Anyone can give me a reply on my question about i2cslave ?
#circuitpython-dev
thorny dove
tulip sleet
Anyone can give me a reply on my question about i2cslave ?
I2CSlave was renamed to I2CPeripheral by us, and now in 8.0.0 we are using the official I2C renaming, which is "target"
tulip sleet
This is our link: https://docs.circuitpython.org/en/latest/README.html
thorny dove
I googled for circuitpython readthedocs and got that 'jake' one .
tulip sleet
we can't control google's use of unofficial rtd sites
thorny dove
Thank you Dan!
onyx hinge
@tulip sleet thank you for the thoughtful review! I appreciate all the things you found. PR is updated now.
tulip sleet
if you searched for "slave", that's probably why you find the "jake" one, since we haven't used that name for a long while
tulip sleet
<@329766224093249548> thank you for the thoughtful review! I appreciate all the ...
yw!
onyx hinge
not sure what to do about hostname validation. I checked, and re1.5 (the implementation of regular expressions we use for the re module) doesn't support the x{min,max} format
and POSIX regexec isn't available everywhere
tulip sleet
not sure what to do about hostname validation. I checked, and re1.5 (the impleme...
man 7 hostname gives a simplified definition:
Each element of the hostname must be from 1 to 63 characters long and the entire hostname, including the dots, can be at
most 253 characters long. Valid characters for hostnames are ASCII(7) letters from a to z, the digits from 0 to 9, and
the hyphen (-). A hostname may not start with a hyphen.
frankly we could just check for a-z, 0-9, -, ., and be done with it
onyx hinge
are you suggesting "just implement a check for it by hand"? Sure I can do that.
tulip sleet
yes, and i"m not sure how strict you really need to be
if you don't drag in regex, maybe that will save a noticeable amount of space on some builds anyway
thorny dove
Dan I readthedocs of the link you gave me, doesn't show up i2cperipheral. Only i2ctarget, but i2cperipheral is in the flashed CPY V8 code
tulip sleet
Dan I readthedocs of the link you gave me, doesn't show up i2cperipheral. Only i...
the name was changed very recently, I guess since the version you flashed
the docs link is "absolute latest", later than 8.0.0-beta.0
also, both names are available right now
i2cperipheral will be removed in 9.0.0
thorny dove
Yes, I have i2cperipheral in the device. It is not in the latest of the docs
tulip sleet
so, it's documented as i2ctarget, but is available as both i2ctarget and i2cperipheral right now, for backward compatibilty in 8.0.0. The compatibility will go away in 9.0.0
thorny dove
thank you
thorny dove
tnx
onyx hinge
@tulip sleet added to the PR https://github.com/adafruit/circuitpython/pull/6933/commits/d59774d7508e40f5ef73e6515b1ea9df0c7e8354
tulip sleet
hostname_valid looks elegant!
@onyx hinge
ports/raspberrypi/boards/raspberry_pi_pico_w/board.c is now an empty file; for other boards I include the comment I put in the suggestion (copy another semi-empty one)
onyx hinge
done!
tulip sleet
done!
okee dokee, I'll approve; did you do a wifi smoke test of some kind (e.g. to test the hostname stuff)?
onyx hinge
after the first rebase, yes. after all these changes just now, no
let me do that again
----------------------------------------
Text Response: it works
```incidentally before I reset it, it had completed >64000 http getsand the http test still works, consider it smoketested
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: invalid hostname
>>> wifi.radio.hostname = "blue.fish"
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'Radio' object has no attribute 'hostname'
```oh that's right, setting hostname is not supported on rp2040 yet (but I can sorta test validity which is weird)excellent - the http support should satisfy some people for a while
onyx hinge
yay! now I'm headed out on my walk, ttyl and thanks again
onyx hinge
Thank you @crimson ferry
crimson ferry
@onyx hinge probably not worth an issue, but same as esp32-s2 originally: reading board.LED output is always False, i.e, led.value = not led.value doesn't toggle the output
onyx hinge
Ooh please do else I'll forget it
crimson ferry
ok
proven garnet
I don't see anything in adabot about assigning the Hacktoberfest accepted label, does that need to be done manually?
crimson ferry
>>> import board
>>> import digitalio
>>> smps = digitalio.DigitalInOut(board.SMPS_MODE)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: SMPS_MODE in use
```but it's only defined in `pins.c`@proven garnet as far as I know yes
@crimson ferry ```c
#if CIRCUITPY_CYW43
never_reset_pin_number(23);
never_reset_pin_number(24);
never_reset_pin_number(25);
never_reset_pin_number(29);
I may have gotten it wrong, but these pins are taken away from their original use and are for interfacing to the cyw43Looks like it's now WL_GPIO1
crimson ferry
I was just poking to see if that pin had something to do with power management
onyx hinge
ah, no, probably nothing to do with cyw43 power management but still a good catch
the state of that pin selects "high efficiency (default)" or "low ripple" at the 3.3v regulator
crimson ferry
ah, thanks. VOLTAGE_MONITOR (29) and VBUS_SENSE (24) in use too with CYW43 ๐ฆ
onyx hinge
Yeah I'm not sure the deal with VOLTAGE_MONITOR, it's supposed to be possible to use it but ๐คท I couldn't get it to work
I should probably study the micropython implementation more closely
crimson ferry
it would be nice if the Raspberry Pi docs were more complete
proven garnet
<@135415127879974912> as far as I know yes
Thanks!
onyx hinge
I wrote a script that can swap the "//!" comment markers that embed Python interface code in C comments with "# C#" comments that embed C code in Python interface code ... by using my editor's "pipe buffer through external program" feature, then overriding the language type manually, I can conveniently edit the docs as Python code before piping it again to switch back. Since the source gets uncrustified and blacked by pre-commit, it doesn't seem to introduce any spurious changes. Not sure "# C#" is the best comment marker, it's just what I implemented... https://gist.github.com/jepler/94825738dff594dff53df169dde98de4
tidal kiln
is installing libraries to non-native USB boards covered generically somewhere? like more generically than here:
https://learn.adafruit.com/circuitpython-with-esp32-quick-start
jaunty juniper
is installing libraries to non-native USB boards covered generically somewhere? ...
that's the only guide about non USB boards currently, isn't it ?
maybe it should be a bigger item so people find it easier ? I only see it mentioned in a paragraph. https://learn.adafruit.com/circuitpython-with-esp32-quick-start/using-web-workflow#step-3124558
tidal kiln
jaunty juniper
ah nice I missed that one
idle owl
I don't see anything in adabot about assigning the Hacktoberfest accepted label,...
Um..... I thought it was automatic.
Maybe Tim was running it manually so I assumed it was Adabot doing it?
@lone axle I merged a new library example in to the HT16K33 library a day and a half ago, and the dynamic screenshot isn't showing up in this list. https://adafruit.github.io/Adafruit_CircuitPython_Bundle/ I'm not sure why. Usually it's quick. I asked Justin first, forgetting you had written this (at least I'm 80% sure you did, I remember Jeff being involved as well somehow, but I'm not sure to what extent). Justin looked, but couldn't quickly figure out what might be going on. Any idea why my new example isn't showing up in the list?
lone axle
<@382939733107408897> I merged a new library example in to the HT16K33 library a...
Nothing really comes to mind sorry. I think the portion I made initially was making those screenshots for learn guide projects. It was later re-purposed to also do examples in the libraries, but I don't think I did that part, or at least don't recall it ๐ .
No ideas about what could be the cause, but I would guess the best way to troubleshoot is start looking into the actions to find when/where these get generated from and try to find some output from it. I could look into it tomorrow afternoon if you'd like.
I do think it may be picky about the example names following the format of having the library name at the beginning of them. But looking into latest commits in that repo it seems all of them do follow that formatting. Not sure of any other gotcha's though.
idle owl
Ok, that would be great. I don't know where to start looking into that, so any help would be appreciated.
The guide page was already live, so there's not really a rush on it. I'll keep the note in my todo list to add it until we figure it out so it won't get forgotten.
@lone axle Tomorrow afternoon would be excellent. Thank you for offering.
lone axle
<@382939733107408897> Tomorrow afternoon would be excellent. Thank you for offer...
Yep, for sure will take a look. Had just a moment now and took a quick peek. I think this is the action run for it and it shows itself as cancelled (although also has green checks that look like success/completion, so kind of ambiguous). https://github.com/adafruit/Adafruit_CircuitPython_Bundle/actions/runs/3150402665 and this is the workflow file for it: https://github.com/adafruit/Adafruit_CircuitPython_Bundle/blob/main/.github/workflows/images.yml
idle owl
@lone axle Ooh nice, ok. I'll rerun it to see if that resolves it. I'll keep you posted.
lone axle
looks like it's been failing for a little while and you are the lucky one who noticed: https://github.com/adafruit/Adafruit_CircuitPython_Bundle/actions?query=workflow%3A"Generate+folder+images"++
lone axle
Yep. This is the most detailed error I've found so far:
Canceling since a higher priority waiting request for 'folder-images' exists
My guess is maybe there is some old run instance of it that has gotten "stuck" and for some reason has higher priority so it's causing these to fail. Okay, off to prep for Javascript class tonight. Will dig further tomorrow.
lone axle
Thank you!
proven garnet
Maybe Tim was running it manually so I assumed it was Adabot doing it?
