#circuitpython-dev
manic glacierBOT
slender iron
@tulip sleet at my comp now
slender iron
@prime flower so you are going?
tulip sleet
@slender iron if you have time, now's good to chat
slender iron
yup, good for me
prime flower
@slender iron I registered using the link and intend to attend
spice crypt
Is there still a way to turn on verbose logging to the REPL? It seems like notro was able to in this issue, but I don't see the command. Is it a special build flag? Only results I'm finding online are for SWD debugging
manic glacierBOT
obsidian compass
Quick question, is there support for USB HID gamepad in CP?
slender iron
@spice crypt I don't know what you mean by verbose logging
@obsidian compass yes but the gamepad descriptors that work with a host can vary
obsidian compass
I am looking for Linux
slender iron
@tulip sleet do we have a learn guide for usb gamepad?
spice crypt
@spice crypt I don't know what you mean by verbose logging
@slender iron output like this assign byte code: code=20002a40 len=27 flags=0
assign byte code: code=20002b20 len=44 flags=0
assign byte code: code=20002b60 len=17 flags=0
make_function_from_raw_code 20002990
mp_obj_new_fun_bc: 20002960
make_function_from_raw_code 2002d970
mp_obj_new_fun_bc: 200028f0
make_function_from_raw_code 20002a90
mp_obj_new_fun_bc: 20002990
idle owl
@tulip sleet Here is the board check I wrote for the Circuit Playground Library. python @staticmethod def circuit_playground_type(is_type=None): if is_type is not None and is_type not in ("Bluefruit", "Express"): raise ValueError("Valid Circuit Playground boards are Bluefruit and Express.") cp_type = None if "Bluefruit" in os.uname().machine: cp_type = "Bluefruit" if "Express" in os.uname().machine: cp_type = "Express" if is_type is not None: return is_type == cp_type return cp_type
Code looks like this: ```python
from adafruit_circuitplayground import cp
if cp.circuit_playground_type("Bluefruit"):
print(cp.circuit_playground_type())
Easily expandable to add new boards as they come.
might rename is_type but otherwise I'm happy with it.
tulip sleet
@slender iron @obsidian compass There's no HID gamepad Learn Guide; sorry; I don't think it worked well or at all on Linux; it MIGHT have worked with the browser support on Linxu
@idle owl I think it would be better not to have it do the comparison itself if the arg is not None. That makes the function do two pretty different things depending on the argument. How about:
from adafruit_circuitplayground import cp
if cp.circuit_playground_type() == "Bluefruit":
# whatever
having it return either a string or a boolean is highly unusual
idle owl
I wanted to force it to require "Express" or "Bluefruit"
So I made it more complicated to require one of those.
So you could put in if cp.circuit_playground_type() == "foo" and it will allow whatever you want. So if you typo'd Express or Bluefruit in your code, you might never know that you're not validating properly because the code will run.
And attempt to work on both boards.
obsidian compass
@slender iron @obsidian compass There's no HID gamepad Learn Guide; sorry; I don't think it worked well or at all on Linux; it MIGHT have worked with the browser support on Linxu
@tulip sleet thanks for info, is there anything I can do to bring it to functioning state?
tulip sleet
ok, so how about having two routines: circuit_playground_is_type(something) and .circuit_playground_type as a proprty (or a no-arg function). What I'm saying is that the API style you have now is highly unusual and potentially confusing
@obsidian compass there was no gamepad description that worked across all of MacOS, Linux, and Windows, so I had to do two out of three. In addition the Linux support seemed flaky in general
or you could have circuit_playground.EXPRESS and circuit_playground.BLUEFRUIT as predefined string constants (or integers!) that the user would compare against.
tulip sleet
sounds ok
@obsidian compass issues were range of joysticks (8 vs 16 bit) and maybe signed vs unsigned
it was kind of a mess; support is poor and there were few hardware examples of non-proprietary game controllers that would work without installing drivers
@ionic elk in general for re patterns, you should always use raw strings, because you might edit an existing one and forget to put the r prefix on when it becomes necessary
a raw string is just a string that's parsed on input differently; it is not a different data type
idle owl
@tulip sleet ```python
@staticmethod
def circuit_playground_type():
if "Bluefruit" in os.uname().machine:
return "Bluefruit"
if "Express" in os.uname().machine:
return "Express"
raise ValueError("Board unsupported by this library.")
