.00) produce two different results
onyx hinge
>>> float(184105) - float("184105.000")
0.0625
>>> float(184105) - float("184105")
0.0
``` well that's interesting! float parsing code is ad-hoc and doesn't always parse to the nearest floating point numberthere are issues about this in the MPy repo
onyx hinge
it's a deliberate trade off because fully accurate parsing is slow and requires a lot of code
tulip sleet
@timber mango If the GPS parsing code could be improved, please file an issue, thanks, maybe even a PR, if it's clear how to fix it
onyx hinge
yeah, not to me, with code space an issue.
timber mango
I'm a bit unsure why the int() is actually in the code.. time_struct seems to handle float ok..?
tulip sleet
it may have been transcribed from some Arduino code
timber mango
Will do some experimenting
onyx hinge
ah if you have long ints you can do better, but this is not true on all our boards by far
https://github.com/micropython/micropython/pull/5832 is probably what dan was thinking of
slender iron
the parsing code really shouldn't use float as an intermediary
tulip sleet
the parsing code really shouldn't use float as an intermediary
the GPS parsing code, you mean? I agree completely
slender iron
yup
onyx hinge
given the limitations of floats, if this little bit can be done in the world of ints you're almost certain to be better off.
timber mango
NMEA specification seems to always use HHMMSS.sss .. So if the format can be assumed to be the same for all chips, the . could just be stripped out, and math done on resulting int
tulip sleet
@onyx hinge sd test program with the debug prints on:
>>> import sd
cmd 0 [00] arg= 0 [00000000] len=0 data wait
cmd 0 [00] arg= 0 [00000000] len=0 data wait
cmd 0 [00] arg= 0 [00000000] len=0 data wait
cmd 0 [00] arg= 0 [00000000] len=0 data wait
cmd 0 [00] arg= 0 [00000000] len=0 data wait
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "sd.py", line 9, in <module>
OSError: no SD card
seems to me that _parse_data should parse into time.struct directly
tulip sleet
NMEA specification seems to always use HHMMSS.sss .. So if the format can be ass...
does it need to be broken down to HH MM SS and sss anyway?
then do int on each of those
using ints makes sense in Arduino, with poor string handling, but in Python, string handling is easy
timber mango
seems to me that _parse_data should parse into time.struct directly
Probably better then trying to do math on it all
slender iron
run time for me. CI is catching up on all my merges
lime trellis
I just don't see any relevant sdcard changes, for this problem that exists in "7...
chiming back in about observed sd card issues. See my message here #circuitpython-dev message
๐ Has something to do with power cycling a 7.2 build. Everything will work upon flashing until power is removed. Maybe pin states are being handled differently upon POR?
tulip sleet
or the card itself is in a somewhat different state
@onyx hinge I can get you some SPI traces, but I need to wire up another board with an SD breakout, I guess, to make the lines accessible
Maybe I should mail you one of these "bad" cards
lime trellis
@tulip sleet have you tried flashing 7.0 and then talking to one of your "bad" cards?
tulip sleet
yes, it works, iI bisected between 7.0.0 and tip and main. The hinge commit is one that adds CMD25 functionality
lime trellis
yes, it works, iI bisected between 7.0.0 and tip and main. The hinge commit is o...
ah ha, ok progress ๐
tulip sleet
which supposedly should not matter, but there it is. Oh yes, this is definitely progress. I can also imagine that other cards may act slightly differently, and give yet other symptoms, such as yours, for exmaple
read back up a ways
lime trellis
read back up a ways
ah I see. man sd cards over spi can be tedious. too many variables
tulip sleet
I don't think they test it extensively, because SDIO is used mostly for speed reasons
sharp jacinth
We fixed a bunch of audio problems on various ports: I'm not sure what you mean.