If nothing exists I'd love to get something up for next year in adabot or something.
idle owl
If nothing exists I'd love to get something up for next year in adabot or someth...
I seriously thought it was already. But apparently I'm wrong.
Wheeeeeeee I am not wrong! My memory is not exceedingly terrible after all. It was in there at some point. https://github.com/adafruit/adabot/pulls?q=is%3Apr+is%3Aclosed+hacktoberfest The 2019 PRs from sommersoft added it. (This was bugging me, I swore I remembered it being in there. I was very confused.)
Code's already written. Might need tweaking to update it, but at least it's mostly there already.
Oh, turns out I completely misunderstood what Tekktrik was saying. ๐
That was already known. It's a different label he was referring to.
Ignore me, nothing to see here, please move along.
onyx hinge
@idle owl I don't know what's up with the folder images. I agree the web-ui says that the update-images task was canceled.. but it also is the case that the latest commit of update-images is very recent and matches what the log of the last run says it pushed.
commit 9fd3cae77f79f3244372a7ed0e8936ec3dc811ff (origin/folder-images)
Author: jepler (github actions cron) <jepler@users.noreply.github.com>
Date: Thu Sep 29 20:12:15 2022 +0000
update images
diff --git a/24lc32_24lc32_simpletest.py.png b/24lc32_24lc32_simpletest.py.png
Log
To https://github.com/adafruit/Adafruit_CircuitPython_Bundle
+ 760d265...9fd3cae HEAD -> folder-images (forced update)```
Since I wasn't in on this whole discussion, what else are you seeing that led you to be looking at this?That my HT16K33 example isn't showing up on the list.
I added it 1.5+ days ago.
So.... it's not running. Or something.
onyx hinge
OK
has the library been released since your changes? It was released September 24, apparently.
idle owl
has the library been _released_ since your changes? It was released September 24...
Oh. No I doubt it.
Uff.
Is that why it's not there?
onyx hinge
Yes, I think so? Since bundle is all about having the latest tagged release. so your image would be missing from and adafruit-circuitpython-bundle-examples-20220929.zip wouldn't have it either. not that many folks get their examples that way.
idle owl
Ok. Released it. I'll check again tomorrow.
idle owl
No worries. I should have remembered that.
onyx hinge
I submitted a support ticket with github about the workflows showing as cancelled. As it's associated with my github account I don't think anyone else can use the URL so I won't share it.
idle owl
I submitted a support ticket with github about the workflows showing as cancelle...
Ok, that works. Thank you.
crimson ferry
@onyx hinge had you seen this https://forums.raspberrypi.com/viewtopic.php?p=2037055#p2037055 (it's in the linked georgerobotics code too)
I'll test uP (and CP when there are artifacts) to see if I can turn it on and off and keep the bits flowing after some large-ish interval. OSError: -2 is a bummer, needs reset to recover afaict.
onyx hinge
@tulip sleet I backed out the black changes from that PR just as you were reviewing it, argh! mind re-reviewing?
tulip sleet
np!
edgy edge
thanks for the detailed response, I hope I'm not missing something, but I'm not sure what you wrote will work for me. My goal is to encapsulate a third-party library so that I can call it from circuit python. The goal is to take that library, unmodified, and put it in lib as a sub module (lib/neuton). My wrapper code is in shared-modules/neutonml and must include lib/neuton/neuton.h. The failure occurs when compiling the shared module because there is no way I can find to set CFLAGS only for shared-modules/neutonml/Neuton.c. I don't understand how the example you provided could work for my code in shared-modules. I don't want to modify the vendor files because that would be a support nightmare - the sub module get rewritten every time Machine learning model is trained. I guess I could come up with my own .h file and just keep that updated as needed with each training.
midnight ember
I might be interested in trying to tackle this issue: https://github.com/adafruit/circuitpython/issues/3364 since I've already laid some groundwork for exploring it. Where can I find the time.localtime class/method in github?
nvm think i found it in shared-bindings/time
yeah that's definitely it. because i've worked with time in some of my projects i think this will be a good entry point for me.
manic glacierBOT
onyx hinge
thanks for the detailed response, I hope I'm not missing something, but I'm not ...
where's the code now? I'll take a look.
I found https://github.com/aulitech/circuitpython/ but as https://github.com/aulitech/Neuton is not a public git repository I can't help you with it.
tulip sleet
@onyx hinge I finished draft release notes for 8.0.0-beta.1. I have tenatively included your 6972 and 6973, and my 6971 (update frozen libs). Happy to wait for a PR for the power-save stuff, since that will reduce support issues considerably.
crimson ferry
(testing now on parallel pico w's)
proven garnet
@gilded cradle is there a use case for the generic x86 version of Blinka? A contributor wants to wrap a board import in a try/except but I know this isn't typically done. They're referencing this file: https://github.com/adafruit/Adafruit_Blinka/blob/main/src/board.py which errors if its a generic x86, but is it even possible to define a display in that implementation?
For reference, it's related to this PR: https://github.com/adafruit/Adafruit_CircuitPython_turtle/pull/31
I don't really know how the generic x86 version of Blinka gets used, and where it's worth accommodating, or where we should just say "if there's no board import, we should actually raise an exception because it doesn't make sense to import this module then".
brazen hatch
Should you need me to test something just go ahead and @ me
I am more than willing to let my picow afk the network connection overnight
midnight ember
@proven garnet perhaps 32-bit windows which runs on stuff like the beagle bone, intel compute stick.. That would be a really long way around to running Linux vm on 32bit windows on a beaglebone. Crazier things have been done though.
proven garnet
I guess it's possible, but what use case is there for Blinka without defining board pins? I figure if you're on Blinka you probably want/need pins defined for use, but maybe I'm missing something.
And then if you can't use pins, can you even instance a display object?
onyx hinge
@tulip sleet I'm back now! working on better documentation for the cyw43 power save mode, and considering whether to set a different value when circuitpython starts
onyx hinge
the good news is I did find the documentation of the power saving mode numbers as well!
what interval have you reached with 0?
and have you had failure?
crimson ferry
with 0 it will generally keep going like the Energizer bunny
with 2 (CP default, right?) it's inconsistent... on uP it was usually between 5-10 minutes. In latest CP, it sometimes goes 1/2 hour before OSError: 2
(various similarities and differences in the implementations, and of course there's always the unknown of network activity affecting something although I can't quite think of the mechanism)
testing CP... code doesn't know if it's connected, so I just always try connect() after failure, counting on its NOP if already connected. That keeps things going, but it's more complicated and different from espressif
onyx hinge
wifi.radio.connect, you mean?
crimson ferry
yes
onyx hinge
I guess it's slightly incompatible with esp32 that wifi doesn't disconnect on soft reset
crimson ferry
that too
and the IP address thing, I think that's a uP bug. uP does recognize in it's status codes that 3 means connected and got IP
but then the IP sticks around after disconnection
onyx hinge
@crimson ferry ^^^ #6976 will set the 0xa11140 mode at startup. your review is appreciated.
or hmmm .. did I break it?
weird it didn't reset after uploading the new uf2
crimson ferry
where did the 4th mode come from? I thought uP had 0xa11140, 0xa11141, and 0xa11142
(I may have glazed over the driver code too fast)
onyx hinge
crimson ferry
oohhhhh, fancy
onyx hinge
those 3 plus 0xa11140
manic glacierBOT
Is there a broad timescale for the inclusion of the ussl module in the Unix port?
upip, etc are non-functional without it.
It would be very useful to prototype apps within a Unix port and then transfer them to the devices, and it would really reduce debugging time, etc
Thank you, Brian :)
PS: I am very happy to help test any Unix port!
midnight ember
I bet my lora boards will work fine now. :/ Everything I've been going through lately just ugh. I thought Circuit Python was falling apart and riddled with bugs. Nope, just my USB controller going insane.
Unplugged every single USB device and it was still there. Delete/uninstall it and it would show right back up with a refresh.
midnight ember
Had to uninstall the main 3 USB controllers in the right order so my mouse was last, then install that one, hard restart and voila. Apparently it made about 100 USB devices in hidden devices.
tulip sleet
Had to uninstall the main 3 USB controllers in the right order so my mouse was l...
did you use the USB cleanup tool?
midnight ember
The crappy part is that now all the bug reports I've submitted have to be wiped out as erroneous.
