#circuitpython-dev
manic glacierBOT
tulip sleet
@onyx hinge note just above
onyx hinge
@tulip sleet do you have any idea about how to make GPIO23 have a known value (I'm still trying to figure out if high or low) during deep sleep on rp2040, without user code intervention? I am not familiar with that code.
tulip sleet
<@329766224093249548> do you have any idea about how to make GPIO23 have a known...
before going into deep sleep, reset_pins() (?not sure of the name) is called, which eventually calls something port-specific. In Espressif there is a special-case routine espressif_board_reset_pin_number() which can be added to board.c which will do whatever special-case pin resetting you want to (a pin number is passed in and it returns true or false saying whether it handled it or not). The same mechanism could be added to raspberrypi
but I don't know if pin state is preserved in deep sleep on RP2040, have to look at the datasheet for that.
in Espressif, this is all quite complicated: there is a "hold pin state" mechanism for sleeping
I would think they would have handled this in the pico-sdk already for the pico w ??
onyx hinge
I think it must not be handled, otherwise it wouldn't use 10mA more
and yes I also don't know what rp2040 deep sleep can do, I should go browse a datasheet
tulip sleet
also in Espressif, some pins are pulled down on reset, and some are pulled up. There is not uniformity, so board designers have to be careful.
onyx hinge
gpio_put(WL_REG_ON, false); // off OK setting that GPIO false turns off the regulator
onyx hinge
ugh I'm really not set up to test deep sleep
tulip sleet
ugh I'm really not set up to test deep sleep
you mean current testing?
onyx hinge
yes current testing in deep sleep specifically
but just having a convenient way to make the device not just enter fake deep sleep too
brazen hatch
wasn't there the need for tinyuf2 to be updated?
tulip sleet
if you have a reasonable ammeter you can just break the red wire in a USB cable and put it in series, and also cut the data lines so it doesn't do fake deep sleep
onyx hinge
and if we are only entering fake deep sleep then we'd need the wifi to not actually be reset ugh
tulip sleet
if you have a some kind of USB breakout that could help
@onyx hinge worth getting a PPK2: in stock at DigiKey: https://www.digikey.com/en/products/detail/nordic-semiconductor-asa/NRF-PPK2/13557476
onyx hinge
I have one, I just haven't learned to use it
tulip sleet
oh, then you are all set. download nrfconnect appimage. load the PPK2 app, set it to source mode, plug in usb to data/power jack plug in header, connect it to ground/+5 on the picow (be careful to disconnect picow USB), set it to 5v in the app with slider, move GUI switch underneath voltage slider to on, press start
I had trouble with version 3.12.0 of nrfconnect recently, try 3.11.1 first
ah, they fixed the problem: https://github.com/NordicSemiconductor/pc-nrfconnect-ppk/releases/tag/v3.5.2
onyx hinge
the VOUT and GND of the 4-pin header go to the pico?
tulip sleet
yes, can you power pico via some pads, or could you use a USB breakout that you can plug the A end of the cable into?
onyx hinge
I don't think I have such a USB breakout but I put a https://www.adafruit.com/product/4449 in my next cart
tulip sleet
the PPK2 can act as a series ammeter with external power (e.g. a battery), or it can supply 1A or so itself (subject to current limitations of the USB port it is plugged into)
i labeled the wires on the 4-pin breakout on mine, since it's kind of confusing otherwise
crimson ferry
I had wondered in a prior issue / PR about whether we might want to expose Pico W GPIO23 / WL_ON to user code. User may want to turn wifi off but not be in a sleep mode, to run non-wifi code but still reduce power. This may not be feasible and mess something else up, but if it could work, it would be cool.
Similar thought about whether user code should (could?) be able to de-init wifi or similar on espressif (for power and / or getting back to a cleaner state).
onyx hinge
I saw that
@crimson ferry yeah the user can't just power cycle the wifi module willy nilly
I mean, I could be wrong and you're welcome to prototype/try it but my gut feel is that it is not going to make anything better
jaunty juniper
wasn't there the need for tinyuf2 to be updated?
but I think it's the edgest of cases: you must have used dualbank while the drive is not extended to switch the firmware to OTA1, and then reformatted in extended mode, which would then make the extended partition use OTA0, which would be overwritten by tinyuf2 
it would be nice to have tinyuf2 updated before 8.0 goes stable nonetheless
onyx hinge
Left is the last 2 half-second LED blinks, then the mash at the right is deep sleep (no time alarm, no pin alarms)
@tulip sleet assuming this graph means anything it's still clearly visibly doing "something" during deep sleep, but what "it" is I can't tell you
last is zoomed way in on the sleep part of the trace
tulip sleet
if the cyw43 is not really sleeping maybe it is looping and sort of sleeping, and doing something periodically
like turning radio on and off?
onyx hinge
nothing is being done to put the cyw43 into any sleep mode at all
at least not outside the sdk, it's vaguely possible (but I don't think so) that the sdk could do something
jaunty juniper
@tulip sleet I'm not clear on what default we settled on for the OTA1/extended drive ? (new unformatted board)
tulip sleet
@onyx hinge I will approve and merge https://github.com/adafruit/circuitpython/pull/7048 (errno fix) if that's ok with you. Looks good to me but did not test. That is the last thing I see for beta.2 after I merge utime()
onyx hinge
<@426200547024961539> I will approve and merge <https://github.com/adafruit/circ...
yes it did look good to me as well, I also didn't test
brazen hatch
I will test rn if you guys want me to
onyx hinge
Just forcing WL_REG_ON to false makes a big difference
tulip sleet
I will test rn if you guys want me to
if you can test the invocation of that error, that would be great. thanks!
brazen hatch
alright brb
tulip sleet
it hasn't built yet though
I can make a build for you quick
unless you are already going to do it
onyx hinge
@tulip sleet ```diff
diff --git a/ports/raspberrypi/common-hal/alarm/init.c b/ports/raspberrypi/common-hal/alarm/init.c
index 31fa58aab3..d94049dcd2 100644
--- a/ports/raspberrypi/common-hal/alarm/init.c
+++ b/ports/raspberrypi/common-hal/alarm/init.c
@@ -204,6 +204,11 @@ void common_hal_alarm_set_deep_sleep_alarms(size_t n_alarms, const mp_obj_t *ala
void NORETURN common_hal_alarm_enter_deep_sleep(void) {
bool timealarm_set = alarm_time_timealarm_is_set();
+#if CIRCUITPY_CYW43
+#define WL_REG_ON 23
- gpio_set_dir(WL_REG_ON, GPIO_OUT);
- gpio_put(WL_REG_ON, false);
+#endif
// If there's a timealarm, just enter a very deep light sleep
if (timealarm_set) {
// Prune the clock for sleep
you could put an ancillary routine in cyw43 that would get called from here, just to move the responsibility. i.e. a non common_hal_ helper routine
brazen hatch
I can make a build for you quick
2.1ghz pi400 overclock go brr
brazen hatch
2.1ghz pi400 overclock go brr
I did reply to the wrong msg, I am doing it myself
tulip sleet
🙂 I guessed that's what you meant
brazen hatch
Just flashed it, kinda unfamiliar with the code, just running it adapting the ssid and passwd
Seems like I needed to load adafruit_minimqtt
did load from latest bundle
but got AttributeError: 'MQTT' object has no attribute 'check_msg'
seems like the script he gave us isn't the same as the one he ran?
or am I too sleepy?
Fresh from my toasty keyboard
I did swap it to _wait_for_msg and now it's on the while True
brazen hatch
I am testing the first script of the issue? Right?
onyx hinge
@tulip sleet we need to draw the line somewhere for the next beta but this is affecting liz and it'd be nice if it was fixed.
tulip sleet
Fine with me to include. If I release tomorrow vs tonight, not a big difference
jaunty juniper
ah, adding .env to my ESP32S3 Box on latest sent it into a boot loop, I'll look into it later and open an issue if it doesn't match another. It doesn't do it on a FeatherS3 (latest)
onyx hinge
@jaunty juniper ouch! that would be good to get to the root of, but if you had longer keys that could sure be related
otherwise let us know what you figure out about it. You can probably redact the file with any alphanumeric just changed to 'x' if it's sensitive.
brazen hatch
I am not quite sure how to raise that error, it's 1 am, my brain is shutting down, imma look at it tomorrow if you guys haven't done it already, sorry.
onyx hinge
of course @brazen hatch -- see you around
jaunty juniper
<@199811382793863168> ouch! that would be good to get to the root of, but if you...
it's the same .env I have been using that has worked for testing Web Workflow so far on other boards, I'll build the latest and test on different boards in a bit to see if it's only the box or some bigger regression - maybe tomorrow morning if I fall asleep in the mean time
onyx hinge
OK, you can open an issue anytime
tulip sleet
@onyx hinge I will merge the errno one anyway
tulip sleet
@onyx hinge should I wait until morning for shakeouts? All the PR's so far have seemed good to go.
long dotenv is only one left
onyx hinge
@tulip sleet I am not likely to do any testing before morning
tulip sleet
I can test the .env easily
onyx hinge
@tulip sleet I did toss it on my rp2040 for a quick test. ```>>> len(os.getenv('sz9'))
512
maybe I will do a wifi smoke test on pico w before I do a release; if that's good, will release
tulip sleet
good night, thanks for all the fixes; the release notes are done except for the recent merges
orchid basinBOT
manic glacierBOT
This reduces power consumption during true deep sleep.
In my measurements with ppk2 and a program that irrevocably entered deep sleep (no time alarm or pin alarm), power usage as measured on a ppk2 decreased from ~10mA to ~1mA.
Closes: #7047 (or at least improves on the status quo)
ppk graph, before (right hand side represents deep sleep):
zoomed in on the right:
midnight ember
Adafruit Feather S2 & S3 WiFi is rock solid right now. Thank you for all the hard work getting that one hammered out.
tulip sleet
@onyx hinge I am looking at the "why did raspberrypi changes build espressif boards" issue, but are you already looking at that?
onyx hinge
@tulip sleet I think the short version is, the script is wrong. I was working on making the script more easily runnable locally, and faster by parallelizing the code to find build flags. I haven't started fixing any bugs
but I think the short version is, if it sees ports/PORT/common-hal/FOO modified, then it rebuilds anything anything with module FOO enabled, not just those within PORT
tulip sleet
I tried to run it locally. There is an f missing on the f-string you added for the set-output changes, but mainly when I give a common-hal file, it just sits. If I listed raspberrypi/audio_dma.c in CHANGED_FILES, it works. Did you rewrite it or you're saying the original script is not right.