def circuit_playground_is_type(self, is_type):
if is_type not in ("Bluefruit", "Express"):
raise ValueError("Valid Circuit Playground boards are Bluefruit and Express.")
return self.circuit_playground_type() == is_type```
from adafruit_circuitplayground import cp
if cp.circuit_playground_is_type("Bluefruit"):
print(cp.circuit_playground_type())
you could make circuit_playground_type a property; that would be consistent with what we make properties.
idle owl
ok
tulip sleet
you could save a tiny amount of code space by calling os.uname().machine only once and assigning it.
i'm just thinking about the frozen size
idle owl
I don't know how to do that and have it return properly.
slender iron
@spice crypt Ah. I'd just look in the source for those messages and turn the define on. I don't know it off the top of my head. (I never use it.)
tulip sleet
@staticmethod
def circuit_playground_type():
m = os.uname().machine
if "Bluefruit" in m:
return "Bluefruit"
if "Express" in m:
return "Express"
raise ValueError("Board unsupported by this library.")
I'm partway through your uheap memory video, hope to do it that way eventually.
tulip sleet
_CP_TYPES = ("Bluefruit", "Express")
@property
def circuit_playground_type():
m = os.uname().machine
for t in self._CP_TYPES:
if t in m:
return t
raise ValueError("Board unsupported by this library.")
def circuit_playground_is_type(self, is_type):
if is_type not in self._CP_TYPES:
raise ValueError("Valid Circuit Playground boards are Bluefruit and Express.")
return self.circuit_playground_type() == is_type
that assumes the type names are actually in .machine, which is true currently
idle owl
Something will be in .machine I would think. And we can return whatever we want, so if it's not necessarily the name, we can still return the colloquial board name.
hmm ok
tulip sleet
again, just saving bytes
idle owl
right. Does _CP_TYPES go in init or outside
tulip sleet
raise ValueError("Valid Circuit Playground boards are", self._CP_TYPES)
it's a class constant
idle owl
ok
tulip sleet
class ...:
_CP_TYPES = ...
yah ok that's where I was putting it.
@tulip sleet So this look ok for the error? Auto-reload is on. Simply save files over USB to run them or enter REPL to disable. code.py output: Traceback (most recent call last): File "code.py", line 3, in <module> File "/lib/adafruit_circuitplayground/circuit_playground_base.py", line 155, in circuit_playground_is_type ValueError: Valid Circuit Playground boards are ('Bluefruit', 'Express')
tulip sleet
I think so, do you think so?
idle owl
Good enough I guess. I used .format otherwise it does weird things with quotes separating each section.
We'll have to update the frozen lib. bleh.
At least we're not still dealing with 4.x I guess. Should probably figure out a way to make it fail gracefully if it's on 4.x still?
"figure out" I mean we already know how to do that. I suppose I simply mean do it.
Actually there's probably no point. Because Bluefruit was never supported on 4.x. We already put a hard stop on that.
idle owl
@gilded cradle or @tidal kiln This looks like something that one of you may be able to assist with: https://forums.adafruit.com/viewtopic.php?f=60&t=164690
gilded cradle
@tidal kiln, want to take a look since you're more familiar with the FT232H?
tidal kiln
@gilded cradle yep
gilded cradle
Thanks
tulip sleet
@slender iron you are right, notifications are only single BLE packet events, so the size is limited to that. And the discovery process does not include a way to determine a characteristic's max length.
I think it would actually be useful to have a Connection.max_data_length property so you could retrieve the negotiated max data length (mtu - 3) for a particular connection. That makes it easier to do length checking in various places.
manic glacierBOT
simple pulsar
I've been dabbling with ulab recently. For a simple bit of analogue sample reading its data storage is the slowest. Is that as expected? Times are in https://forums.adafruit.com/viewtopic.php?f=60&t=164758
pseudo pebble
Does anyone know if the CircuitPython port for the Xenon / Argon supports the bluetooth functionality?
idle owl
@pseudo pebble Last I knew, no. It utilises the microcontroller successfully but does not include support for the radio. It's been a while since I looked into it though.
pseudo pebble
@idle owl I knwo the Argon supports wifi, wasnt sure about the bluetooth
idle owl
Ah hmm ok. @solar whale might know better than me.
pseudo pebble
With Adafruit not yet supporting a feathewr of their own with onboard wifi that works with CP, my stack of Argons is still relevant
solar whale
@pseudo pebble the Argon is an nrf52840 and the Circuitpython bleio functions should be supported.
idle owl
Like I said, been a while since I looked into it. I'm guessing it was pre-nRF52840 CircuitPython support.