Wasted a lot of my time and others, sorry about that to anyone who had to deal with one of my confusing bug reports... because I was more confused than anyone.
there was a tool? thought it was just a viewer?
tulip sleet
no, it removes all the unconnected devices
midnight ember
oh yeah i tried that. it refused to remove the device
had to remove all the controllers instead, that's what ended up doing the trick.
apparently there's some kind of enumeration bug that will get stuck at that specific registry key
tulip sleet
https://www.uwe-sieber.de/misc_tools_e.html scroll down to "Device Cleanup Tool". It removes the usb-attached devices, not the native drivers
midnight ember
lol welp that could have saved me a lot of time fighting with it
i got the tree viewer and thought that was the tool you were pointing me towards :/
this is why i can't have nice things
tulip sleet
ok, I made that clearer on the Learn Guide page
midnight ember
thank you for your help danh, i've been fighting against that one for a while, weeks.
tulip sleet
try that tool anyway. It will remove all the stale CIRCUITPY's, COMnn, etc.
you may have done that in a more forceful way
midnight ember
i think the reason i missed it is because in device manager it doesn't have a yellow exclamation mark, it's a failure but grayed out one, sneaky unless you're looking for it.
lone axle
@idle owl if you're around. No worries if not... Was your new HT16K33 example this one? ht16k33_segments_14x4_demo.py
That one does appear in the list now.
manic glacierBOT
edgy edge
I found https://github.com/aulitech/circuitpython/ but as https://github.com/aul...
https://github.com/aulitech/circuitpython/tree/neuton is public now. I double checked from an incognito browser. The other repository you were looking at is the unadulterated vendor code pulled in as a sub module under lib/neuton
onyx hinge
@crimson ferry 1800 second delay successful with the automatic power saving here
from 300 to 1800 by 100 second intervals in a single test over the last couple hours
I expect it to fail overnight when my wifi router resets, no surprise there
@edgy edge This is the problem I am encountering while trying to check into your situation. In my experience this indicates either https://github.com/aulitech/Neuton.git does not exist or it is not public. ```jepler@bert:~/src/circuitpython$ git checkout aulitech/main
HEAD is now at ecaccd0dbc Merge pull request #1 from aulitech/neuton
jepler@bert:~/src/circuitpython$ tail -3 .gitmodules
[submodule "lib/neuton"]
path = lib/neuton
url = https://github.com/aulitech/Neuton.git
jepler@bert:~/src/circuitpython$ git submodule update --init lib/neuton
Submodule 'lib/neuton' (https://github.com/aulitech/Neuton.git) registered for path 'lib/neuton'
Cloning into '/home/jepler/src/circuitpython/lib/neuton'...
Username for 'https://github.com':
Just fixed that
onyx hinge
thanks, I'm now able to build your code
crimson ferry
<@360465312518504448> 1800 second delay successful with the automatic power savi...
with mode 2? what changed? I can test the PR when there are artifacts
onyx hinge
@crimson ferry no with the 0x...0 mode
crimson ferry
ah, good... yeah that one is rock-solid ๐
onyx hinge
@edgy edge I think you were really close! ../../lib/neuton/neuton/neuton.c:340:10: error: no previous prototype for 'neuton_model_get_inputs_ptr' [-Werror=missing-prototypes] 340 | input_t* neuton_model_get_inputs_ptr() | ^~~~~~~~~~~~~~~~~~~~~~~~~~~ note how the kind of diagnostic is shown in the brackets in the error. You had -Wno-strict-prototypes but not -Wno-missing-prototypes.
I think that with this change I just PR'd to you the compiler diagnostics problem should be solved in a way that is palatable
@crimson ferry yeah I'll return to that PR in the morning and fix whatever's stopping it building. It's getting late here, so I'm gonna call it quits
crimson ferry
of course, have a good one
onyx hinge
haha what a coincidence, the diagnostic is ... ./bindings/cyw43/__init__.h:53:1: error: function declaration isn't a prototype [-Werror=strict-prototypes]
well I guess there are more flags that you changed but .. I don't think the situation is hopeless, I think they just need to be dealt with one at a time
edgy edge
<@588158503453982733> I think you were really close! ```../../lib/neuton/neuton/...
Actually I had it compiling and building. Not sure why your results are different. To get rid of that error it took a change to the makefile where -Werror=missing-prototypes is specified
so I don't have a problem getting it to work, it's just that the CFLAGS are applied to all shared modules
onyx hinge
https://github.com/aulitech/circuitpython/pull/3 updated to, I think, fix the rest of the global CFLAGS changes so that they're local to your new code
now for real this time, goodnight ๐
lone sandal
thorny dove
CPY V8.0.0-beta.0-49, a call to I2CTarget.request() results in: 'TypeError: extra positional arguments given' (see the compiled image)
onyx hinge
@thorny dove that sounds like a bug, but I'm not sure if it's a documentation bug or not. Please submit a github issue.
thorny dove
Thank you for your reply. Will submit.
I viewed the sources. There also define one param as float. So, at least the doc should give as example and description ... (-1.0) or so.
onyx hinge
Please provide a full script to reproduce the issue
thorny dove
OK
onyx hinge
I viewed the sources. There also define one param as float. So, at least the doc...
Please feel free to contribute improvements to the documentation as well
thorny dove
Yep
onyx hinge
It's a doc bug. The signature should read: def request(self, *, timeout: float = -1) -> I2CTargetRequest:, i.e., timeout is a keyword-only argument.
thorny dove
That means that a call should look like this: 'target.request(timeout=-1.0) ?
onyx hinge
@thorny dove yes
>>> device.request(timeout=5)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
OSError: [Errno 116] ETIMEDOUT
An issue, or better yet a PR to fix it, would be appreciated from you once you've verified it as well ๐
thorny dove
OK
thorny dove
target.request(timeout=-1) is accepted.
midnight ember
Looks like most of my issues started happening when I installed WSL. Getting Hyper-V errors the same day the enumeration driver got corrupted. Could also explain why I've been getting duplicate USB IO warnings because I'm literally running 2 OS's even if Linux is always in the background it's still there. I'm setup with VSCode & Windows GIT desktop now so I really don't need WSL unless I'm going to try to build a custom board (much easier in Linux than Windows python). Going to uninstall WSL and revert the Virtualization setting in BIOS to off.
If WSL is to blame I'm going to take back all the nice things I've been saying about WSL the past month.
blissful pollen
Strange, I've had zero issues with WSL
proven garnet
Interesting issue regard GitHub Actions @lone axle. I love peaking at repositories' CI, so I'll have to watch one of your previous streams about it.
onyx hinge
I cannot understand why the last actions build of https://github.com/adafruit/circuitpython/pull/6976 did espressif boards!
tulip sleet
Looks like most of my issues started happening when I installed WSL. Getting Hyp...
I don't think you need to turn off the Virtualization setting in the BIOS. That's pretty harmless.
I've found WSL, even WSL2 to be slower than a conventional VM running Linux. But I just use a native Linux machine for development. A conventional VM might also work better. I like VMWare Workstation (free for personal use) better than VirtualBox.
tulip sleet
@onyx hinge @crimson ferry worth adding a note to https://github.com/micropython/micropython/issues/9455 to point to our solution
crimson ferry
I'll add a comment
or actually @onyx hinge, if you'd prefer, you'll be able to explain it better
onyx hinge
sure! (already did, though it was just a quick answer)
tulip sleet
I think we crossed ๐
@onyx hinge I think I will release 8.0.0-beta.1 today after the currently building PR's finish. Anything else you can think of?
onyx hinge
I think picow is at a great spot to release and get more feedback. Let me take a quick glance and see if anything else pops up to me
I don't see anything else in PRs that it would make sense to merge right now.
@tulip sleet ^
tulip sleet
sounds good, and there will always be more ๐
midnight ember
QT Py S3 works flawlessly but I don't have any error correction setup in my code. Looking into it.
edgy edge
now for real this time, goodnight ๐
s thanks again for all your help. The changes you made appeared to work. After a bit of testing I decided to merge my branch with main on my fork, then pull in all commits since the fork. Boy did that screw things up. Time to watch football and come back later
midnight ember
Left a message in their discord with a link to the github issue though he's tagged in github too now. Thanks for all your help with this one Dan. It's greatly appreciated.
edgy edge
how do you find Errors caught by the formatter on commit?It barfs on Neuton.c, but that file compiles and there is nothing obvious to the naked eye.```
sperok@Speros-Mac-mini nrf % git commit -m "Neuton parameter fix up"
Check Yaml...........................................(no files to check)Skipped
Fix End of Files.........................................................Passed
Trim Trailing Whitespace.................................................Passed
Translations.............................................................Passed
Formatting...............................................................Failed
- hook id: formatting
- files were modified by this hook
/Volumes/CPDev/circuitpython/tools/uncrustify.cfg:408: option 'sp_type_question' is deprecated; did you want to use 'sp_before_ptr_star' instead?
/Volumes/CPDev/circuitpython/tools/uncrustify.cfg:902: option 'sp_before_tr_emb_cmt' is deprecated; did you want to use 'sp_before_tr_cmt' instead?
/Volumes/CPDev/circuitpython/tools/uncrustify.cfg:905: option 'sp_num_before_tr_emb_cmt' is deprecated; did you want to use 'sp_num_before_tr_cmt' instead?