I think just removing common-hal here might do it:
module_pattern = re.compile(
r"^(ports/[^/]+/common-hal|shared-bindings|shared-module)/([^/]+)/"
)
I cannot think of a case where you want to recompile other ports if common-hal has changed without shared-bindings changing.
onyx hinge
it 'just sits' because it's doing lengthy work of getting all board compile flags
removing the first alternative of that first group I think will fix the immediate problem
tulip sleet
so that would be parallelized in shared_bindings_matrix.py?
onyx hinge
because it'll just build "all in port PORT" instead of "all with module FOO"
yes I want to do is parallelize get_settings_from_makefile
tulip sleet
ok, 1.5 minutes to do that, I see. Are you working on that now or should I submit a simple PR for this other thing?
onyx hinge
I think go ahead and submit the simple change.
onyx hinge
@tulip sleet I think I'm close to PR'ing my changes so maybe not merge yours for a bit?
Using files list on commandline
Adding docs/boards to build based on changed files
Building docs: False
Would set GitHub actions output build-doc to 'False'
Building boards:
raspberry_pi_pico_w
Would set GitHub actions output boards-aarch to '[]'
Would set GitHub actions output boards-arm to '["raspberry_pi_pico_w"]'
Would set GitHub actions output boards-riscv to '[]'
Would set GitHub actions output boards-espressif to '[]'
oops we crossed. I added a revert of #7055 to #7057
tulip sleet
oops we crossed. I added a revert of #7055 to #7057
but I don't understand why ports/common-hal should trigger a recompile of all boards that match that module. Or does it no longer do that due to new line 136? (and falls into the full check)
a common-hal change or a bindings change should only affect that port
onyx hinge
@tulip sleet what's a path that you're wondering about? I can show you what it results in with my change.
$ GITHUB_OUTPUT=/dev/null ./tools/ci_set_matrix.py ports/raspberrypi/common-hal/ssl/SSLSocket.c
Using files list on commandline
Adding docs/boards to build based on changed files
Building docs: False
Building boards:
raspberry_pi_pico_w``` isn't this the case you're wondering about?sorry, i am testing it by hand, and it's working right. I didn't see the logic change you made in the makefile-based determination logic, which fixed what it did before.
hurrah for @functools.cache !!
onyx hinge
it's down at new lines 152ff that "specific to port & module" is handled
boo for accidental mentions
tulip sleet
yes, I missed those changes the first time I looked, ok, is good!!
onyx hinge
OK, thanks for checking -- it's easy to overlook something
I think it https://github.com/adafruit/circuitpython/issues/6922 closes this issue
tulip sleet
i think so too, can link it
onyx hinge
yes I checked and that now just builds stm boards
orchid basinBOT
crimson ferry
@onyx hinge can we not turn the wifi module off for light sleep and fake deep sleep, as long as there's a way to turn it back on? I'm think something like asyncio with time alarms, where the wifi module is off while waiting for tasks.
tulip sleet
<@426200547024961539> can we not turn the wifi module off for light sleep and fa...
light sleep is defined to keep things in the same state, such as the current wifi state. So turning it off would change the state. We had the same thing with Espressif boards; they have a "light sleep" which turns off wifi. We though we wanted that but realized it wasn't that useful. In retrospect we might never have bothered with "light sleep" in the alarm module
digital shoreBOT
onyx hinge
@crimson ferry My understanding is that until a pico-sdk bug is fixed we CANNOT power cycle & re-initialize the wifi coprocessor without re-initializing CircuitPython. https://github.com/raspberrypi/pico-sdk/issues/980 "The second call to cyw43_arch_init() hangs." If that is a wrong or incomplete understanding (or once they fix it), then we have more options available.
tulip sleet
has anyone used a QT Py ESP32-C3 with CircuitPython recently? beta.2 is just bootlooping on mine(repeats ESP-ROM:esp32c3-api1-20210207)
blissful pollen
has anyone used a QT Py ESP32-C3 with CircuitPython recently? beta.2 is just boo...
Tried it on the SeeedStudio ESP32-C3 and seeing the same thing
Build:Feb 7 2021
rst:0x3 (RTC_SW_SYS_RST),boot:0x8 (SPI_FAST_FLASH_BOOT)
Saved PC:0x4038201e
SPIWP:0xee
mode:DIO, clock div:2
load:0x3fcd5810,len:0x1350
load:0x403cc710,len:0x8f0
load:0x403ce710,len:0x2c18
entry 0x403cc710
thanks dan and jeff, i’ll look at that issue more… maybe what I’m really after, when that is fixed, for pico and espresif, is a Wi-Fi deinit in the user API
tulip sleet
Tried it on the SeeedStudio ESP32-C3 and seeing the same thing
```ESP-ROM:esp32c...
beta.1 is OK, so something..., probably the ESP-IDF update 😦
I'll bisect
slender pivot
Hi, is https://learn.adafruit.com/how-to-add-a-new-board-to-the-circuitpython-org-website available on GitHub to make a PR to?
tulip sleet
Hi, is https://learn.adafruit.com/how-to-add-a-new-board-to-the-circuitpython-or...
The guide is not on github, but you can click the "Feedback" link on the left sidebar to make a suggestion
slender pivot
Ok, thanks for your prompt response
What about https://circuitpython.org? Is that just index.html?
tidal kiln
website repo is here:
https://github.com/adafruit/circuitpython-org
slender pivot
Thanks, is the landing page https://circuitpython.org/? itself just index.html?
If I have minor suggestions to make should the PR touch index.html directly? I am a bit of a git noob which is why I ask
tidal kiln
you can open an issue as a general way to make a suggestion
and discussion there can help answer that
slender pivot
Ok, my change is just some typos on the landing page such as "built in interpreter" --> "built-in interpreter", and things like that
I can make the changes and open PR directly if it is allowed
Meaning no need to open issue for discussion as changes are minor. If necessary, we can discuss under the open PR?
Does this seem feasible or do you have a better suggestion? Open to ideas
tidal kiln
yah, for minor things like spelling, can just make a pr
slender pivot
Great, thanks for your help @tidal kiln
onyx hinge
"-fno-branch-probability" saves ~400 bytes on trinket m0. I wonder how bad it is for performance.
onyx hinge
just can't catch a break with esp-idf can we 😦
tulip sleet
just can't catch a break with esp-idf can we 😦
it is pretty frustrating. I'll research their issues, and maybe I have to do a bisect inside esp-idf
well, that fix is in the latest v4.4 I am using, so it doesn't appear to be that
tulip sleet
@onyx hinge could we video about asyncio yield?
onyx hinge
@tulip sleet sure! will 10-15 minutes from now be OK? I'm trying to see if I can make the doc build go
tulip sleet
sure, I will eat some lunch
onyx hinge
@tulip sleet ping me when you're ready
onyx hinge
Amelia channel OK?
midnight ember
@tulip sleet your post about the beta 2 release in the circuit Python forum should probably be pinned and replace the currently pinned beta 1.
midnight ember
@marble notch Murk this is the developer channel. Recommend you switch to channel https://discord.com/channels/327254708534116352/537365702651150357 for code support.
marble notch
midnight ember
😉 no worries
onyx hinge
@tulip sleet Adafruit_CircuitPython_asyncio doesn't work with python3.9 because select.poll isn't compatible: ``` File "/home/jepler/src/circuitpython/frozen/Adafruit_CircuitPython_asyncio/asyncio/core.py", line 169, in wait_io_event
for s, ev in self.poller.ipoll(dt):
AttributeError: 'select.poll' object has no attribute 'ipoll'
this does not affect running the tests, hmm? too bad, maybe should look at that later as well
onyx hinge
no, the core tests run Python3 with python3's asyncio as reference, not with 'our' asyncio
tulip sleet
<@329766224093249548> your post about the beta 2 release in the circuit Python f...
thanks, I forgot to finish that off
manic glacierBOT
onyx hinge
@proven garnet if you're around can you please take a look at this run? The error isn't related to my change in an obvious way, but the most recent builds on main didn't have a problem, so .. I'm puzzled.
A clone of CircuitPython is down in ./build_deps/ and it has a submodule named nvm.toml. Hmmmm, it IS a recent change that nvm.toml is moved, and maybe it's moved to a place that it's being populated by the CI?
This build script is different than e.g., requests, fairly radically, not sure which one is 'more current'
er I take back the thing I was saying about a recent change, I was thinking of nina-fw
tulip sleet
@onyx hinge some related links I had mentioned:
https://www.youtube.com/watch?v=Y4Gt3Xjd7G8
https://vorpus.org/blog/notes-on-structured-concurrency-or-go-statement-considered-harmful/
https://trio.readthedocs.io/en/stable/tutorial.html
random junco
I think Tekktrik is moving this weekend
onyx hinge
oh, could be. I think the CI is green but it really merits a look by someone who is more conversant with it.
sharp jacinth
How hard would it be to replace the build of circuitpython on a rp2040 stamp with an identical build that just appears as a different USB name on the PC?
jaunty juniper
How hard would it be to replace the build of circuitpython on a rp2040 stamp wit...
we have supervisor.set_usb_identification() in CP8
sharp jacinth
ooooh
jaunty juniper
timber mango
@prime flower hey brother, can you accept my friend request? id like to ask you about something if thats alright?
sharp jacinth
Most folks around here don't do DMs
timber mango
fair enough, i just need to talk to him about something privately.
prime flower
<@311339734339289090> hey brother, can you accept my friend request? id like to ...
Hi - I don’t Direct Message and will not accept friend requests. If it’s a matter of community - please take it up with the moderators or within #help-with-community . If it’s a code issue - you can ask it here or in the forums.
timber mango
Hi - I don’t Direct Message and will not accept friend requests. If it’s a matte...
it's neither, it's regarding something completely different.
thats why i was hoping to talk to you privately.
cunning crypt
it's neither, it's regarding something completely different.
Good evening,
It's generally considered rude to DM someone without their permission, and nobody has any requirement to accept your DMs. It's particularly fishy when someone joins a server and, on the same day, asks a specific person to accept their messages. Especially when that's the only messages you've made on the sever.
As this is the #circuitpython-dev channel, it's generally presumed that anything brought up in this channel, hence the comments about code or community.
If you have a more personal, urgent matter, you can contact one of us moderators and we can then contact the person in question. This allows us to make sure nobody's getting spammed, while also allowing more information to be passed without it being public. Additionally, we can set up non-public spaces so people don't have to accept permanent friend requests.
So, in short:
If you have something code related, it should be in a channel.