Thanks for chiming in!
solar whale
The Wifi via the ESP32 is only minimally supported
pseudo pebble
:: nods :: YEah, it works fine for basic stuff... but I gues particle is dead long term
Still, i have like 11 xenons and 8 argons and 3 borons... so if I can at least use the xenons ? Boonus
solar whale
the xenon and argon should be functionally the same.
I even the Boron should work the same -- just no support for the celluar part.
pseudo pebble
I flashed up an argon, I'll try some BLE code on it, see what happens
solar whale
you can use the wifi on the argon if you reflash the esp32 with the AT command firmware but it is not as well supported as the ESP32SPI capable boards.
pseudo pebble
:: nods :: Yeah, I am using it that way for some basic data logging. Qorks fine but the connection isnt super stable
solar whale
In my opinion,you are better off adding an airlift featherwing....
pseudo pebble
Well, a xenon + airlift woudl not be a horrible option. I hate having to go to a two board solution, evne with stacking, but it is what it is
solar whale
Good luck with BLE -- it should behave much like the fearther nrf52840 express
with a few pin name differences
pseudo pebble
:: nods :: I am lookign at some sample code now.
Big win foer me if it works. I cant go to production with them ofg course, but for dev? Perfect
Ok. It works
idle owl
Good to know!
pseudo pebble
Yeah. No real muss
ionic elk
@slender iron are RGBMatrix and FramebufferIO SAMD specific?
onyx hinge
@ionic elk ladyada has asked me to look into porting it to stm, starting with the feather.
but right now, it isn't on that port, just the ones tannewt mentioned
ionic elk
That's what I thought. I wasn't sure about FrameBufferIO though which is why I asked. That's part of RGBMatrix?
onyx hinge
right now framebufferio only works with rgbmatrix. in the future, we imagine that for microcontrollers with built-in LCD driving might implement something else that would work with framebufferio.
ionic elk
Ok, sounds good. Thanks!
onyx hinge
happy to answer any more questions you have.
slender iron
@ionic elk rgbmatrix and framebufferio are on nRF too
ionic elk
@slender iron cool! it's not a super big deal just mentioning it in my conversation with @tulip sleet in the docs PR. There's minor ambiguity for some modules about whether they should be "default" or not but I'd hardly describe it as urgent.
slender iron
just making sure you know after seeing "Chatted with @narrow dirge about this on Discord to verify that FrameBufferIO is only used with RGBMatrix on the SAMD for now." above
ionic elk
yeah I'll edit that to remove any confusion
Shouldn't impact the code, it's basically just whether the modules are set to 1, 0, or FULL_BUILD in circuitpy_mpconfig.mk. I figured in this case they should be 0 and turned on per-port.
slender iron
@vale spire @atomic summit got the build setup enough to crash the linker
atomic summit
oooer, progress!!! Awesome!
slender iron
you can make BOARD=saola menuconfig and the sdkconfig file ends up in the board folder
had to make minor changes to the idf to get it happy with -Wundef and -Wsigned-compare
vale spire
@slender iron neat! just reading throught he Makefile now. so you call the idf cmake with your sdkconfig and get a bunch of .a files for the relevant components and then you do the rest in the Makefile?
slender iron
@vale spire that's the theory
orchid basin
lucid solar
@slender iron Github check mpy-cross-mac is failing, I think it's because msgpack isn't installed now for some reason.
onyx hinge
@lucid solar noted, I'll look into it
we'll just drop that macos job if necessary, it's only a curiosity
https://github.com/adafruit/circuitpython/pull/2794 is an example of a job that is failing due to this
lucid solar
@onyx hinge we could just brew install msgpack
onyx hinge
are you familiar with macos and brew? I'm not, though I did write that section of the actions file. I would be happy to see a PR from someone who is, to make it work right.