Let me re find the library. You're right there's a ton of audio stuff
orchid basinBOT
onyx hinge
here's a somewhat speculative ~~fix~~ change ```patch
diff --git a/shared-module...
if you still have some time to spare on this, can you try with this speculative fix? I think the change I introduced during card init is that a failed 'wait for ready' causes the whole 'cmd' operation to fail... that's not how it used to be.
sharp jacinth
Let me re find the library. You're right there's a ton of audio stuff
https://github.com/adafruit/Adafruit_CircuitPython_VS1053 Here's the library. My goal is to have audio output from a headphone jack with similar quality to the audio from a pi
I'm not liking the random jitters and stops that I can't control when using the pi
timber mango
<@456226577798135808> If the GPS parsing code could be improved, please file an ...
@tulip sleet Tried my first at this, and will really appreciate feedback: https://github.com/adafruit/Adafruit_CircuitPython_GPS/pull/76
tulip sleet
@onyx hinge somehow i missed your speculative early this afternoon. I will try it
onyx hinge
Thanks! I didn't @ you the first time.
sharp jacinth
Should I be in #help-with-circuitpython? This seemed more dev related
tulip sleet
Should I be in <#537365702651150357>? This seemed more dev related
are complaining about the pops and clicks?
sharp jacinth
No thats not CPs fault. I'm just trying to discern if there's a way to replicate what I'm doing with a pi, with any cp mcu libs
tulip sleet
are you complaining about VS1053 or about native audio
sharp jacinth
I'm just trying to understand if it's possible currently to use a CP lib on a CP mcu to output audio from a headphone jack. Not intending to complain at all, just wanting to know the lay of the land
I can always use arduino I just love cp
tulip sleet
you can play wav files and mp3 files (not all boards support mp3). The audio will be output as PWM audio (lower quality), as DAC audio (high quality), or as I2S audio (digital output)
this is without a VS1053, right?
in all cases you need some kind of amplifier (except for a very few boards that have a headphone jack)
sharp jacinth
this is without a VS1053, right?
Yeah I think since the lib is giving folks trouble trying to get audio to work with Python I'd have to go without that chip. So I'd need some kind of amplifier IC that feeds into a headphone plug?
My plan was to buy a circuitbrains m4 board which I believe from my reading can handle mp3s and also include a sd card reader for the music. The VS1053 was attractive because it has all the hardware I need.
I can bodge in an SD card reader by cribbing off of adafruit boards though
tulip sleet
yes, you don't need the VS1053 board. There are several boards with SD cards. Then you connect an external amplifier. Or the PyGamer has both
do you want to drive speakers or headphones?
sharp jacinth
Speakers but the line goes to an external amp
tulip sleet
then you could use the DAC output directly, because it is 3.3V max
as a line output to the external amp
sharp jacinth
Ooh ok. What can I google to make sure the Amp is sufficient? And can the DAC be connected to a headphone plug?
Thanks!
tulip sleet
what is the amp/
i have used a little battery-powered Radio Shack amp with a speaker, and I have used breakout amps from the Adafruit store to drive a speaker, both with good results
sharp jacinth
Let me get the model.
tulip sleet
@onyx hinge your patch works on the GC4 with the "bad" cards. I am still having trouble with the sd breakout on the MM4. I wonder if there is a marginal timing problem there
i still see couldn't determine SD card version on the MM4
onyx hinge
@tulip sleet that first is good news. you might try lowering the SPI clock rate, it defaults to 8MHz while the old adafruit_sdcard defaulted to 1_320_000 or something
sharp jacinth
It is a "70V/25V PA Amplifier"
tulip sleet
and it has a line-level input? Does it have a mic input too?
the specs are talking about the output
tulip sleet
<@!329766224093249548> that first is good news. you might try lowering the SPI...
The MM4 with breakout seems quite flaky. I can't even get 7.0.0 to work with some cards
but the fix seems good for on-board sockets. I could try it on something else like a pygamer as well
i have to do something non-work for a bit
onyx hinge
thank you so much for testing!
tulip sleet
I tried lowering the clock rate to 1MHz and even 100kHz, unsucessfully with the breakout
sharp jacinth
It does, 2 of them