If you have something community related, you can talk to a moderator or #help-with-community
If you have something personal, you can talk with a moderator and we can go from there.
timber mango
Good evening,
It's generally considered rude to DM someone without their permis...
yeah i apologize, hence why i asked if it was appropriate to speak with him privately, i really did not mean to disrespect anyone, i was just wondering if we could chat privately.
cunning crypt
And yet so far you have not made any indications of what you want to talk about. As I said, you can DM one of the mods (Oh look, I'm conveniently available), and we can pass on the information and see if the other party is willing to talk about it.
You have to understand that we are here to protect our community members - and unwanted DMs are so frequent, and so outright dangerous, that people can't just accept them.
And at the end of the day, if someone says "No" you have to accept that.
prime flower
@timber mango nope! I don’t accept DMs. You can talk the matter over with @cunning crypt and they can pass it to me. Thanks!
timber mango
that's totally understandable, and i apologize if i annoyed you in anyway.
@cunning crypt may i dm you?
cunning crypt
<@85813260359565312> may i dm you?
Go ahead.
proven garnet
<@135415127879974912> if you're around can you please take a look at this run? T...
I can take a look! @random junco is right in that I'm moving but I would do anything not to touch another box for the rest of the evening 
onyx hinge
@proven garnet OK I think the immediate fire is put out but either this library is an oddball and it would be good to bring it in line; or this is common and it'd be good to correctly deal with the problem.
proven garnet
Yeah, just caught up, interesting that it's just happening now, I'm about to make a junk PR to a library and see what happens.
I tagged you on it just for an easy link if you need
Yup, looks like this is gonna be an issue.
Your fix looked good though. I think adabot can fix this pretty quickly. Even better is turning this into that composite action!
I can get an adabot patch tonight, probably within an hour if you (or anyone) wants to review; I can run it after. Also happy to wait a day since it is the evening, but this sounds way better than moving furniture.
No worries either way, I'll have it ready tonight though.
I'll also prioritize that composite action since this would be awesome for this sort of thing. I think I was just about ready to PR it too.
orchid basinBOT
proven garnet
That's weird I could have sworn the CI failed when I submitted that PR a little while ago, but now it shows as passing.
@onyx hinge no need to respond now but curious if anything changed in the last little bit, since the library CI is passing now for the test PR.
spiral elk
I think I'm gonna close the ESP32-CAM PR and open a new one once I've tried my code on updated esp-idf. Its sat for a long time and needs directory/device-name renames anyhow.
I'm suspecting the issues I've been having with the camera modules are due to wrong settings in sdkconfig
Been trying to build the example code from espressif but I think my idf install is somehow hosed.
foggy lake
Is it possible to clear the UART buffer?
jaunty juniper
Is it possible to clear the UART buffer?
there's UART.reset_input_buffer(), is that what you're looking for ?
https://docs.circuitpython.org/en/latest/shared-bindings/busio/index.html#busio.UART.reset_input_buffer
foggy lake
@jaunty juniper looks like what I'm looking for. Thanks
onyx hinge
<@426200547024961539> no need to respond now but curious if anything changed in ...
weird, I don't know what's up.
ornate breach
hey @tulip sleet this issue right here, should it be closed since there was a PR merged referencing it? https://github.com/adafruit/circuitpython/issues/5734
tulip sleet
hey <@329766224093249548> this issue right here, should it be closed since there...
Yes, thanks!
ornate breach
np! I was looking to see if there was a good Hacktoberfest issue to work on and noticed it 🙂
manic glacierBOT
tulip sleet
@analog bridge are you testing beta.2 on your C3 board? This is my highest priority thing, but if you think you have a lead on it, go ahead. I have to be afk for a while. Thanks.
ornate breach
Yes, thanks!
I think this one can be closed as well: https://github.com/adafruit/circuitpython/issues/3515
If I recall right, we moved away from devices.h usage.
analog bridge
<@747673663419187281> are you testing beta.2 on your C3 board? This is my highes...
I am looking into it but so far no breakthrough in sight. I suspect its something to with flash (specifically boards with 4MB).
Can you test on a 4MB s2/s3 board?
tulip sleet
will do
ornate breach
tulip sleet
I think this one can be closed as well: https://github.com/adafruit/circuitpytho...
Closed that too. This is the best kind of issue 🙂 . Are you @ornate breach on GitHub or something else? I want to credit you.
ornate breach
yeah skerr92
i'm finding a few with merged PRs like this too lol https://github.com/adafruit/circuitpython/issues/1270
tulip sleet
yeah skerr92
Post comments pointing to what you think closes these issue in the issues, and I can close them from there. That will get the xref directly in. I will see all these here or in my email and will close them.
starting with the tilegrid above -- thanks!
tulip sleet
Can you test on a 4MB `s2/s3` board?
4mB flash, like an S2 wrover?
tulip sleet
I am looking into it but so far no breakthrough in sight. I suspect its somethin...
Metro ESP32-S2, which uses an ESP32-S2-WROVER (4MB flash, 2MB PSRAM) is fine with beta.2
(i was testing the wifi scanning problem on this board.)
analog bridge
@tulip sleet Can you do a custom build on it with
-CIRCUITPY_ESP_FLASH_FREQ = 40m
+CIRCUITPY_ESP_FLASH_FREQ = 80m
tulip sleet
<@329766224093249548> Can you do a custom build on it with
```patch
-CIRCUITPY_...
Didn't make a difference, and also I remembered Scott told me that param is just used to construct the shell flash command, not used in the build itself that I can see
in ports/espressif/Makefile
there are two patches in our version of esp-idf, both having to do with sdkconfig params. it would be interesting to back those out (and compensate if necessarY). One for JTAG serial, and one for the workaround you mentioned the other day
(and then deleted)
analog bridge
there are two patches in our version of esp-idf, both having to do with sdkconfi...
for config ESP_PHY_ENABLE_USB, I was thinking of PRing the following upstream:
config ESP_PHY_ENABLE_USB
bool "Enable USB when phy init"
- depends on USB_OTG_SUPPORTED || ESP_CONSOLE_USB_SERIAL_JTAG || ESP_CONSOLE_SECONDARY_USB_SERIAL_JTAG
+ depends on USB_OTG_SUPPORTED || SOC_USB_SERIAL_JTAG_SUPPORTED
default y if IDF_TARGET_ESP32C3 || IDF_TARGET_ESP32S3
default n
help
one for the workaround you mentioned the other day
(and then deleted)
I forgot, which one is this?
tulip sleet
> one for the workaround you mentioned the other day
> (and then deleted)
I forg...
(ESP32_ECO3_CACHE_LOCK_FIX)
These are the only diffs from upstream:
analog bridge
Oh okay, I deleted that comment as I found your commit message:
This was necessary to get CircuitPython not to boot-loop when starting up and then disabling the SPI flash cache during filesystem_init().
right, there seemed to be something depending on that even without SPI RAM being present
analog bridge
@tulip sleet does adafruit_feather_huzzah32 work with 8.0-beta2?
tulip sleet
i will test
analog bridge
quoting mwalimu on github:
I able to get it to work on M5-ATOM by rewriting Huzzah32. I also got it to work on M5-STAMP-c3u by rewriting QTYPY-esp32-c3.
They worked rather well with REPL and Workflow. However, since the last update none of them work.
https://github.com/adafruit/circuitpython/issues/7060#issuecomment-1280014229
tulip sleet
@analog bridge HUZZAH32 does not work - hangs after "Serial console Setup". reconfirmed it works on beta.1. testing qt py esp32 in a minute. Also will test an S2 board with no PSRAM
QT Py ESP32 PICO works with beta.2
analog bridge
Neradoc tested Feather ESP32 V2 and it also works, both QT Py ESP32 PICO and Feather ESP32 V2 have 8MB flash
meanwhile huzzah32 and the esp32c3 boards tested so far boards have 4MB flash
tulip sleet
ESP32S3 with 8MB flash and no PSRAM works
analog bridge
I did a custom build for microS2 with 4MB flash and it also works, note: the chip is still 16MB, I defined 4MB in firmware to restrict the size
tulip sleet
by using the 4MB .csv?
analog bridge
by using the 4MB .csv?
yes
jaunty juniper
feather ESP32-S2 has 4MB (plus 2 PSRAM) and seems fine
analog bridge
so, the conclusion I am reaching is that this issue is present on esp32/esp32c3 boards with 4MB flash
tulip sleet
Both Feather S3 8mB/none and 4MB/2MB work with beta.2
is your c3 board different in some way than the qt py c3? I have a qt py c3, and I also have an Espressif ESP32.C3-DevKitC-02, not tested yet
analog bridge
microC3 uses ESP32C3-MINI-1 module with the ESP32-C3FN4 chip, I believe this chip is the same as on QT PY C3
tulip sleet
that espressif c3 doesn't work either, has 4MB flash
analog bridge
I did a DEBUG=1 build of the microC3 and I get the following: (not sure if these were present earlier)
SHA-256 comparison failed:
Calculated: ccb0d00bac7e84e1d90a12e4f75f4ab6c5f7e71bb209afd5819c4c9557a6db71
Expected: c9cf160580940ec7801c73b16423824e72ad12055c732e83ce66332240af42a7
Attempting to boot anyway...
we could compare the generated build-*/esp-idf/sdkconfig for these boards between beta.1 and beta.2 and see if there is a difference
I have to go to the store but will be back in an hour or so.
proven garnet
weird, I don't know what's up.
Gotchya, it's still worth addressing, but if it isn't occurring right now then maybe I can finish up that composite action stuff and get that out with the fix. Two birds, one PR.
ornate breach
Curious, do we have Matter support for the ESP32 in circuitpython?
tulip sleet
Curious, do we have Matter support for the ESP32 in circuitpython?
Not intentionally, no
ornate breach
there appears to be python matter libs
I'm just curious if it's worthwhile looking into
blissful pollen
there appears to be python matter libs
Something I had vaguely thought about it but not high on my list to look into
ornate breach
the only real support we would get with the nRF7 chips is being a wifi coprocessor since that's the intended end use.