Option<NUM>: at /Volumes/CPDev/circuitpython/tools/uncrustify.cfg:1055: Expected number , for 'indent_shift'; got 'false'
/Volumes/CPDev/circuitpython/tools/uncrustify.cfg:1121: option 'indent_sing_line_comments' is deprecated; did you want to use 'indent_single_line_comments_before' instead?
Option<NUM>: at /Volumes/CPDev/circuitpython/tools/uncrustify.cfg:1202: Expected number , for 'indent_comma_paren'; got 'false'
Option<NUM>: at /Volumes/CPDev/circuitpython/tools/uncrustify.cfg:1206: Expected number , for 'indent_bool_paren'; got 'false'
Option<UNUM>: at /Volumes/CPDev/circuitpython/tools/uncrustify.cfg:2724: Expected unsigned number , for 'mod_full_brace_if_chain'; got 'false'
Option<NUM>: at /Volumes/CPDev/circuitpython/tools/uncrustify.cfg:2944: Expected number , for 'pp_indent_brace'; got 'true'
orchid basinBOT
tulip sleet
how do you find Errors caught by the formatter on commit?It barfs on Neuton.c, b...
Those are errors in the uncrustify.cfg, seems like you might be using a version different than what we are using.
we are using uncrustify 0.71.0_f or 0.72.0
what does uncrustify --version say?
edgy edge
Uncrustify-0.75.1_f
tulip sleet
roll back to an earlier version, then
thorny dove
It's a doc bug. The signature should read: `def request(self, *, timeout: float ...
don't you think the function signature for request() should read: 'def request(self, *, timeout: int = -1) -> I2CTargetRequest:' ? See the annotated screenshot of the part of the file I2CTarget.c
onyx hinge
don't you think the function signature for request() should read: 'def request(s...
Circuitpython always has floats enabled
thorny dove
Circuitpython always has floats enabled
OK, so the signature stays with 'float'
onyx hinge
thank you Danh!
proven garnet
I wrote a GitHub Acttion script to compile MPY files and create a ZIP file as part of a release process. It's similar to what the libraries do, but it's geared more for projects than libraries (I write CircuitPython code but want to get MPY files to actually put on the board). I'm happy to add it to under adafruit, but there's also the option to add it to the circuitpython org.
The major difference at that point is how they're called in a CI, whether it's adafruit/build-mpy@v1 or circuitpython/build-mpy@v1. Does one seem more intuitive or convey the purpose better? I guess circuitpython/build-mpy@v1 sounds less like an Adafruit internal tool?
For reference, here's that action: https://github.com/tekktrik/build-mpy
We could also add it to the GitHub Marketplace (free of course, I really think it just makes it searchable in that system in that case, but didn't know if there was any strong opinion either way or if anyone knows more)
And I guess I'm looking for feedback on the name - build-mpy? release-mpy?
onyx hinge
@proven garnet I am assuming it could potentially build the mpy files even if it was not related to a tagged release, so I'd say it's build-mpy
orchid basinBOT
proven garnet
Thanks! I think that makes sense.
manic glacierBOT
manic glacierBOT
manic glacierBOT
thorny dove
I am (still) busy trying to get I2CTarget working in a testscript. While studying the contents of the I2CTarget class inside the file CPY/shared_bindings/i2ctarget/I2CTarget.c, there is the function 'request'. In the docs text (also available in CPY readthedocs) the function definition defines parameter timeout and defaults it to -1. However in the function timeout is checked for a value of 0, while - looking at the function definition - timeout should be checked for a value of <= 0. The else if part checks for >0. Or do I misinterpret this code chunk?
In relation to the I2CTarget class. Where is taken care of the locking/unlocking of the I2C bus? Like in busio.I2C there are try_lock() and unlock(). Also bitbangio does have both lock() and unlock().
manic glacierBOT
blissful pollen
I am (still) busy trying to get I2CTarget working in a testscript. While studyin...
I would have to look at it closer but at first glance the timeout does look like an error.
Again off the top of my head but the bus locking is in shared-bindings/busio/I2C.c as you said with try_lock() and unlock().
stuck elbow
@onyx hinge (he/they)๏ปฟ perhaps you could reopen https://github.com/espressif/esp-idf/issues/6333 for visibility?
it doesn't look like they are going to reply otherwise
I mean it's 20 months...
onyx hinge
@stuck elbow no there's no reopen button for me
we basically threw up our hands and moved on from doing audio on espressif family chips, and haven't returned to it either
stuck elbow
@onyx hinge do you still have the code?
onyx hinge
@stuck elbow not that I can find right away
stuck elbow
oh well, thanks
thorny dove
@onyx hinge PR created. See: https://github.com/adafruit/circuitpython/pull/6985
thorny dove
I would have to look at it closer but at first glance the timeout does look like...
I just created a PR that also takes into account the range checking of the 'timeout' parameter of function 'request'
blissful pollen
I just created a PR that also takes into account the range checking of the 'time...
Great, and thanks! I'll take a look at the PR in more depth later but a quick glance it looks good (though I'm not sure on the .gitignore part)
proven garnet
Early hug report to @lone axle for the PR reviews today, I got behind on those type annotation ones ๐ตโ๐ซ
lone axle
Happy to get some reviews in. I've been focused more on single issues last few weeks and hadn't made a pass thu the open PRs list in a minute. It's good to get thru some today. We need more hours in the day!
Testing out your fix in logger next.
blissful pollen
Early hug report to <@382939733107408897> for the PR reviews today, I got behind...
I'll try to look at more of these too, they're usually not too hard. Quick question though (as I just saw this) for an init call would you have -> None as the return annotation? I just saw it in one PR but haven't in others.
proven garnet
Yeah, __init__() is called after actually creating the new object and only initializes it, so it doesn't actually return anything specific (hence None)
jaunty juniper
made clearer when you call super().__init__()
blissful pollen
Cool just wanted to make sure it makes sense to use it there and it wasn't a special case because it was __init__
lone axle
Yep, that was (and still is) one of the weirdest things to me, but is correct that __init__() will return None.
onyx hinge
speaking as a python language lawayer, __init__ doesn't have to return None. It just does so by convention. This very contrived example prints "7" by returning something other than None: ```class Base:
def init(self):
return 7
class Derived(Base):
def init(self):
self.attr = super().init()
print(Derived().attr)
I can't think of an actual GOOD use for this though
proven garnet
This is the kind of rules lawyering I live for. That's interesting, I never thought to actually store the results of super().__init__().
But to your point, why would you haha
digital shoreBOT
idle owl
<@330227457296957440> if you're around. No worries if not... Was your new HT16K3...
Yep, found it now. I had to release the library. I either forgot or was unaware.
lone axle
There is something odd going on where github is reporting the workflows as successfully but also cancelled. I made a guess at the root cause and put in a PR in the Bundle repo for it. But after some further research I'm thinking it's possible that it's something more complex than what I tried to address.
onyx hinge
@tulip sleet I was wondering if there was some way we could potentially combine blocks of doc from ports/.../common-hal next to the shared-bindings documentation, as part of the pyi extraction steps. I just don't know quite what it'd look like. Would you repeat the def ... exactly, and have the docstring be the added caveat? I might prototype this.
I don't think detecting that the implementation is going to throw NotImplemented is likely to work out super well
tulip sleet
there were cases where NotImplementedError was not callled directly, and there were also a few false-flag cases
I was just going to put something in shared-bindings. We have to update that anyway when we add new arguments, etc., so I think it's the same kind of maintenance
so if someone added a new capability or fixed the "not implemented", it's part of the job to check the doc
onyx hinge
@tulip sleet do you want to do the core section or shall I?
tulip sleet
i can do it this week, I've skipped a couple of weeks
onyx hinge
thanks!
tulip sleet
i can talk about beta.1
onyx hinge
yes please
you are much better informed than me on any given week
<@&356864093652516868> Hey! we're looking forward to chatting in about 25 minutes in the circuitpython voice channel! Please add your notes to the shared document: https://docs.google.com/document/d/1p3AthJvTLLbLhqsiL6_ZHBQkj7CUSQ_IOJdc9Dnu4Hw/edit?usp=sharing -- and if you just want to listen in, that's great too. See you in a bit
proven garnet
The easiest way to manage Hacktoberfest for the libraries is to add the Hacktoberfest topic to the repositories. I can easily add (and later remove) that using the gh CLI. This way reviewers don't need to worry about adding the hacktoberfest-accepted label. @idle owl does that sound good?
idle owl
The easiest way to manage Hacktoberfest for the libraries is to add the Hacktobe...
Yes please. And please update Adabot to do that automatically for the future.
proven garnet
You got it!
onyx hinge
@proven garnet also how much work would it be to apply the hacktoberfest labels on the core just like in the libraries?
from what dan said it sounded like maybe that doesn't happen
onyx hinge
thank you
errant grail
The easiest way to manage Hacktoberfest for the libraries is to add the Hacktobe...
Does hacktoberfest also include the Community Bundle? Iโve been spending a lot of time in there lately. ๐
idle owl
Does hacktoberfest also include the Community Bundle? Iโve been spending a lot o...