lucid solar
msgfmt is part of msgpack I think...I could be wrong will check anyway
onyx hinge
on linux msgfmt comes from gettext, and we were running 'brew link --force gettext'. Until some recent change, that made the msgfmt program available. but now brew just criticizes you and doesn't do what was requested, even with --force
lucid solar
gettext was updated 2 days ago on brew, must have been that
onyx hinge
could be? yesterday a run said this: ``` brew link --force gettext
shell: /bin/bash -e {0}
Linking /usr/local/Cellar/gettext/0.20.1... 181 symlinks created
If you need to have this software first in your PATH instead consider running:
echo 'export PATH="/usr/local/opt/gettext/bin:$PATH"' >> ~/.bash_profile```
today a run said this: brew link --force gettext shell: /bin/bash -e {0} Warning: Refusing to link macOS provided/shadowed software: gettext If you need to have gettext first in your PATH run: echo 'export PATH="/usr/local/opt/gettext/bin:$PATH"' >> ~/.bash_profile
lucid solar
maybe we don't even need gettext installed if macOS provides it...but I guess it doesn't provide msgfmt...which I guess we use?
tulip sleet
do either of you have a Mac to update and check this on? If not I do (not used on a daily basis)?
lucid solar
I'm updating now 😉
onyx hinge
fwiw this seems to work today on github actions, so I can also PR it if @lucid solar's tack doesn't pan out. ```- brew link --force gettext
-
echo "::set-env name=PATH::/usr/local/opt/gettext/bin:$PATH"
weird way to set an environment variable, but what are you gonna do
lucid solar
msgfmt is in /usr/local/opt/gettext/bin on my mac, so I think that's a good fix. can you do the PR @onyx hinge please?
onyx hinge
sure thing
onyx hinge
@lucid solar are you familiar with how to merge or rebase your PR to get this change once it's merged?
tulip sleet
@lucid solar @onyx hinge I have a very vanilla Mac with only XCode installed, and there is no gettext available in /usr/local/
so I think you have it from a previous instsall
lucid solar
I was just going to say that...it's installed by brew
I didn't mean it was preinstalled
so we need the other command also
onyx hinge
"brew install gettext" might be needed on a fresh mac ? It seems to be preinstalled on the runner, though it's not explicitly listed at https://github.com/actions/virtual-environments/blob/master/images/macos/macos-10.15-Readme.md
onyx hinge
it's installed on the runner but is not placed on the PATH.
lucid solar
lets see if the test fails
tulip sleet
ok, got it
lucid solar
@onyx hinge yes, I can rebase my PRs when this is confirmed working.
onyx hinge
@lucid solar excellent. I wanted to make sure in case you were unfamiliar.
onyx hinge
I can change the PR to ``` run: |
-
brew install gettext echo "::set-env name=PATH::/usr/local/opt/gettext/bin:$PATH"
``` though that prints a warning: Warning: gettext 0.20.2 is already installed and up-to-date which seems a bit silly
it might be better, instead of assuming gettext is installed when I can't point at documentation that says it is
tulip sleet
this is odd: I did brew install gettext and now I have both gettext and msgfmt on my path under /usr/local/bin
onyx hinge
https://github.com/Homebrew/homebrew-core/pull/53489 may have changed the behavior a second time?
we aren't doing brew update/upgrade so we wouldn't get it until the base runner image was updated.
(look at me pretending I know much about brew :-/)
lucid solar
I'm assuming, the actions macos image installs some of the "Utilities" via homebrew, so it probably installs gettext as a dep. of one of them.
onyx hinge
@lucid solar my theory too
tulip sleet
yes, my brew updated itself before trying to fetch, so it may be fetching the older one?
lucid solar
"Where is the macOS source? We are in the process of preparing our macOS source to live in this repo so we can take contributions from the community. Until then, we appreciate your patience and ask you continue to make tool requests by filing issues."
tulip sleet
anyway, your pr is working, that job passed
i'll approve and merge once the test job works, unless you want to do it a different way
lucid solar
I think it's fine
onyx hinge
I don't need to fiddle with it further, thanks
onyx hinge
I think this is the change in brew that happened: https://github.com/Homebrew/brew/pull/7328
lucid solar
@ionic elk whats your plans for the STM32F7 ports? I'm starting to enable the missing modules, done pulseio, doing os now...