I am inclined to think that adding some basic Matter support might be worth while.
tulip sleet
the only real support we would get with the nRF7 chips is being a wifi coprocess...
what is the protocol used to talk to the chips, then, is it Matter? I am surprised the uses would be only as a coprocessor, since they are many applications for integrated wifi on a single chip, as we see from ESP32xx.
ornate breach
it's listed as a wifi coprocessor for the public release documentation
"The nRF7002 is a companion IC, providing seamless Wi-Fi connectivity and Wi-Fi-based locationing (SSID sniffing of local Wi-Fi hubs). It is designed to be used alongside Nordic’s existing nRF52® and nRF53® Series Bluetooth Systems-on-Chip (SoCs)"
tulip sleet
as for the nRF SDK, we don't use it. We use only the nrfx lower-level library, which calls the SoftDevice. The nRF SDK has a lot of convenience stuff, but we didn't use it.
aha
ornate breach
it works essentially like the Pico W WiFi coprocessor
Matter is one of the primary protocols they are pushing on it because it's the major IoT standard major tech is pushing.
tulip sleet
https://www.nordicsemi.com/Products/Development-software/nrf-connect-sdk -- says it is based on zephyr. we don't use this at all
we looked at it; zephyr had issues about dynamic assignment of pins to functions. We brought that up, they are moving in the direction of making that possible, but I don't know where it is. Zephyr is complicated and very Linux "inspired"
.
ornate breach
that's understandable. I imagine building without Zephyr might be an undertaking
I imagine like all things, it isn't impossible; just a bit more trivial.
tulip sleet
it's not possible to omit zephyr -- it provides the BLE stack, the RTOS, etc.
ornate breach
hmm, I would imagine that you would have the nRF7 loaded with it's base firmware and then you would communicate over a serial interface like SPI
tulip sleet
(afk to eat dinner)
ornate breach
(for when you get back)
You would essentially have to have a firmware image like this https://github.com/adafruit/Adafruit_nRF52_Bootloader built, programmed onto the nRF7002 and then develop our own commands to talk back and forth. lots of moving pieces.
tulip sleet
(for when you get back)
You would essentially have to have a firmware image lik...
yes, so we need the details of the protocol exchanges with the co-processor
ornate breach
I'm still waiting to see if Nordic will be able to get me a nRF7 DK which would be helpful for making some initial progress. I'm hoping I can get one by the end of the year
looks like the DK at least uses QSPI to communicate
tulip sleet
the physical protocol i'm not worried about, it's the messages
ornate breach
yeah, they also appear to use some control pins
HOST IRQ = P0.23
COEX_REQ = P0.28
COEX_STATUS0 = P0.30
COEX_STATUS1 = P0.29
COEX_GRANT = P0.24
so that will have to be considered as well
onyx hinge
busy day in here!
tulip sleet
HCI is kinda standard for Bluetooth, but I don't know of such a standard for wifi (e.g., the Pico W CYW43 uses a bespoke protocol, I think).
ornate breach
i guess we'll have to wait to see what new info they release. there doesn't appear to be much about wifi other than it's shipping with Matter spport
using Matter seems like it might be okay. though that might actually get implemented on the nRF7 itself.
tulip sleet
using Matter seems like it might be okay. though that might actually get impleme...
my guess is that Matter is in layers above that. e.g. MQTT support is going to be above the building blocks supplied by the nRF7xxx
ornate breach
my guess is that Matter is in layers above that. e.g. MQTT support is going to b...
Gotcha, I’m still trying get grips on all this. 🙂
kind river
I'd guess that Matter runs on the Zephyr TCP/IP stack that has drivers for the nRF7 WiFi and the internal 802.15.4/thread stack
ornate breach
I'd guess that Matter runs on the Zephyr TCP/IP stack that has drivers for the n...
Looks like you are correct
It runs on top of the zephyr application on the chip it’s on
Looks like there would need to be some support in the Adafruit Bluefruit app for the QR code portion of setup
Or maybe an Adafruit Matter app for matter setup that is independent
manic glacierBOT
analog bridge
@tulip sleet does the huzzah32 hang with storage extension off?
tulip sleet
<@329766224093249548> does the huzzah32 hang with storage extension off?
I did not try that yet.
I'm guessing it will not, based on the bisect commit.
i can do a quick test
tulip sleet
<@329766224093249548> does the huzzah32 hang with storage extension off?
No, it does not hang. I tested on the tip of main with CIRCUITPY_STORAGE_EXTEND = 0
analog bridge
I don't have a huzzah32 or any other 4MB esp32 available for testing, this is a very specific issue.
tulip sleet
do you have some idea of what it might be? I can test for you. My first thing would be to look at the partition .csv files to see if they're unusual compared with the others.
could possibly be related to PSRAM as opposed to 4MB/8MB? Though I don't see an obvious reason.
analog bridge
yesterday I tested 4MB partition.csv on microS2 and storage extension worked fine
could USB play a role?
tulip sleet
yes, it's peculiar to the plain ESP32, so I wonder what's different. Could be a low-level thing due to hw otr ESP-IDF internal differences
could turn off USB or microS2, though it's a bit hard to test then (have to enable UART console or debug prints). I haven't yet built a debug version to get a backtrace or anything. Finished the bisect late last night here.
i am going out shopping in a bit, will be back in an hour or two.
analog bridge
I'll test with CIRCUITPY_USB=0 and 4MB partitions.csv on microS2
tulip sleet
I'll test with `CIRCUITPY_USB=0` and 4MB partitions.csv on `microS2`
It is not going to find a partition that matches app/ota_1, so the data in _partition[1] is not useful.
but there is no test for CIRCUITPY_STORAGE_EXTEND = 0 in your initial commit, so I'm confused why turning it off would make a difference.
ok, I am going out. ttyl!
analog bridge
@tulip sleet I think you found the bug 👍
<=4MB no-uf2 partition tables don't have ota partitions
also in Makefile the condition to use no-uf2 partition table seems wrong:
-UF2_BOOTLOADER ?= $(if $(filter $(IDF_TARGET),esp32s2 esp32s3),1)
+UF2_BOOTLOADER ?= $(CIRCUITPY_USB)
I am doing this now
It already works async, I just need to tweak the cases a bit more.
onyx hinge
awesome thanks @brazen hatch !
brazen hatch
yea something is very wrong with the return codes of cyw43_wifi_link_status
analog bridge
@tulip sleet can you test #7073
proven garnet
@onyx hinge I finalized the composite actions, hoping we can roll them out to a test library this week! @idle owl, that's the repo I transferred to you yesterday.
tulip sleet
<@329766224093249548> can you test #7073
sure, I will. We might discuss whether to shrink CIRCUITPY or just disable dualbank in 4MB as well. Some people might complain about the reduction in size of CIRCUITPY. I will discuss that in an internal mtg today and we can talk about it in the regular Monday public meeting as well.
Should
_partition[1] = esp_partition_find_first(ESP_PARTITION_TYPE_APP,
ESP_PARTITION_SUBTYPE_APP_OTA_1,
NULL);
check that the partition was not found and act accordingly?
i mean, we already made a decision not to have OTA on those boards, though maybe that was before dualbank was a possibility.
hmm, well, the whole point of extended storage is to make a larger CIRCUITPY possible; I missed the point there.
so ok to have an OTA partition that can be reused
@analog bridge how would you feel about the default, at least on some boards, to be that the OTA partition is part of CIRCUITPY instead of by-default reserved for dualbank? I know Limor's feeling is that use of dualbank is rare, so if we can make it available, but not the default, then those who want it can set up their boards that way, and most people, who don't need it, can benefit from a larger CIRCUITPY initially?
analog bridge
<@747673663419187281> how would you feel about the default, at least on some boa...
storage is extended by default and dualbank is off
tulip sleet
ah great, ok, I did not realize that
so my only remaining q is about the _partition[1] above
analog bridge
so my only remaining q is about the `_partition[1]` above
esp-idf says the following about return value of esp_ota_get_next_update_partition
NULL result indicates invalid OTA data partition, or that no eligible OTA app slot partition was found.
the only reason to have "no eligible OTA app slot" is incorrect partition table as we witnessed
otadata partition may be corrupted though
tulip sleet
maybe safer to check for _partitition[1] being NULL in supervisor_flash_read_blocks() before trying to call esp_partition_read() ?
I should just put this in a review
I'll do that, mention it in a review
blissful pollen
I'll do that, mention it in a review
Is this the one that was breaking the C3? If you want me to retest the one I have just ping me here or on GH and I'll try to do it later
tulip sleet
Is this the one that was breaking the C3? If you want me to retest the one I hav...
no, this is ESP32 only. The ESP32-C3 issue is different, and has to do with a toolchain update
random junco
<@&356864093652516868> We are back to our normal time of 2pm EDT / 11am PDT for the CircuitPython Weekly Meeting, which is 3 hours from now! I look forward to seeing your hug reports and status updates. Doc link is here: https://docs.google.com/document/d/1WzDqCFGtObTzJMjDfAjHY0ujFAD_jQ3Q_zowSrH404o/edit?usp=sharing - see you then!
tulip sleet
both ESP32 and ESP32-C3 were not working, but for diff reasons
analog bridge
no, this is ESP32 only. The ESP32-C3 issue is different, and has to do with a to...
I believe if esp32c3 didn't have the other issue, this one would be present on c3 as well
blissful pollen
Ah okay. Well still holds when a fix is found for that still feel free to ping me about it
tulip sleet
i wonder if the compilation change is actually causing this error to manifest in a different way.
so go ahead and test c3 on this new PR too
tulip sleet
@analog bridge ehh, I was making comments on the commit, not the new PR (which doesn't even have that file, so I can't put in review comments about it).
analog bridge
<@747673663419187281> ehh, I was making comments on the commit, not the new PR (...
lets discuss it separately?
tulip sleet
(could not understand why I was not seeing "Start a Review") 🙂
analog bridge
@tulip sleet what should be done if it is null? we cannot raise an error in that part of code I believe? should we revert to less storage but than how will the user know?
tulip sleet
just use partition 0, or maybe go into safe mode with a new safe mode condition code?
blissful pollen
Quick question that probably is too short for in the weeds: Has anyone ever done any work on optimizing the display on CP? Just noticed the Arduino eyeball project runs at 50ish FPS with full refreshes and the equivalent CP code is about 5 FPS.
(I could add this to in the weeds but I'm missing today)
tulip sleet
safe mode errors are sometimes "here's this bad thing that happened". or it could be an assert which would require debugging but at least it would be clear what the error is
tulip sleet
see supervisor/shared/safe_mode.h
it could also check for _partition[0] being NULL. Could do that right after the find_partition calls and then just go into safe mode. have to prevent recursive safe_mode happenings.
it's ok by me to open an issue to fix this as opposed to fixing this now
i am just thinking about someone who somehow gets circuipython loaded (maybe by .bin) on a chip with a wonky partition table
analog bridge
it's ok by me to open an issue to fix this as opposed to fixing this now
yes, lets discuss it in a separate issue and not block #7073 on this
also no bonus fixes with #7073... tested on microC3 and #7060 is still present
analog bridge
i am just thinking about someone who somehow gets circuipython loaded (maybe by ...
in dualbank, there is
update_partition = esp_ota_get_next_update_partition(NULL);
assert(update_partition != NULL);
my opinion would be to check partition-table during build instead of runtime
tulip sleet
my opinion would be to check partition-table during build instead of runtime
that's a good idea
solar whale
<@&356864093652516868> We are back to our normal time of 2pm EDT / 11am PDT for ...