If it doesn't, we can make it happen.
lone axle
I think it's tied to issues / PRs so as long as those are created / labeled I think it should count. At minimum the PRs to add to the Community Bundle itself if labeled.
blissful pollen
Does hacktoberfest also include the Community Bundle? Iโve been spending a lot o...
If you're PRing to the repos included in the bundle you probably need those authors to tag it (though the merge upstream to the bundle itself could count too)
idle owl
If you're PRing to the repos included in the bundle you probably need those auth...
Oh right, good call. We can't make it happen.
midnight ember
I like how even documentation is included, encourages participation where ever you can.
onyx hinge
Ahhhhh I tried that Tandy keyboard for a few days but I'm so happy to be back on my usual tenkeyless.
the alt/ctrl key placement was just too weird
blissful pollen
I think the repo maintainers could tag it though... maybe
idle owl
I think the repo maintainers could tag it though... maybe
Yes, if the authors/hosts of the repos do the labeling/topic adding, it would count.
But that's up to the person whose account hosts the actual library code.
onyx hinge
stuff under https://github.com/circuitpython/ should be fair game too
idle owl
@lone axle Can you do a backup recording today?
lone axle
<@382939733107408897> Can you do a backup recording today?
Yep, am all set up for it.
onyx hinge
It's a good moment to mention that next week's meeting is on a Tuesday, due to a US holiday on Monday (Indigenous Peoples' Day)
midnight ember
the blog is amazing, my favorite thing to read everyday
proven garnet
Expect those to be assigned before tomorrow!
onyx hinge
@gilded cradle blinka is fair game for hacktoberfest pull requests right?
gilded cradle
onyx hinge
Yes, either the PR or the repo needs to be tagged "hacktoberfest" as I understand it
@turbid radish should we tag the circuitpython-weekly-newsletter as hacktoberfest, so that people who pull request news get credit for their contributions towards their hacktoberfest work?
turbid radish
No, I'd rather they be more substantive if that's ok
onyx hinge
that's fine too, I wanted to ask!
onyx hinge
[from djdevon3's updates]
midnight ember
glad to hear you made it through fine too Anne, and SLS too. ๐
thank you for that jepler, i always forget to do that stuff, multitasking not my forte
onyx hinge
Continue turning our own CI process into composite actions so adabot isnโt needed to update them all anymore
โค๏ธ
oh my creating a gist from circuitpython creates a lot of possibilities
lone axle
proven garnet
Yeah that sounds good, it'll work the same either way!
errant grail
Multiple instances for sample mixing is useful for music, too.
austere acorn
Just add in big red letters then, in all guides
that only one instance should be created ๐
onyx hinge
@austere acorn thank you for the question though!
austere acorn
there are boards for music, that do have that much amount of flash and ram
but for "regular" use it should be made clear
that multiple instances kill a project
also, that issue can probably be closed then
cos there is no solution
(sorry, @idle owl, i'm done :D)
onyx hinge
this space and time, and especially in the weeds, is for anyone in the community ๐
midnight ember
you bring up a good point for trying to streamline but audio mixer guys it's a touchy subject of possibly taking away a future hope/dream/tool ๐
austere acorn
@midnight ember this is the board for music ๐ https://www.electro-smith.com/daisy/daisy
onyx hinge
@austere acorn I'll write a few words on that PR and close it. Thanks for putting it In The Weeds because I was not concentrating on it; this made me think it through, which just having the issue hanging there might not have done.
๐
austere acorn
64MB of SDRAM for up to 10 minute long audio buffers, 8MB external flash + sd card
proven garnet
Thanks!
lone axle
Thanks everyone, have a good day ๐
errant grail
Thank you!
tidal kiln
nope. wrong button ๐
unique python
thanks!
proven garnet
Also I think I need expanded permission to edit topics in the libraries
proven garnet
Either that can happen or if anyone has that and wants to run the script, that would work too
onyx hinge
thanks kattni!
midnight ember
does it run circuit python though?
idle owl
You are all quite welcome!
austere acorn
does it run circuit python though?
not YET ๐
idle owl
Either that can happen or if anyone has that and wants to run the script, that w...
Um.... I'll think on that.
midnight ember
ohh it's an M7 ๐คค
idle owl
@lone axle Checking recording now to see if I need yours.
proven garnet
Sounds good, if you can selectively choose permissions it should only be the topics permission or something like that. I'm not sure how it's split up.
onyx hinge
@lone axle good "catch" that the actions happen in forks, and thinking about the consequences.
idle owl
Does anyone know how to trim a few seconds off the beginning of a mkv file?
austere acorn
<@196904073255976961> I'll write a few words on that PR and close it. Thanks for...
Sure, no problem, I was investigating it since I started with the Mp3Decoder issues, I wanted to check if I could help further
idle owl
It's not necessary but would be nice ๐
uncut nexus
Is anyone else having problems building the current main for Pico-W? I am getting an error in the pioasm module where cmake fails trying to build the C test (undefined reference to `_exit')
onyx hinge
@idle owl I know it's possible with ffmpeg but I have to spend 10 minutes with the documentation to do even simple things
lone axle
I've used an application called OpenShot video editor to make simialr small changes to videos. Not sure if it's available outside of linux though.
tulip sleet
i have used vlc to do that
onyx hinge
@uncut nexus that's odd. we're not seeing it when we build the core. what system are you on (mac/win/linux)? or ci builds on linux of course
idle owl
I've used an application called OpenShot video editor to make simialr small chan...
It is!
idle owl
i have used vlc to do that
Wait, really?
lone axle
If it's quick / easy enough for you to upload it and share a link I can try to trim it and send back.
tulip sleet
i think so, i don't remember how
https://www.vlchelp.com/cut-trim-videos-with-vlc-media-player/ ugh, it's playing and recording, never mind
midnight ember
does the output format have to remain mkv?
tulip sleet
ugh, that's a bad way, let me find another
midnight ember
yeah vlc is not good for editing
uncut nexus
@onyx hinge - using Linux (ubuntu 20.04) I've even tried spinning up a fresh VM and a fresh CP clone.
@idle owl - I think Audacity can handle mkv files (as imports)
onyx hinge
@uncut nexus OK .. can you go ahead and do a "make BOARD=raspberry_pi_pico_w clean" and then a "make -j1 BOARD=raspberry_pi_pico_w" and then show the full output?
tulip sleet
idle owl
Oh fair enough.
onyx hinge
our actions also uses ubuntu 20.04 as its linux flavor
idle owl
Nice, I'll see if the file uploads. Thanks Dan!
Unsupported media
I'll try this OpenShot editor.
uncut nexus
@onyx hinge - had to put is as a file; too long for chat
midnight ember
i just used clideo.com editor today, just searching for a free online editor. worked fine, adds a watermark though and unsure if it'll export to mkv.
uncut nexus
If I edit build-raspberry_pi_pico_w/pioasm/CMakeCache.txt and add "CMAKE_C_FLAGS:STRING=--specs=nosys.specs" then it works
onyx hinge
@uncut nexus that's odd! It's trying to use the embedded C copiler arm-non-eabi-gcc to build this program, which needs to run on the host computer. This could happen if your environment has the CC variable set to /nvme/circuitpython/gcc-arm/bin/arm-none-eabi-gcc, or at least that's the first possible cause I thought of.
The succeeding build says: -- Check for working C compiler: /usr/bin/cc - skipped while yours got a different value. Once you figure out WHY it got that different value, you'll be on track to resolve the problem.
uncut nexus
I do have that set... building has always worked OK that way. Let me unset and see what happens
onyx hinge
I think the recomendation is to ensure that the location with the embedded C compiler is on your PATH, not to specify it in the CC environment variable.
uncut nexus
@onyx hinge - unsetting CC worked - thanks!!
onyx hinge
OK! I'm glad it was something simple, even if not evident
uncut nexus
I was setting CC in a setup script for building CP - will remove it.
thorny dove
@onyx hinge the PR is ready to be reviewed
manic glacierBOT
midnight ember
Just listened to the weekly meeting on YouTube and my mic sounded horrible. Audio was coming from my new webcam 5 feet away I used for the first time this morning instead of the headset. ๐คฆโโ๏ธ sorry about that
idle owl
hacktoberfest topic added to all CircuitPython libs!
tulip sleet
@onyx hinge .env syntax is kinda ridiculous.
k1=#
k2=##
k3= #
k4=# #
oh no are you looking at that issue now?