lucid solar
@ionic elk https://github.com/adafruit/circuitpython/blob/1f034f0f59e4263a52cd376ed53888eeaca3c0a4/ports/stm/mpconfigport.mk#L53
Is SPI really needed for displayio? when the driver for LTDC is made, it won't use it. But I guess they might still want to add a SPI display? just checking what you think.
onyx hinge
@lucid solar right now enabling displayio probably expects that spi and i2c are available. You're right, it could be separated out, but I suspect there's work to do first
ionic elk
@lucid solar Yeah I saw your quick PR to add PulseIO in, looks good. There are probably a number of modules that can very easily be added - OS and thus Random and Storage are among them. ADC support is probably among the more difficult.
They didn't make my original todo list because I like to double check the Micropython and STM32Duino implementations to make sure there aren't any initialization gotchas or other major hidden differences from the F4 that could mess users up unwittingly. Making sure of that takes testing and reading and it isn't at the top of my priority list right now, but I'm happy to review anything you feel is low hanging fruit.
lucid solar
@ionic elk yeah, I'm just enabling simple things just now. Ones that I know won't break other ports. I'll get some F4 boards so I can test more complicated things in the future.
ionic elk
As for DisplayIO, that comment could probably be a little cleaner, yeah. Right now it fails because PulseIO isn't in and there's some backlight brightness thing somewhere that needs it factored out...
If you want, you could just breeze through the initialization settings between the CubeMX exports for the F4 and F7 for the basic/PWM timers and see if anything stands out to you. Micropython doesn't have an great equivalent to our PulseIO so it's less ideal for gotcha checking compared to, say, flash, where we're basically identical
Just FYI I'm working on a PCB project I promised a friend I'd do by the end of the week so I'm going to have sporadic availability today, but I'll be doing some reviews tomorrow and I'll be fully back in gear on Monday as well.
lucid solar
Ok, take your time. None of this is really priority, it's just I'm wanting to understand how CPY works so trying to 'touch' as much of the codebase as I can.
ionic elk
No it's great! It's awesome to have someone else exploring the STM32 port in detail. And I'd also admit that I didn't spend as much time on the F7 implementations as the H7 ones so it's good to have you pointing out issues like these
lucid solar
I don't understand the shared-bindings/shared-module system yet.
I'd like to look at the L4P5 port at some point....seems @meager fog is interested in that mcu and it's very similar to f7 in terms of features.
ionic elk
I think they like it for the similarity in stats to the SAMD51. But that said the Micropython is littered with structural differences for the L series, so I definitely wouldn't underestimate the difficulty of adding it. I think it'd likely be as much if not more work than the H7 port.
They have different flash memory structure, different bus settings, all kinds of stuff.
But I'm excited to explore the L series now that we have more low power focus and support. I've got friends who use Circuitpython for environmental sensors so this kind of stuff is definitely intriguing to me
lucid solar
I'm more into graphics, that's why I want to get LTDC working on the discof7 board...the screen is nice!
I think I'll leave it for a few days though, I've got another CPY issue I want to tackle first.
Actually, I want to get openocd into the Makefile first, as it's very useful for debugging...plus the USART/VCP thing we talked about.
tulip sleet
@lucid solar shared-bindings/ provides the Python visible API layer for native modules. A native module is implemented per port in port/<port-name>/common-hal/<module-name>. If part or all of the module can be implemented in a port-independent way, then that port-independent code is in shared-module/
onyx hinge
so any knowledge of the underlying datastructures should be kept out of shared-bindings. I keep falling afoul of this, and tannewt gently corrects me
spice crypt
It's not in the same directory as mylog.txt resides
slender iron
hrm, what did you run in gdb?
slender iron
looks like the ram.bin bit is commented out
slender iron
line 13 may need to be updated too
that's how this stuff goes. it always needs a little updating because it's not fully auto
spice crypt
It's not happy after uncommenting the ram dump
`Breakpoint 1 at 0x310e6: file ../../py/gc.c, line 106.
Breakpoint 2 at 0x62d88: file ../../main.c, line 251.