Aren’t the meetings normally at 2PM Eastern?
stuck elbow
time is an illusion
random junco
@tulip sleet: Can you speak to the core in today's meeting?
tulip sleet
sure
random junco
thanks!
proven garnet
1pm my time, need more coffee!
I feel this in my bones, not even noon and I'm already ready for my third 
random junco
I'm trying to work on typing in the adafruit_avrprog.py library. How can I tell the type for spi_bus and rst_pin in this code:
_spi = None
_rst = None
def init(self, spi_bus, rst_pin) -> None:
"""
Initialize the programmer with an SPI port that will be used to
communicate with the chip. Make sure your SPI supports 'write_readinto'
Also pass in a reset pin that will be used to get into programming mode
"""
self._spi = spi_bus
self._rst = DigitalInOut(rst_pin)
self._rst.direction = Direction.OUTPUT
self._rst.value = True
Outside of just knowing and based on the code/doc not sure. They are busio.SPI and microcontroller.Pin. Some of the other libraries, like sensor ones, probably have similar items to use as a template
random junco
Outside of just knowing and based on the code/doc not sure. They are busio.SPI a...
Thanks
blissful pollen
If you can't find a sample let me know. I've seen it around but don't have one handy offhand
random junco
I will if I can't find it, appreciate it
proven garnet
I'm trying to work on typing in the `adafruit_avrprog.py` library. How can I te...
Variables like spi_bus and i2c_bus, especially in the __init__() methods of drivers, are almost (if not completely) always there respective types from busio.
It's more difficult to tell about pin arguments, but a good tell is how it gets used: if it has .switch_to_input() or .value used, it's probably a digitalio.DigitalInOut. If it gets turned into a digitalio.DigitalInOut like it does here, it's probably a microcontroller.Pin.
random junco
It's more difficult to tell about pin arguments, but a good tell is how it gets ...
thanks for the explanation, I'm starting to figure it out I think
proven garnet
No problem! There are a couple of exceptions like drivers for GPIO extenders that define their own DigitalInOut class, but otherwise you can bank on the above.
tulip sleet
@analog bridge I was waiting for the artifacts but one build is stuck doing a network fetch. Also a few boards are failing due to size considerations. I will build HUZZAH32 from your repo and test it.
analog bridge
<@747673663419187281> I was waiting for the artifacts but one build is stuck doi...
the overflow is because sdkconfig-esp32.defaults is setting its own LOG LEVEL
#7052 addressed the log level issue in -opt and -debug sdkconfigs
tulip sleet
i saw something about that last night, but didn't have a chance to check it yet.
analog bridge
I'll push a commit to fix it
tulip sleet
<@329766224093249548> can you test #7073
It comes up on HUZZAH32! Also tested creating a file and then hard-rebooting, and the file is preserved, so the filesystem is being reused instead of being re-created.
idle owl
<@426200547024961539> I finalized the composite actions, hoping we can roll them...
Please bug me today to accept that, or you'll have to transfer it again.
thorny jay
<@&356864093652516868> We are back to our normal time of 2pm EDT / 11am PDT for ...
Thank you Paul, I'll skip (as in no notes) this week meeting as I did not do much except testing that flip clock during previous meeting. And if I find my Feather S3+TFT, then I will try during this meeting the PR that @lone axle made 5 days ago.
manic glacierBOT
I'm happy to test for you, but I've never got artifacts from a PR build before. So I will appreciate your instructions.
Click on the "Details" link on a board build below.
Then click on the Summary link:
Then scroll on the Summary page to see the "Artifacts se...
onyx hinge
thanks for that ^^ dan
tulip sleet
I should probably put it in a guide
modern wing
Good afternoon, happily lurking.
onyx hinge
That message I think is about the label being newly assigned, it's not a count of labeled issues.
modern wing
welcome back 🙂
idle owl
Thank you 😊
onyx hinge
@deʃhipu for implementing PNG support for adafruit_imageload
💯
stuck elbow
@lone axle ouch, sorry, I was meaning to get back to it, but life happened
blissful pollen
I’m out for an appointment so can’t edit the notes doc for in the weeds but I had a general question if anyone has done any performance on displayio speeds. More if it’s worth me poking at.
idle owl
I’m out for an appointment so can’t edit the notes doc for in the weeds but I ha...
I'll add it
lone axle
No worries. I have a followup one in for the last few bits. I am super excited to have the functionality in displayio 🎉
errant grail
I never settle for being a single point of failure. I usually bring on multiple points.
onyx hinge
@analog bridge if you're motivated, i'm sure that PR could use someone "kicking the tires" as it were -- I only wrote some basic tests and ran on the host, didn't even put it on a device yet.
stuck elbow
No worries. I have a followup one in for the last few bits. I am super excited t...
that code still misses handling for filters, which is required for the non-indexed images (grayscale and rgb)
onyx hinge
as long as there's a clear and reliable way to create compatible files it's a great step forward
analog bridge
<@747673663419187281> if you're motivated, i'm sure that PR could use someone "k...
sure, chained exceptions were on my radar earlier when traceback module was implemented, super excited for this, thanks! 👍
onyx hinge
particularly, I'm expecting that there are other places that the code for chaining needs to be added -- or, alternately, a better way to centralize the chaining.
lone axle
that code still misses handling for filters, which is required for the non-index...
We can always gain implementations of that functionality in the future. Having indexed pngs with easy support for the transparency is really nice to have 😄.
onyx hinge
This 'echo' thing is very weird for today's computers but made a lot of sense in the days when you would hook 16 VT-100 terminals to your one UNIX server. And Linux dutifully copied this way of working with serial connections.
tulip sleet
or teletypes!
onyx hinge
green-screen VT terminals were a thing during my university time, so that's the model I think back to
tulip sleet
interesting we don't really use terminals anymore, or we use them the other way around
delayed echoing can be maddening if you don't see your typing right away
stuck elbow
it's fine with vim, drives you crazy with emacs
idle owl
"If I do say so myself" 😂
Jeff came up with CowBell, if his hubris did not make that obvious.
gilded cradle
I wondered
idle owl
(It's a name completely worth the hubris. So well done.)
stuck elbow
CowBell™ (now with more hubris)
modern wing
CowBell+, the new streaming subscription service 🐄
stuck elbow
CowBell+, the new streaming subscription service 🐄
live from Switzerland
modern wing
Kattni: "I had covid, and I'm working on sharing...." oh no please, don't share covid "...the experience on my blog." whew. 🌡️ 🙂
stuck elbow
Nurgle is the god of generosity
onyx hinge
Happy Birthday @minor plume !
minor plume
Happy Birthday <@353373147199832074> !
Thanks!
minor plume
it still has more RAM than the pico though
That’s true. Not by as much as one might expect, though… 😉
random junco
@onyx hinge - I agree on the Lucy Peanuts desk, first thing that came to me too
tulip sleet
i had an IBM Z50 for a while, that I could run some kind of UNIX on. But I sold it on to the next collector
turbid radish
@lone axle did you get your newsletter weekly report in?
onyx hinge
stuck elbow
I did look at some things briefly
one surprising thing was that enabling async dma transfers didn't help much
tulip sleet
the base pixel-writing routine does a lot of work
stuck elbow
it's mostly cpu-bound
midnight ember
didn't intend on missing the meeting but things happen :/
idle owl
@random junco
gilded cradle
I think the last time optimizations were done in CP 4.1.0
idle owl
Tim was talking
analog bridge
Thanks!
idle owl
@lone axle I would type it out. I think Paul couldn't hear you.
lone axle
I have tested a few scripts that try to show which things are fast and which things are slower. But never gone further than that to figure out where improvements can be made. @blissful pollen
blissful pollen
On my phone while waiting for an appointment. Just comparing the arduino eyes project to CP it’s about 1/10th. Just more wondering if someone had hit a hard limit already
lone axle
<@382939733107408897> did you get your newsletter weekly report in?
I will check right now and get it in if not.
ember iris
Of late sometimes discord doesn't pipe the audio for everyone sometimes. I've found refreshing my page get's the audio to work if I notice someone isn't making audio but others are responding
onyx hinge
@blissful pollen I'd love to chat with you about this sometime that's convenient for us both
stuck elbow
one thing that could potentially help with displayio would be running the rendering in the background on the second core on devices that have two cores
ember iris
cheers all
onyx hinge
👋
random junco
really sorry @lone axle
lone axle
really sorry <@382939733107408897>
No worries! I am quite suspicious of my tab crashing immediately before that (it cut off while the topic was being introduced) being related some how. Maybe it doesn't send out "person re-joined" events correctly in all cases internally when a client comes back.
I think you heard me before when I read my hug and status so it was only after that crash that you didn't.
onyx hinge
If I was working on optimizing displayio I'd look at how e.g., displayio_tilegrid_fill_area has to make multiple conditional function calls for each pixel. And figure out how to change things so that instead of common_hal_displayio_bitmap_get_pixel being called (say) 32 times for an area, a new call could do 32 pixels at once. This would eliminate a lot of function call/return, the type-check if statements, etc. The same would be done for the "pixel shader" step, it would work with a group of pixels in one call, not a call per pixel... Probably it should become a call "per row, up to some maximum pixel width". But that's just based on gut instincts, not on actual profiling or anything.
blissful pollen
<@228322543507865601> I'd love to chat with you about this sometime that's conve...
Sure thing. My thought was to look at it more closer to / just after Halloween but really anytime I’m on usually after 4ish or weekends works
analog bridge
Any suggestions regarding the coproc API? PR #6902
import coproc
with open("Program_A.bin", "rb") as f:
a = coproc.Coproc(buffer=f.read(), address=0x500007fc, length=1024)
with open("Program_B.bin", "rb") as f:
b = coproc.Coproc(buffer=f.read(), address=0x500007fc, length=1024)
coproc.run(a)
coproc.halt(a)
coproc.run(b)
# OR
a.run()
a.halt()
b.run()
@onyx hinge ^
onyx hinge
@tulip sleet @idle owl do we have enough uncategorized issues in the core that we should do a bug triage session this week? I am guessing that the 8.0.0 list has had all the stuff booted from it that can reasonbly be deferred. Or, I can unilaterally go label the unlabeled issues if nobody's done it yet.
blissful pollen
Any suggestions regarding the `coproc` API? PR #6902
```py
import coproc
with op...