I can't even guess at the content of those variables
tulip sleet
k1 is #; k2 is ##; k3 is the empty string, and k4 is #
that's what bash does
so they are attempting to copy that. It's kind of execrable
it's a comment only if it has preceding whitespace
onyx hinge
oh yes when I put on my shell glasses it makes "perfect sense"
tulip sleet
right, breaking up into words or something
onyx hinge
$ k5=x#y; echo $k5
x#y
oof
at first I was worried it was counting to the last '#' or something
tulip sleet
i half-fixed the issue, including the # inside single quotes
unichar_isspace() is true for \n; maybe Scott didn't realize that. That was one fix, and then I did a better check for # inside quotes
but I haven't yet fixed abc=#
onyx hinge
if abc='#' works I think it's sufficient
or at least a good improvement over what's there
tulip sleet
I think so too. it's a morass of loops and flags
onyx hinge
there might still not be a way to include a literal ' ?
tulip sleet
i didn't try that yet
idle owl
@onyx hinge Is there any PiCow docs on using the port? I know there's no guide or anything, but did you doc it within the core?
onyx hinge
@idle owl I didn't write any specific documentation to it yet. The response from Anecdata on https://github.com/adafruit/circuitpython/issues/6558#issuecomment-1266078988 is spot on though. The process of connecting & using requests or socketpool SHOULD exactly match esp32-s2 except where functionality isn't implemented yet
idle owl
Keen, thank you.
tulip sleet
@onyx hinge one final haha:
in bash abc='def'#hi means abc is def#hi, but in python dotenv, it is def
:throws up hands:
onyx hinge
yeah abc=d'e'f"g"h is "perfectly good" shell ๐ฌ
idle owl
@tulip sleet Hey so... for the deep sleep thing. You said it only drops to 75uA if you do not "hold" any pins. Does that apply to shutting off the red LED, NeoPixel and/or I2C power pin? I guess I don't understand what "hold" means.
tulip sleet
https://github.com/theskumar/python-dotenv#file-format "The format is not formally specified and still improves over time."
onyx hinge
like a fine wine
tulip sleet
<@329766224093249548> Hey so... for the deep sleep thing. You said it only drops...
I mean if you don't use preserve_dios in exit_and_deep_sleep_until_alarms()
idle owl
I mean if you don't use `preserve_dios` in `exit_and_deep_sleep_until_alarms()`
Ok, so changing pins in your code.py is not even close to what you mean. Got it!
tulip sleet
on the ESP32 V2, If I don't fiddle with any pins, the "right thing" happens and it's 75uA
that is not true on the TFT boards
idle owl
Ok, thanks. Wanted to make sure before testing it.
Hmm.
Weird.
Or maybe it's not weird. I wouldn't know.
tulip sleet
I mean, you can set the pins to anything you want, and when you go into deep sleep, they get reset to a reasonable value, kind of by accident, on the ESP32 V2
assuming you don't use preserve_dios
idle owl
Ok got it
warm stump
I've just been trying to get time.localtime() to set the rtc on a macropad with unexpectedly mixed results,
most importantly setting the rtc with a tuple of 9 zeros can result in the board failing its next boot.
workaround: switch to the minimal rp2040 image on the macropad with the uf2 bootloader that still works
which will be enough to back up your code.py and remove it so you can switch back to the usual image.
That whole process and more about the USB Cable Time Protocol, I have documented here:
https://github.com/TheMindVirus/macropad/blob/archive/sketches/CableTimeProtocol/README.md
jaunty juniper
I've just been trying to get `time.localtime()` to set the rtc on a macropad wit...
have you tried safe mode ?
warm stump
have you tried safe mode ?
I saw what safe mode looked like, it didn't seem to enter it, it just locked up instead, on circuitpython 7 and 8 but I had cp6 for the rp2040 with no peripherals so that was safe mode in this case
depending on where this happens, it either will or won't be able to action the switch to safe mode at the yellow LED blink, I think mine reached printing "code.py output" and everything before it
but also any C which is dependent on the RTC at that stage may get confused by an invalid value, mostly related to the rtc and time circuitpython modules
alpine rapids
Hi! long time, first time, etc. I'm working on a PR for a variant of the esp32-s2-devkitc-1 that hasn't been added yet, and I just wanna make sure I'm editing the ci_check_duplicate_usb_vid_pid.py file correctly. I'm on Windows, so I plugged the board into the UART micro usb on the board (and into my computer), and this is what shows up - do I need to add a new entry for this VID/PID, because this combination isn't in that file yet
thorny dove
Question: After a successful merge of a PR yesterday, I downloaded and flashed the 'latest' for an Adafruit PyPortal Titano and an UM PROS3 with the intention to try the changes. The boot_out.txt of the PROS3 shows ยซAdafruit CircuitPython 8.0.0-beta.1-4-g1fd09cb6d on 2022-10-04; ProS3 with ESP32S3' while boot_out.txt on the Titano seems not to be 'latest': Adafruit CircuitPython 8.0.0-beta.0-49-g14fc4a079 on 2022-09-14; Adafruit PyPortal Titano with samd51j20'. How come?
tulip sleet
Question: After a successful merge of a PR yesterday, I downloaded and flashed t...
Try looking at https://adafruit-circuit-python.s3.amazonaws.com/index.html?prefix=bin/pyportal_titano/en_US/ again. The build you are listing is way down the list (20220914). What is the name of the UF2 you loaded on the Titano?
tulip sleet
maybe try loading again
onyx hinge
Thank you for contacting GitHub Support and sorry for the time taken to get back to you, we have had quite a backlog.
I got a reply on the github support ticket I filed about the cancelled jobs. No actual update.
thorny dove
Try looking at <https://adafruit-circuit-python.s3.amazonaws.com/index.html?pref...
Excuse me I made an error! Flashed a wrong one. Thank you for your advice!
tulip sleet
np ๐
crimson ferry
Hadn't seen this before beta-1, but I didn't think anything related had changed for espressif:
On reload of Adafruit CircuitPython 8.0.0-beta.1 on 2022-10-01; Adafruit QT Py ESP32S2 with ESP32S2 (also replicated on Feather ESP32-S2 TFT):
import time
import traceback
import microcontroller
import wifi
from secrets import secrets
time.sleep(5) #wait for serial
print(f"{"*"*40}")
errs = 0
while not wifi.radio.ipv4_address:
try:
wifi.radio.connect(secrets['ssid'], secrets['password'])
time.sleep(5)
except ConnectionError as e:
# ConnectionError: No network with that ssid
# ConnectionError: Unknown failure 2
traceback.print_exception(e, e, e.__traceback__)
errs += 1
if errs > 8:
microcontroller.reset()
while True:
print(f".", end="")
Result:
ConnectionError: No network with that ssid (usually)
or
ConnectionError: Unknown failure 2 (once)
Fixed by microcontroller.reset() but not by reload.
Web Workflow is not enabled.
onyx hinge
@tulip sleet if you can take a look at ^^^ if you don't mind, I'll just drop test case "e15" as unsupported in our version and mark it ready for review.
jaunty juniper
Hadn't seen this before beta-1, but I didn't think anything related had changed ...
that looks like https://github.com/adafruit/circuitpython/issues/6866
tulip sleet
<@329766224093249548> if you can take a look at ^^^ if you don't mind, I'll just...
I think that is fine. I think I can fix that, but can do that in another PR.
jaunty juniper
we've been having that reload instability for wifi for a little while now, mitigated by enabling auto connect as I assume it keeps the connection during reloads so there's no reconnect to fail
crimson ferry
that looks like <https://github.com/adafruit/circuitpython/issues/6866>
yes, I had seen it with scanning, but not connection. Now it's consistent with connection, on several boards, but again only when web workflow is Off. With web workflow on, like scanning, connection behaves fine.
Append to same issue, or separate issue?
crimson ferry
we've been having that reload instability for wifi for a little while now, mitig...
even doing py wifi.radio.stop_station() wifi.radio.enabled = False wifi.radio.enabled = True wifi.radio.start_station() which is the closest thing we have to deinit, allows connection to work on reload with web workflow on, so I think it's a little deeper than the autoconnect
jaunty juniper
even doing ```py
wifi.radio.stop_station()
wifi.radio.enabled = False
wifi.radio...
I think at this point we need somebody to dive in the code, look at the changes made in the IDF in PR 6531, run a debug build and such ๐คท
crimson ferry
disabling the radio before connection also makes it work without web workflow
wifi.radio.enabled = False
wifi.radio.enabled = True
```fixes scanning too when web workflow is not enabledsomething is not de-initing or initing right in the non-web-workflow path
I'll add this to the issue
proven garnet
Late hug report to @idle owl for helping add the hacktoberfest topic to the libraries!
Also I don't have topic access, so I know I said I'd add it to the core repo but I actually can't
idle owl
Late hug report to <@330227457296957440> for helping add the hacktoberfest topic...
Hug to you for the script!
idle owl
Also I don't have topic access, so I know I said I'd add it to the core repo but...
I'll do it.
Oh someone else did.
onyx hinge
def print_string_foolishly(s):
for c in s:
# status display update could occur between writing any two characters
# wrecking the escape code we're trying to write in a foolish way
print(end=c)
print_string_foolishly("\033[H\033[2H") # clear screen code
```am I wrong about this being a potential problem?I guess anything sending escape codes itself should just disable the status line
idle owl
@proven garnet There's a step for setting up library repos that you're missing or maybe can't do anyway. We need to find out if you can, and if not, add it to the things you ask me to do in the end before closing the setup issue.
tulip sleet
```py
def print_string_foolishly(s):
for c in s:
# status display up...
yes, it could be a problem. These in-band status things are ... not great. Yet another reason to be able to disable status bars, which is now available.
random junco
I don't know if it's a bug, but Hacktoberfest is listed twice in the sort issues drop down on circuitpython.org, not sure it's related
idle owl
Well, there's that at least. Um.... file an issue on the circuitpython-org repo about it.
Not sure if it's an Adabot or -org issue.
idle owl
Wait.... isn't time.monotonic() in ms?
This "fractional seconds" crap is useless to me, in the first doc that popped up.
jaunty juniper
no, seconds
idle owl
Ok. Got it.
jaunty juniper
yes, it could be a problem. These in-band status things are ... not great. Yet a...
TBH I am not confortable with the status bar being sent outside of the REPL (or "Code done running." state) by default, even if it can be disabled. (Not to mention that you can't easily enable the Data CDC on ESP)
stuck elbow
there is ticks_ms somewhere too
in microcontroller or in supervisor or something like that
tulip sleet
TBH I am not confortable with the status bar being sent outside of the REPL (or ...
The disabling is sticky across VM restarts. The primary reason for doing it outside is I think to show status when the web workflow is starting up, since there can be a significant delay when that is happening. But feel free to open an issue about this.
idle owl
@tulip sleet I'm testing this. https://github.com/adafruit/circuitpython/issues/6676 They ran into the same thing I did with the value parameter on PinAlarm. It cannot be set to True apparently. Nearly everything was not working properly - the TimeAlarm did not execute, the Pretending to deep sleep until alarm message did not show up, there was no way to interact with the board through the serial console (ctrl+C/D did nothing), and if I made changes to the code and saved, I had to press the reset button on the board to make the changes happen. THEN.... I changed the PinAlarm parameter to value=False, and suddenly everything worked right again. So this is not a bug with running Time and Pin alarms in conjunction, it's a bug with being able to set a PinAlarm to trigger when high. I ran into this with my original LoRa Mailbox project, and was told I must use a NC reed switch and if not switch.value because that wasn't "how PinAlarm worked."
So at this point, we either need to document that it can't be used that way, or figure out the reason and fix it.
Unless I'm missing something.
I can put all of the facts in the issue, but I don't want to issue the "solution" if I'm wrong and completely missing something.
tulip sleet
<@329766224093249548> I'm testing this. https://github.com/adafruit/circuitpytho...
different pins have different requirements about when they trigger. Triggering when brought low is typical; which pin on which board is what to know.
idle owl
different pins have different requirements about when they trigger. Triggering w...
Ok....... it's D12 on a Feather ESP32-S2. How the heck is anyone supposed to know which pin can do what?
tulip sleet
well, we need to document it, if that's the case, and maybe it's a bug that we thought it would work rising, not falling. Is that the only pin specified? Sometimes it's "one pin can be either" if only one pin, but for two pins, it might be "falling only"
it could be our misinterpretation of the doc, or it could be a coding bug
idle owl
What do you mean the "only pin specified"?
tulip sleet
Anyway, a really simple example with D12 in an issue would be great, and I will test it.
You can specify more than one pin alarm at once
idle owl
No, it's only one.
One pin, one time.
Let me simplify this further, and I'll send it to you.
@tulip sleet Do you want the time alarm stuff in it at all? It does all the same weird stuff without it.
tulip sleet
nope, simplest possible
proven garnet
<@135415127879974912> There's a step for setting up library repos that you're mi...
Do you mean the transferring?
idle owl
Do you mean the transferring?
No, I mean giving the CircuitPythonLibrarians team write access.
proven garnet
Ohhhh gotchya
I typically don't have access to the Settings tab in a repo, if that's where you do that
idle owl
@tulip sleet
idle owl
I typically don't have access to the Settings tab in a repo, if that's where you...
That is where you do that, and it requires further authentication inside that. So, when you ask me to add the webhook, please remind me to update the settings, and add the CPLibians with write access. Thank you!
tulip sleet
๐ and just specify the board in an issue, can be a very short issue. There are no external pullups/pulldowns on the pin, is that right?
idle owl
๐ and just specify the board in an issue, can be a very short issue. There are ...
No, I literally used a jumper wire to tie it to GND or 3V
There's nothing else.
Create a new issue?
idle owl
No, it starts off nothing and then I try to connect it to the opposite pin
tulip sleet
try starting off low and then bring it high
tulip sleet
because it was already kinda high, maybe
proven garnet
That is where you do that, and it requires further authentication inside that. S...
you got it!
idle owl
Right now, it's showing the "pretending to deep sleep" message even
Which it doesn't do on save otherwise.
tulip sleet
set pull=True when creating the alarm, that will set a pull opposite from what value= is
idle owl
That is so confusing.
tulip sleet
also edge defaults to false,
idle owl
I've never set edge
tulip sleet
I would set edge to True, since you are looking for a transition from low to high. But i am not sure if edge is available on ESP32-S2
idle owl
File "code.py", line 7, in <module>
ValueError: Cannot wake on pin edge. Only level.```ok, then just set pull
tulip sleet
value and pull to True?
sorry, sudden dishwasher repair diagnosis needed. Yes, value and pull to true, and then move jumper to +3v to trigger. If that doesn't work try value=False, pull=True, and move jumper to ground.
idle owl
sorry, sudden dishwasher repair diagnosis needed. Yes, value and pull to true, a...
Sorry I didn't expect this to be more instruction, so I filed the issue.
I closed the other issue as well.
Assigned the PinAlarm one to you, @tulip sleet - you can unassign it if that wasn't the plan.
tulip sleet
i'll try it and see what's going on, might be a doc issue if there are restrictions on what the pin can do.
idle owl
i'll try it and see what's going on, might be a doc issue if there are restricti...
Right, that's what I was saying earlier.... maybe we need to document it better, or maybe it needs a fix.
proven garnet
@lone axle https://github.com/adafruit/Adafruit_CircuitPython_DisplayIO_FlipClock was going to get a release after a PR merge, but doesn't have one yet. Want me to release it at 1.0.0?
lone axle
<@382939733107408897> https://github.com/adafruit/Adafruit_CircuitPython_Display...
Yes please, and thank you!
midnight ember
```py
wifi.radio.enabled = False
wifi.radio.enabled = True
```fi...
oh thank god, that issue is driving me insane. finally someone was able to confirm it's not just me.
happens on my UM featherS3 too
but not on my QT Py S3
crimson ferry
I had never seen it and I've been trying to figure out why... probably because I only do a scan at startup in some code, but it's rote and I don't usually watch that, and; I almost alwatys disconnect wifi after I use it, then reconnect when I need it again
midnight ember
maybe worth looking into the qt py to see what the difference is? qt py works great for me.
crimson ferry
but not on my QT Py S3
I've focused on S2 for now, since we know that to be more stable... once the S2 quirks are well understood, then why S3 may behave differently becomes an easier question hopefully?
it is weird for different boards with the same basic chip to behave differently in fundamental ways
generally, when I get something working on one S2, I have high confidence it will work on all S2s
midnight ember
well it happens on both my S2/S3 so the qt py s3 is the only espressif board i have that works right. figured maybe it could be relevant. i'm totally fine if you can fix the S2 first... i just need one of those boards to work right don't care which.
crimson ferry
yes I do hope someone can narrow down and fix 6866
midnight ember
sounds good. i'm just relieved someone else was able to see it exists. everything was pointing at my PC and not the boards and i was coming up completely empty and frustrated. this is a HUGE weight off my chest just to know it's a real bug.
crimson ferry
I think you raise several distinct issues in 6791, the CP8 quirks with non-web-workflow only explains a couple of them
midnight ember
yeah the hard fault is a separate leftfield issue. the more pressing issue for CP in general is that hard reset bug.
but combine both the hard reset bug and hard fault bug for my UM FeatherS3 and it's practically unusable. ๐ฆ just the way it goes sometimes.
crimson ferry
I'll try some S3s for grins
I think all I have is Espressif DevKits and UM boards
oh, and Adafruit Feather S3 TFT ๐
(I need my PSRAM, so I didn't jump on the S3 QT Py yet)
crimson ferry
well, got unexpected results on S3 so went back to S2 and now I can't even fully reproduce what I did an hour ago. So. Many. Variables. Web Workflow is a cure for many ills. Not so much for disabling wifi.
we need a diff for simulating executing one code path in the core vs another
ember iris
well, got unexpected results on S3 so went back to S2 and now I can't even fully...
Well you're not alone. Using a metro Esp 32s2 I was on the nightly from sept11, tested the wifi radio enable toggle and it worked. I reliably got the same number of wifi access points as I did from a hard reset. There were some ap's that varied but they were the ones with a weak signal and they were inconsistently in the hard reset scan so it just seemed like it was consistently difficult to see signal.