Forgot to comment on the PR but I did look at it and in general I liked it myself though didn’t get to spend time trying it or thinking of other uses.
idle owl
<@329766224093249548> <@330227457296957440> do we have enough uncategorized issu...
I'm up for it if you both think it would be helpful.
analog bridge
Forgot to comment on the PR but I did look at it and in general I liked it mysel...
thanks for testing it, I'll be pushing a more advanced version of it upstream with access to shared memory and pin conflict resolution
tulip sleet
<@329766224093249548> <@330227457296957440> do we have enough uncategorized issu...
I was just starting to label the unmilestoned issues. I am just behind on that. When it gets lower we can go through it again.
blissful pollen
thanks for testing it, I'll be pushing a more advanced version of it upstream wi...
Cool shared memory / python objects would be great for that.
proven garnet
I think I asked before but asking again since there's still time for new contributors - do we want to share our participation in Hacktoberfest on their Discord server? Do we think we have the capacity for an influx if we publicize we're looking for help?
Sorry meant to bring it up in the meeting
onyx hinge
@analog bridge I haven't really thought about it since writing that feedback on the PR last month. I think it would be helpful to me to talk about a couple of 'real' uses of this functionality because right now it feels very abstract and instead of being helpful I immediately jumped to "here are some potential sources of trouble" which doesn't really help you move the PR forward
Again, with pio, we had a number of interesting and well motivated examples that were in the pico examples and in their manuals .. are any of those potentially applicable to the risc v co-processor? Like, a bit banged i2c or neopixel, say.
analog bridge
<@747673663419187281> I haven't really thought about it since writing that feedb...
wait for me to PR my local changes upstream... it contains solutions to the issues you raised
ember iris
I think I asked before but asking again since there's still time for new contrib...
Do you know of a project that did share on the Hacktoberfest Discord, so we can see what volume of PRs they received?
onyx hinge
oh you have updates to the PR incoming? that's excellent news!
proven garnet
Do you know of a project that did share on the Hacktoberfest Discord, so we can ...
Hm, of our size? Not sure, I k ow a bunch of small projects filled up quickly. Not sure what the follow through was, or how much handholding was needed either. I'm definitely more available this week to help with that.
Not sure what it was like last year though since I was solely on the contributing side haha
analog bridge
oh you have updates to the PR incoming? that's excellent news!
yes, I'll update the PR by tomorrow and ping you then
proven garnet
ember iris
Hm, of our size? Not sure, I k ow a bunch of small projects filled up quickly. ...
That'd be great, but make sure you take time to balance with life and school--I need to pick up the slack on my end (I'm hoping for a quiet week but life isn't predictable). I don't think I have enough perspective to suggest to or not to share on the other discord unfortunately.
proven garnet
@lone axle mind if I use your PR as guinea pig for so CI changes? I'd get something into main and use yours (which we know should pass CI) to test the new build and release changes.
Specifically the one for imageload
It might mean it sits for a little bit, while the new CI gets any wrinkles ironed out.
lone axle
<@382939733107408897> mind if I use your PR as guinea pig for so CI changes? I'...
Don't mind at all. Sounds good.
proven garnet
Thanks!
random junco
@turbid radish I requested a review on the notes doc PR from you here: https://github.com/adafruit/adafruit-circuitpython-weekly-meeting/pull/86
@turbid radish If I did anything wrong or need to change anything, just let me know. Thanks!
onyx hinge
@lone axle there's a lot less RAM on the rp2040 than on an esp32s2 with psram. if large json responses are needed, we should really look into "json streaming with visitor pattern" so the whole json data doesn't have to fit in RAM. https://github.com/adafruit/circuitpython/issues/5512 -- the portalbase way of saying 'set this label from this json path' would fit well with it.
lone axle
<@382939733107408897> there's a lot less RAM on the rp2040 than on an esp32s2 wi...
In my case it was an HTML file returned instead of JSON. Using this constructor from httpserver library: HTTPResponse(filename="sizetest.html")
onyx hinge
ah, well, a similar thing would be needed, because RAM is RAM and we like to use it.
@proven garnet this looks fishy
pattern: "bundles/*"
github-token: ${{ inputs. }}```
inputs dot what?Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1YtDimxJ-FXu-88WqLRq-TFmE1q-xTAp-OIEiUpwkFGw/edit?usp=sharing
lone axle
@onyx hinge Would it be expected to raise a memory error if it ran out trying to return a response? I was not getting an error on the circuitpython side. It seemed to think that the response was returned successfully, on the client side it knew the content-length, but believed that it had received 0 bytes.
onyx hinge
huh! yes I would have expected a MemoryError in that kind of situation. I guess I only assumed that was part of what was happening
lone axle
I'll get an issue created with everything I found about it.
onyx hinge
OK
...
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
MemoryError: memory allocation failed, allocating 158208 bytes
```Just reading a big file into memory gives a MemoryError. But .. you're working with HTTP server, not HTTP client?Is the limit above or below 8kB?
lone axle
```>>> with open("bigfile.txt") as f: content = f.read()
...
Traceback (most re...
Correct httpserver and the limit I was seeing was 2920 bytes is okay and gets returned successfully, 2921 bytes is too much and consistently doesn't get any of the response content.
idle owl
@lone axle Hey. So, I'm working with the LTR-303 and 329 (which are basically the same thing except 303 has an int pin and interrupt capabilities). The int pin is broken out to the header on the 303. My assumption was, you create an object with the pin, and then poll the pin and then do something with that. I seem to be wrong, or not understanding it properly, because in the example, "interrupt capability" is simply enabled, with no pin assigned, and evidently used. Here is the example. https://github.com/adafruit/Adafruit_CircuitPython_LTR329_LTR303/blob/main/examples/ltr303_advancedtest.py
Here is a link to the PCB files if you need to look at the board: https://github.com/adafruit/Adafruit-LTR-329-LTR-303-PCB and here is the product link, if you simply want to look at it: https://www.adafruit.com/product/5610
I guess what's confusing me is, if you're not polling that pin, and the interrupt capability is an internal thing, why is the pin broken out to the header?
Or maybe it's an internal thing that you can also poll if you wanted to?
lone axle
Or maybe it's an internal thing that you can _also_ poll if you wanted to?
Yep, I think it's exposed and you can poll it if you want to. But even if you don't poll that pin, assuming your reading values in your main loop, and not spending too much time on other stuff it would essentially behave the same.
idle owl
(I'm writing the guide for this, and want to make sure I'm understanding it properly, and explaining what needs to be explained, and avoiding what doesn't.)
lone axle
As far as I can tell that example that you linked enables the mode so that the sensor will flip the INT pin high and low for you to notice, but the example doesn't actually poll it. So it's enabled but ignoring it if I understand correctly.
lone axle
I think in order to use it (beyond just enabling) you'd need to connect this pin to an IO on your microcontroller:
idle owl
That's what I thought
lone axle
then set up a DigitalInOut with which ever pin you connected and then you'd be able to pull it's value to see the change.
idle owl
Exactly what I thought
I was even prepared to wire it up that way over here for testing.
ember iris
Looking at the code, it reads to me that you can set the conditions on the LTR-303 so it toggles the pin under specific light conditions. That way, if your board is suppose to change behavior when the light changes, that pin can be a simple way to identify that the light has changed without requiring your code to constantly check the LTR-303
idle owl
But then there was no pin in the code.
idle owl
Looking at the code, it reads to me that you can set the conditions on the LTR-3...
But that only makes sense if the pin is connected to the micro, right?
Am I understanding what you're saying correctly?
ember iris
But that only makes sense if the pin is connected to the micro, right?
Yes, but some micro's have hardware interrupts (which I don't think is easily exposed in circuit python) so when that hardware interrupt triggers new code can run. That lets you never poll a pin, but gives you the ability to change your micro's behavior on a pin change
idle owl
Yes, but some micro's have hardware interrupts (which I don't think is easily ex...
Even so, still needs to be connected, right?
(Sorry I'm being redundant here, simply trying to understand.)
lone axle
Yep, I believe you're understanding is right. It would need to be physically connected to an IO on the microcontroller and have code somewhere "watching" it in order for it to make any real difference.
idle owl
(Dealing with brain fog, so things are more difficult to grasp at the moment. 😕 )
ember iris
(No problem I'm lifting my head out of a mountain of code and very slowly re-realizing how to word things)
Yes, if you're going to change your device behavior on the interrupt it should be connected, but looking at the example I think it's just showing it's an alternate option to interact with the LTR-303, not a required way to engage with the LTR-303
idle owl
Ok.
I think I'll ask Limor why the code does nothing physically with the int pin, so I know from the author what the reasoning was. I was going to include an example in the guide using the int pin, but there isn't one, not really anyway. So that might not end up happening.
ember iris
I think it's presence in the advanced test is to test that those commands are being set
idle owl
Either that, or update that example if she wants me to. But I'm basically certain that won't be the answer.
Fair enough ok
Welp, it's snowing here.
Thanks @lone axle and @ember iris . It seems I was right in what I was thinking, and now I can ask Limor a smarter question, which is always better.
lone axle
yep, you're welcome 👍
ember iris
https://www.mouser.com/datasheet/2/239/Lite-On_LTR-303ALS-01_DS_ver 1.1-1175269.pdf
On page 25 of the data sheet it shows a flow diagram of a suggested usage of the interrupt pin
Basically, power on the device,
set interrupt conditions,
Only read the interrupt pin
If it triggers begin reading the data via i2c
onyx hinge
@lone axle should probably try decreasing the 8192 in the file down to something like 2048 or lower: ``` def _send_file_response(self, conn, filename, root, file_length):
self._send_bytes(
conn,
self._HEADERS_FORMAT.format(
self.status, MIMEType.mime_type(filename), file_length
),
)
with open(root + filename, "rb") as file:
while bytes_read := file.read(8192):
self._send_bytes(conn, bytes_read)
I'll give it a try
lone axle
https://www.mouser.com/datasheet/2/239/Lite-On_LTR-303ALS-01_DS_ver%201.1-117526...
thats neat that you can set up the conditions. So you can use it as a simple switch but still tune it to your own needs.
ember iris
https://www.mouser.com/datasheet/2/239/Lite-On_LTR-303ALS-01_DS_ver%201.1-117526...