But then I wanted to make sure the 8.1 beta behaved the same, so I upgraded to that, and couldn't get the scan working off of soft resets even with wifi.radio.enable toggled between false and true. So then I erased flash and tried the beta, and toggling the wifi and it didn't work. So now I've erased flash and went back to the sept11 nightly and have the toggle in and it's not working any more.
So I can't duplicate what I was doing before with the same code on the same version of circuit python.
I think I'm going to try the 7.3 release, not erase flash, 8.1 release, and the toggle and see if that works, but it's really me grasping at straws trying to figure out how to get back to the wifi.radio.enable toggle working
ember iris
Well you're not alone. Using a metro Esp 32s2 I was on the nightly from sept11, ...
Well now that didn't work either. I have no idea how to get it back into a state where the wifi.radio.enable being set to false then true gets the scan working again.
tulip sleet
@crimson ferry @ember iris I took a brief look at the wifi init code in 7.3.3 vs 8.0.0-beta.1, and I nothing missing stands out immediately. A couple of things are done only once, ever (even between soft restarts), but that is true in both versions. I'll look at this more in the near future.
crimson ferry
thanks @tulip sleet, this one is having me question the nature of my reality ๐
jaunty juniper
I did a bisect that landed on an update to IDF
crimson ferry
yeah, and there were a half dozen CP changes along with it, but none that seemd relevant
tulip sleet
I missed that you did a bisect. That says a lot.
jaunty juniper
crimson ferry
Well now that didn't work either. I have no idea how to get it back into a state...
same
ember iris
<@360465312518504448> <@459119350851567626> I took a brief look at the wifi init...
Thank you, the other bit I think we've confirmed is that in the circuit python 8 builds, each soft reset uses up a bit of the ram set aside for initializaiton and after a number of loops there's a memory error as a result. @crimson ferry On that note, while I can't make this happen again, when the wifi.radio.enable toggle was working, the code still ran out of memory eventually. That was the initial reason I moved from the sept11 nightly to the 8.1 beta--I didn't want to say it was still an issue if it had been resolved.
crimson ferry
@ember iris gotcha, I didn't run it long enough to see the memory issues... CP heap (gc) and not esp-idf?
ember iris
<@459119350851567626> gotcha, I didn't run it long enough to see the memory issu...
It was the esp-idf heap that ran out, the CP gc.mem_free() showed the same amount the whole way through
crimson ferry
ok, leak in the idf, that's interesting
ember iris
Yup, at around 32 bytes each soft reset, but it does vary to 30 and 34 and more/less some times
random junco
I'm trying to add a library to the CircuitPython bundle. In the Learn Guide (https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/sharing-in-the-community-bundle) it says And then we'll verify it worked by looking at the build log and versions file. - where / how do I do that? I'm not understanding the included screenshot. Do I have to build something? (This is the library /issue: https://github.com/adafruit/Adafruit_CircuitPython_BLE_File_Transfer/issues/15)
blissful pollen
I'm trying to add a library to the CircuitPython bundle. In the Learn Guide (ht...
Are you trying to add to the Adafruit CP bundle (which I'm assuming as that library is an Adafruit one). That guide is to add to the community bundle I belive
jaunty juniper
you can skip that, it's possible to build the bundle locally to test, but the Adafruit bundle is not setup like the community bundle is
the CI will do the test
random junco
Yes - Tekktrik linked to the Community Bundle learn guide in that issue, so I was guessing the process was the same
jaunty juniper
the guide mixes things, it references docs/drivers.rst which is for the Adafruit bundle
random junco
so I did the git submodule add, now I'm looking at the drivers.rst file
proven garnet
Oh wow that's oversight on my part
I should have specified the Adafruit CircuitPython Bundle
I didn't see that the guide referenced the Community Bundle
Sorry!
jaunty juniper
is there a guide for the Adafruit Bundle ?
random junco
Sorry!
Just my first time, wanted to make sure I get it right. :). CI is building it now, we'll find out
proven garnet
I don't believe so, that's the general guide for add to bundles in general, or just general library creation really
jaunty juniper
I don't think there is ? It was mentioned that the guide should be updated, or the bundles should change structure to match with each other
random junco
It passed CI if you want to take a look to see if I got it right: https://github.com/adafruit/Adafruit_CircuitPython_Bundle/pull/406
proven garnet
@jaunty juniper you look at them often but I'm free if you want me to take a look
jaunty juniper
maybe have the BLE libraries in alphabetical order ?
random junco
didn't even think of that, good idea!
I think I'll leave Bluefruit at the top though, probably one of the more popular?
proven garnet
~~You're welcome ~~ sorry ๐
manic glacierBOT
dusk gull
I want to quickly test RAM benefits of freezing some python code in firmware on a custom board. I see there is not much in the way of suggestions including my own code (shows suggestions for using existing Circuitpython libraries)
Should I just use the suggestions methods from Micropython, or is there a recommended method for this. Ideally I keep these with my board folder that holds my custom board information.
For example using the manifest file
https://docs.micropython.org/en/latest/reference/manifest.html
analog bridge
๐ฅณ #7000 is here ๐
proven garnet
Is anybody else seeing GitHub Actions not running when it otherwise should be?
tulip sleet
@onyx hinge @analog bridge Something is up with GitHub Actions. The two most recent PR's have not run anything.
githubstatus.com reports problems that are being fixed
analog bridge
GitHub Actions is down
tulip sleet
GitHub Actions is down
do you have info other than githubstatus.com ? (I should look at twitter or something)
analog bridge
checked it on githubstatus.com only... earlier webhooks was also down
I believe they will eventually catch up, on my cp fork... workflows from an hour ago have just now started showing up and are queued to run.
tulip sleet
https://github.com/adafruit/circuitpython/pull/6999 is green to merge, but there were no runs. That is weird. Maybe the Actions are supposed to intervene and unset the "ok to merge" status
tulip sleet
i removed my review so it wouldn't be merged yet
analog bridge
it may require git commit --amend --no-edit and a force push
tulip sleet
i will try that
manic glacierBOT
Already built it.
Ljinux had to be severly trimmed to fit on this tiney tiny storage.
I will now try outside of my driver and ljinux.
edgy edge
how do I set VSCode to stop fighting uncrustify with formatting? Please tell me answer is not" read uncrustify.CFG" and have VS code Match settings ...
tulip sleet
how do I set VSCode to stop fighting uncrustify with formatting? Please tell ...
how much difference is there? You can always let uncrustify clean things up. If it's something big like 2 vs 4-space indentation, yeah, you should try to match that.
I am using emacs with a particular formatting scheme, with only a few tweaks, and uncrustify complains very little.
;;; Indent curly braces on their own lines to line up under "if", etc.
;;; https://www.emacswiki.org/emacs/IndentingC
(setq c-default-style "k&r")
;;; Indent case inside switch.
;;; https://www.gnu.org/software/emacs/manual/html_node/efaq/Indenting-switch-statements.html
;;; //, not /* */ comments.
(add-hook 'c-mode-hook
(lambda ()
(c-set-offset 'case-label '+)
(c-set-offset 'arglist-cont-nonempty 4)
(setq comment-start "//"
comment-end "")))
;;; Indent case inside switch.
;;; https://www.gnu.org/software/emacs/manual/html_node/efaq/Indenting-switch-statements.html
(add-hook 'c++-mode-hook
(lambda ()
(c-set-offset 'case-label '+)))
;;; Indent preprocessor directives to match MicroPython style
(c-set-offset (quote cpp-macro) 0 nil)
;;; spaces instead of tabs
(setq-default indent-tabs-mode nil)
every commit fails formatting. It's easy enough to let uncrustify do its thing, then restage the changes. But it would be nicer if VS Code didn't mess everything up. At a minimum brace handling is incorrect. both formatters have a ton of options, I was hoping somebody here Will already have done the work for code like you have for emacs.
tulip sleet
every commit fails formatting. It's easy enough to let uncrustify do its thing, ...
https://stackoverflow.com/questions/45823734/visual-studio-code-formatting-for for example, but you can probably find more. I was websearching for << vscode k&r >>
basically k&r style with tweaks
hmm, you can even use uncrustify inside vscode: https://www.sfdcnotes.com/2019/12/18/vs-code-extension-uncrustify-code-formatter/<>
onyx hinge
@tulip sleet are there any special consequences when renaming a submodule? I want to use nina-fw for certificates with picow
but the current location is under ports/espressif
edgy edge
thanks , i'll give them a try. No no need to keep things in sync if I only use uncrustify
tulip sleet
<@329766224093249548> are there any special consequences when _renaming_ a submo...
you mean you want to move it from espressif/certificates?
onyx hinge
right, to .. lib/certificates maybe?
tulip sleet
that sounds fine, though is there some circumstance where we might want version skew for that between the two ports (I can't think of one).
onyx hinge
hopefully there wouldn't be
tulip sleet
I think moving to lib/certificates is fine. I -have- found that moving or renaming submodules can confuse my local git something bad, and I end up having to re-clone. There are various webpages about how to do it safely, seems more complicated than I want, and they might be outdated.