Advanced test probably tests the i2c register settings for the 303, but doesn't look like it takes advantage of the interrupt pin, which probably is done so end users always see their light data regardless of their lighting conditions. (The code could look broken if it's only suppose to run in bright light, but the user has low light when they're testing it)
ember iris
thats neat that you can set up the conditions. So you can use it as a simple swi...
it's such a clever design, especially if you need to save a degree of processing power or power management. Heck even keeping the i2c bus open when the device isn't in use without needed to spend clock cycles to confirm it's not in use.
onyx hinge
@lone axle the other thing that's suspicious about 2920 is that it's 2 times the "maximum segment size" of 1460 bytes. https://superuser.com/questions/1652024/why-is-the-tcp-mss-maximum-segment-size-1460-bytes
onyx hinge
@lone axle OK I reproduced the "2920 problem" without requests. I still think a smaller buffer might be in order there -- 8192 bytes can be a lot of heap -- but there's clearly a core problem.
send(2921 bytes) waits awhile, says it succeeded, but nothing is received on the other end
lone axle
@onyx hinge changing to a smaller buffer size inside httpserver does seem to fix it! I was able to return the 2921 sized file.
And for what it's worth it seems consistently faster than it was before. Taking about 1/4 to 1/3 the amount of time to transfer same data as old version (plus succeeding on 2921). I'll PR httpserver.
onyx hinge
OK please file a core issue though, the 2921 thing is real
lone axle
Is it expected to get [Errno 112] EADDRINUSE when resetting with soft reset by saving a file and/or ctrl-c / ctrl-d on the pico w? I've noticed I get that most times when it resets for those reasons. microcontroller.reset() fixes it when it comes back up from that.
made one here. I'll change the title to include the 2920 reference though.
onyx hinge
EADDRINUSE with a server? sort of. I'm not clear on the details. There's supposed to be a socket call that you can make to change the behavior. maybe CP should always make it implicitly; I haven't checked if espressif port does this, or done servers on espressif much either https://stackoverflow.com/questions/3229860/what-is-the-meaning-of-so-reuseaddr-setsockopt-option-linux
lone axle
EADDRINUSE with a server? sort of. I'm not clear on the details. There's suppose...
Yep, I see it with the HTTPServer test script. I eventually added an except block for it with microcontroller.reset() inside so it will retry after a few more seconds.
idle owl
@ember iris @lone axle The final decision is to use that example, but show wiring an LED directly to the interrupt pin, to see when it's high and low. Thanks again!
lone axle
PR for HTTPServer library is here if anyone has a chance to look over or try it out. It gets the larger responses working on Pico W. https://github.com/adafruit/Adafruit_CircuitPython_HTTPServer/pull/15
Nice, I like that idea. Makes it possible for user to see what it does without adding any complexity to the code.
idle owl
Nice, I like that idea. Makes it possible for user to see what it does without a...
Agreed.
ember iris
<@459119350851567626> <@382939733107408897> The final decision is to use that ex...
Woo is a really nice and minimal change that makes that functionality make more sense
proven garnet
<@135415127879974912> this looks fishy
```
pattern: "bundles/*"
gith...
Oops, yeah, I'll fix that
onyx hinge
@lone axle can you approve https://github.com/adafruit/circuitpython/pull/7075 ?
it has two good test results as far as I understand the responses.
blissful pollen
if he's not here i can approve it in a couple min
lone axle
<@382939733107408897> can you approve https://github.com/adafruit/circuitpython/...
yep, done. 👍
onyx hinge
Thanks!
proven garnet
<@135415127879974912> this looks fishy
```
pattern: "bundles/*"
gith...
Fixed, this references the action input
onyx hinge
[tio 19:50:25] tio v1.32
[tio 19:50:25] Press ctrl-t q to quit
[tio 19:50:25] Connected
^[]0;🐍
``` I just got an incomplete status bar message escape code with tip-of-main. I had been at the REPL ">>>" prompt but stopped & re-started my terminal program. Can't reproduce it.though the fact that it says "^[" makes me wonder if somehow it was still the stty echo problem that's a linux bug. I'm gonna choose "not caring" on my bingo card
manic glacierBOT
midnight ember
@onyx hinge thank you for the hug report during the meeting. I’ve been hashtagging circuit Python on twitter to submit stuff to the newsletter. I assumed that was a valid way to submit? Is there a different preferred method? Finding the newsletter on GitHub through searching I’ve found to be slow and often frustrating. Then I think I have to make a pr which gives me anxiety. Does the newsletter have its own repo I can fork? I LOVE the google doc format the meeting uses. It works phenomenally well. If it was as simple as that to submit to the newsletter I’d definitely use that method instead.
thorny jay
<@426200547024961539> thank you for the hug report during the meeting. I’ve been...
That would be a question for @turbid radish.
My preferred way to contribute is to tag #CircuitPython when I am microblogging my project (that is very useful as I am only blogging there an nowhere else except gist).
I think if you send to cpnews@adafruit.com it will reach Anne (and Mister Lady Ada) or whoever is on duty, that would be how I contribute for project I am not the author.
I don't think I ever touched the newsletter on the github (maybe once I had to remove an entry because of a misunderstanding and it was not a CircuitPython project at all). First I am lazy, and I did not feel right to self-promote one of my project in someone else newsletter (even if it is a PR that could be refused), I prefer a human to decide if it is worst it or not... even if most of the time I hope to benefit from the newsletter publicity. 🙂
I am not sure the Google Doc workflow is right for the newsletter... but I can think or more convoluted way to produce a newsletter, where there would be "news snippets" (a picture, a description, a link, a few keyword) and then building the newsletter would be about selecting the one to talk about, the order, and some highlight. Maybe this is more or less what is happening, and some news/projects make it to the newsletter, other to the blog.
🏗️ 🚧 🧱 sorry for the wall of text.
random junco
<@426200547024961539> thank you for the hug report during the meeting. I’ve been...
You can fork this repo for the newsletter: https://github.com/adafruit/circuitpython-weekly-newsletter. I don't see Anne changing her workflow. You can also email cpnews@adafruit.com if PRs make you anxious, that may be the easiest way.
thorny jay
You can fork this repo for the newsletter: https://github.com/adafruit/circuitp...
At one point (if not done already) Anne will follow you and pick your project for the newsletter without even you noticing it. 🙂
midnight ember
So I forked the newsletter, do not recommend. It was like 4GB. Then read on the newsletter github page they recommend editing the draft in markup from the site, not forking it. That should be easier to work with as I prefer using github site over submitting nailbiting PR's anyway.
random junco
Sorry! I send in a PR every week, never thought about editing it straight in markup on Github
brazen hatch
My uni got enterprise.
Basically you need username and password for it.
midnight ember
Do I get credit for https://github.com/adafruit/circuitpython/issues/3364 or no because it doesn't involve a PR or MR? That one wasn't an easy one, I did have to put in the work on it.
tulip sleet
Do I get credit for https://github.com/adafruit/circuitpython/issues/3364 or no ...
You have our eternal gratitude 🙂 . I'm not sure how to get Hacktoberfest credit for this.
random junco
They talk about testing as a low-code solution in the table here: https://hacktoberfest.com/about/
midnight ember
Coming from you Dan that's good enough for me. 🙂
random junco
I don't understand what they mean by create a PR to track low code or non-code contributions though
tulip sleet
@midnight ember I can transfer the issue, but it still is a pin mapping thing. Is there something the library can do about it.
for the localtime thing, is there a place to contribute an example that could be a PR? I'm not sure what a no-code PR would be.
midnight ember
I honestly don't know because I can't trace the physical pins easily to compare the pinout from the physical ESP chip vs Adafruit board that is working. I have no idea why it's not working. Could be user error sure, I've had my fair share of making mistakes lately especially with the lora board wiring. All I know is it's not working, no one in discord can explain why, so I really don't know what to do. There's an incompatibility that exists somehow and it's over my head.
tulip sleet
I think if you compare the colorful pin diagram for the Feather S3 (s2?) vs our similar Feather, you can see what the differences are. The peripheral we are working only works on certain pins or certain sequences.
see what the pins are with our Feather, and then try to use the same on the UM Feather. But it may be that some pins are not exposed, since the Feather does not have enough room to break out all the pins
midnight ember
the pretty pin diagram only shows the pinouts which could be coming from any gpio out of the esp chip. i need to verify each pinout of the esp is the same. while i'd love to take UM at his word i really do need to verify the routing.
tulip sleet
the pin diagram labels which GPIO is which. It should be correct. I don't think UM would let a mistake stand, because someone writing code depending on the diagram would say "GPIO13 isn't working", etc.
you could just toggle each pin as GPIO to verify the routing
just in the REPL, etc.
midnight ember
ok so if the pins are all 100% correct, then has to be a software bug, hence the bug report?
tulip sleet
I wouldn't call it a bug, it's that the pins the featherwing needs and takes from our feather are not the same as the pins on the UM Feather
tulip sleet
i didn't know he said they were the same; his board was first, and we didn't try to copy his board
when we did our own layout
midnight ember
oh i didn't know that, thought it was the other way around
tulip sleet
just put the diagrams next to each other and look
midnight ember
well he said it was the same in the UM discord when i asked him. again, i have no way to verify. i guess i could toggle each gpio, will probably have to do that.
i have and they match up!
hopefully you can see why this is confusing, something doesn't add up.
tulip sleet
yes, I would say to toggle each one, on each board
ok now I understand; I was making no assumptions about pin matchup. Which one is this? FeatherS3 or S2?
midnight ember
UM FeatherS3
compared to Feather M0 & ESP32-S2 & S3 since that's what i know works with the matrix featherwing.
tulip sleet
it would be good to compare the board files in CircuitPython as well. There could be a typo in one or the other
midnight ember
didn't think of that.
tulip sleet
and it could be something about how protomatter is detecting the board type, maybe there is a bug in the compile-time flags
midnight ember
i know the arduino part of the library has variants explicitly in it but couldn't find something like that in the CP side.
tulip sleet
so protomatter is being compiled incorrectly for the FeatherS3 but not our Feather. But are you saying it works on the Adafruit Feather ESP32-S3, and not on the UM FeatherS3?
like and like?
midnight ember
was looking through arch.h and that was over my head.
correct.
i have video demos to prove it, was confirming they work because they're not specifically listed in the rgb matrix featherwing guide.
so was doing a nice thing to have those added as confirmed working.
dynobot you vex me.
if there's a mislabeled pin on the M0/M4/RP2040/ESP32-S2/ESP32-S3 feathers it would have to exist on all of them. because the matrix featherwing works with all of them, it only doesn't work with the UM FeatherS3.
analog bridge
What would be the place to put code that is shared between multiple modules at the shared-bindings level?
Example: microcontroller.nvm and alarm.sleep_memory have very similar code in shared-bindings
proven garnet
Any PR you submit with the hacktoberfest-accepted label will count. So either an empty commit or even a junk commit you submit should be good, we should draft it to prevent merging, and we'll just slap that label on it
tulip sleet
What would be the place to put code that is shared between multiple modules at t...
There is a shared/ directory
seems like that might be fine; its entries are not tied to module names
proven garnet
So either of those options should work @midnight ember
Get that sweet Hacktoberfest cred
I can help you with that if you submit the PR 🙂
midnight ember
I don’t see how I could turn that into a pr other than submitting it as a class for the time library. I don’t want to cheat and do a junk pr. Would rather everything be up to snuff. If it’s not good enough to be accepted by hacktoberfest then it’s not good enough.
tulip sleet
if you find a guide it could be put in, it could be a PR to the learning guides repo
it could even be a PR to that as an example that might be used in the future
idle owl
@tulip sleet Good afternoon. Jeff will be back in about an hour. Are you interested in doing a bug triage session then?
tulip sleet
<@329766224093249548> Good afternoon. Jeff will be back in about an hour. Are yo...
I don't think we actually need a bug triage session. I cleaned up a bunch yesterday that were not labeled.
idle owl
I don't think we actually need a bug triage session. I cleaned up a bunch yester...
Ah ok.
tulip sleet
did Jeff suggest one?
idle owl
Yesterday, yes. I'm following up on that.
tulip sleet
I think he saw about half a dozen that had no milestone, but they are all gone now
idle owl
Alright that works too.
tulip sleet
I would like to talk to you about where to put the "CircuitPython limitations" information. I think it belongs somewhere, either in guides or rtd or both.
idle owl
We can talk about that, sure
idle owl
Let's vid.
proven garnet
I don’t see how I could turn that into a pr other than submitting it as a class ...
@tulip sleet had good suggestions, and if none of those work, even just a dummy PR with any change at all is enough for us to attach the label needed to count it (so don't worry about contents in that case). We can close a PR like that after Hacktoberfest ends in that case.
old smelt
we keepp bumping into S3 BLE limitations for instance.
@tulip sleet - this could not be a more timely comment as I brought up Discord specifically to poke around regarding BLE on the ESP32-S3. I guess I have my answer.
jaunty juniper
yeah, can't start a service, so no HID, no UART server (no bluefruit app), etc. but you can connect to another board or send advertisement between boards like adafruit_BLE_radio does for example
glad mantle
so I've been trying to set up a build environment on linux following this guide:
https://learn.adafruit.com/building-circuitpython/linux
ultimately I need to create a .uf2 for a pcb with an embedded rp2040
I think I'm stuck on installing the toolchain or adding it to my PATH correctly
glad mantle
so when I just went to do a test build I get an error about no such file or directory "arm-none-eabi-gcc"
since that was file in the example in the tutorial, I'm not sure I got the right toolchain
jaunty juniper
that would be that step:
export PATH=/home/$USER/bin/gcc-arm-none-eabi-10-2020-q4-major/bin:$PATH
I think I may have made another error
that precipitated this
I used the cortex a build, because I believe I misread the description
cortex a should be for the "normal" pi boards, right?
not the rp2040 based boards
jaunty juniper
yeah it's the linux Raspberry Pis
jaunty juniper
the RP2040 is M
glad mantle
ok.... let me see if all of that makes a difference and I'll check back later if I'm still stuck
thanks
lone sandalBOT
candid sun
i wanted to see if anyone had any thoughts here on this issue that i'm investigating from the forums: https://forums.adafruit.com/viewtopic.php?p=944133 . basically, it's this project for the pyportal (https://learn.adafruit.com/using-microsoft-azure-iot-with-circuitpython) that throws an error "no module named 'ssl'" while importing the azureiot lib and i've been getting the same error on cp7 and 8. it looks like ssl is not a built-in module on the pyportal, so would that be causing that issue? and if so, are there plans to add it to the pyportal?
jaunty juniper
i wanted to see if anyone had any thoughts here on this issue that i'm investiga...
ssl is for builtin wifi
glad mantle
ok... so I got the (hopefully) correct toolchain now, and extracted that into a folder
and then I thought added the correct thing to my bash profile
but it still doesn't seem to be pointing to it correctly
or whatever the right terminology is
candid sun
ssl is for builtin wifi
ah, okay. i'll look deeper then. thanks!
jaunty juniper
ah, okay. i'll look deeper then. thanks!
there's 2 different ways to use the azure library, vias ESP32SPI (pyportal) or via native wifi, the guide should use the proper one for the pyportal
unless there is a bug in the library with imports
which seems to be the case since the import ssl has been added but is not used I think ?
candid sun
yes, i was also trying the esp32spi simpletest for azure but am getting another error when esp.get_time() is called (returning 0)
jaunty juniper
yeah that's part of the API, esp.get_time() must be in a try/except, it takes a few seconds before it starts returning correct values
candid sun
okay, thank you. i'll put in a try/except and see if i can get further with the simpletest at least. i've only used azure with native wifi previously
jaunty juniper
ah and the error has changed from ValueError
I saw that recently, lemme check
yeah it's an OSError now
yeah the try/except is there, but the error has changed
it makes sense, since a value error implies a value passed by the caller, but now we have to update things 😉
candid sun
so i added a try/except for OSError and was able to get the time (woo!) but the same ImportError: no module named 'ssl' occurred when it tried to import IoTCentralDevice..
glad mantle
I'm still getting this error if I try to just make one of the boards that are already in circuit python:
FileNotFoundError: [Errno 2] No such file or directory: 'arm-none-eabi-gcc'
proven garnet
I'm still getting this error if I try to just make one of the boards that are al...
What happens if you type which arm-none-eabi-gcc into the command line?
jaunty juniper
so i added a try/except for OSError and was able to get the time (woo!) but the ...
yeah I think it's a bug in the library, if adafruit_azureiot.iot_mqtt is supposed to work on ESP32SPI, this line (and the import ssl at the top) has to go or be try/excepted in some way
https://github.com/adafruit/Adafruit_CircuitPython_AzureIoT/blob/de016a19919b654ddd5e3c6c8776c3198fbd5aa5/adafruit_azureiot/iot_mqtt.py#L138
proven garnet
You'll need to follow @jaunty juniper's advice, and add the location of that file to your path,
jaunty juniper
did you start a new shell ?
candid sun
yeah I think it's a bug in the library, if `adafruit_azureiot.iot_mqtt` is suppo...
okay, thank you!
glad mantle
I restarted the computer
if that does it
I added this to my .profile
export PATH=/home/ziptyze/bin/gcc-arm-none-eabi-10.3-2021.10/bin:$PATH
if that looks correct
jaunty juniper
- I woud check it's correct by looking at what's in it
- I would check that your shell correctly uses .profile by adding an echo in it or something
glad mantle
you may have to point me in the right direction for doing those things
I am more of a hardware guy, so this is all pretty new to me
jaunty juniper
well first what does this do ?
ls /home/ziptyze/bin/gcc-arm-none-eabi-10.3-2021.10/bin
is there arm-none-eabi-gcc in there ?
glad mantle
it is
jaunty juniper
ok then what does this do ? do you see the gcc-arm path in it ?
echo $PATH
nope
glad mantle
/bin/bash
jaunty juniper
is there a /home/ziptyze/.bash_profile ?
glad mantle
originally no
then I made one
which didn't seem to make a difference
so then I got rid of it
glad mantle
yes, there is that
jaunty juniper
try putting it there
jaunty juniper
meh anywhere
glad mantle
k
did that and saved
closed the terminal
and now reopening
ok.... now if I do the echo $PATH it shows that line in there
jaunty juniper
progress !
glad mantle
yes! it compiled a uf2 for a pico
now to set one up for my board
it was not quick....
thank you for all of your help
I'm trying to build one for this: https://1upkeyboards.com/shop/controllers/1upkeyboards-pi60hse-60-pcb/
jaunty juniper
ah keyboards !
midnight ember
quick someone hide that before jepler sees it and takes on another keyboard project
tulip sleet
<@329766224093249548> - this could not be a more timely comment as I brought up ...
kattni and I discussed where to put that information -- the answer is in several places and I'll be doing that soon
glad mantle
ah keyboards !
yeah, we now have two 60%'s that have embedded rp2040s. currently we are using qmk firmware, but we want to give the option for circuit python as well, particularly because we put a qwiic/stemma connector on it. I'm very interested to see what people can do with it
I had honestly never used python before the picos/2040 came out, and I tried using one of the pro micro footprint rp2040 boards on one of our macro pads when 7 was still in beta, and I was impressed how quickly I was able to get a working keyboard going using the default libraries and very little prior knowledge
now switching our lineup to rp2040s instead of 32u4's is a no-brainer, both with capability and price
glad mantle
so... following the guide:
https://learn.adafruit.com/how-to-add-a-new-board-to-circuitpython/customizing-the-board-files
when I get to the part about configuring the led status pins etc., if I'm following along the steps using an existing board like the pi pico, they don't seem to have these options configured in the mpconfigboard.h
is that somewhere else, or is that guide not up to date
glad mantle
both of those boards should have a status led though, and I don't see it defined on either
I'm curious about that because my board has no status led, so I want to make sure it's not defined somewhere else and I'm wasting a pin
looks like it is defined in the pins.c file though
so I don't know if it needs to be defined both places
jaunty juniper
both of those boards should have a status led though, and I don't see it defined...
you mean CIRCUITPY_STATUS_LED ? yeah the QT PY has a neopixel only and... the pico doesn't use it for some reason
glad mantle
so even though it's on there, it'
it's not setup by default to be used for the status?
it's just there if people feel like turning it on
jaunty juniper
if it's defined that LED will be used as a monochrome status LED
glad mantle
but in the default uf2, then no?
jaunty juniper
I think it should be defined in the pico now that I think about it, don't know if there was a reason to omit it
glad mantle
well... just making sure I understand it correctly I guess
jaunty juniper
if you don't have an LED, you don't need to do anything about it
glad mantle
the way it's defined in the pins.c file is just giving it an extra way to be called in code right?
like you can refer to it as led or gp25
glad mantle
so if it has two definitions you can use either?