Breakpoint 1, gc_log_change (start_block=start_block@entry=0, length=length@entry=2) at ../../py/gc.c:106
106 change_me += length; // Break on this line.
$1 = 0x0
$2 = 0x2
$3 = 0x6b0
#1 0x00037a20 in gc_alloc (n_bytes=<optimized out>, has_finaliser=<optimized out>, long_lived=<optimized out>) at ../../py/gc.c:597
597 gc_log_change(start_block, end_block - start_block + 1);
#2 0x0004441e in m_malloc (num_bytes=<optimized out>, long_lived=<optimized out>, num_bytes=<optimized out>) at ../../py/malloc.c:81
81 void *ptr = malloc_ll(num_bytes, long_lived);
#3 0x00053d88 in m_malloc0 (long_lived=false, num_bytes=24) at ../../py/malloc.c:122
122 void *ptr = m_malloc(num_bytes, long_lived);
#4 mp_map_init (n=3, map=0x20014f5c <mp_state_ctx+76>) at ../../py/map.c:87
87 map->table = m_new0(mp_map_elem_t, map->alloc);
../../tools/output_gc_until_repl.txt:32: Error in sourced command file:
No symbol "_srelocate" in current context.`
Also, I had to compile with -ggdb or else gdb would never break on gc.c:106
spice crypt
yes
slender iron
I don't think I've done it with nrf before
you should be able to watch ram.bin grow
spice crypt
it is indeed
slender iron
it dumps all of ram every gc action 🙂
does your code finish? you may need to ctrl-c it at some point
the break in main.c is meant to stop gdb after the user code finishes
spice crypt
I had originally tried to run this on my full code and Jlink just segfaulted after a few minutes. I did change the code to something very minimal
spice crypt
Yep, I see that. I'll give it a few minutes
do yo have analyze_heap_dump.py command in your history so I can avoid transcribing from youtube
slender iron
let me see
spice crypt
BBL, will report back
slender iron
@spice crypt I don't have it in my history. It's been a while since I did it
spice crypt
np. Thanks for looking
sour lynx
(I realize it's late.)
@slender iron I'm here, sorry, didn't get the notification yesterday
slender iron
no problem 🙂
I finally got some focused time to work on things
one thing I ran into is a few compiler errors due to use using -Wundef and -Wsigned-compare
sour lynx
sure, can pull these changes in
slender iron
the other question I had for you was how I could disable overwriting the given sdkconfig
my ideal would be to have a full sdkconfig.defaults for the port and then individual sdkconfigs for each board
where the sdkconfig only has the differences
sour lynx
you can pass multiple sdkconfig defaults file via SDKCONFIG_DEFAULTS variable, like so:
cmake -D SDKCONFIG_DEFAULTS="sdkconfig.defaults;sdkconfig.myboard"
(one sec, will double check the syntax...)
sour lynx
it will create sdkconfig file from the contents of both specified sdkconfig.defaults files
usually it is recommended to add sdkconfig to .gitignore
slender iron
kk, then it can be in the build folder
sour lynx
i think you should be able to customize sdkconfig file name, which makes it possible to move it to the build folder
slender iron
yup. I sorted that out yesterday
just hadn't thought about the board one being a default as well
sour lynx
cmake -DSDKCONFIG=build/sdkconfig -DSDKCONFIG_DEFAULTS="sdkconfig.a;sdkconfig.b"
slender iron
my current state is that the compiler is segfaulting when I try to use it for linking
so I've gotta mimic the link command the idf is doing
I'm happy with the structure
sour lynx
oh. that's interesting. if you find what change in the command line makes it segfault/not segfault, let me know
slender iron
kk
sour lynx
Possibly some improvement in the future about this part: https://github.com/tannewt/circuitpython/blob/bab671a45e8c327d170787c86549f3c2a2bdfb9e/ports/esp32s2/Makefile#L63-L87
to derive it from the build system, instead of hardcoding
slender iron
yup! That was just expedient
sour lynx
We tend to reorganize the files/directories once in a while. We do this in a way that doesn't break apps compiled with IDF build system, but if you are hardcoding directories then it may break unexpectedly.
slender iron
had to do that first since tinyusb pulls in the headers
ya, we can always update it too
sour lynx
i've logged a task in our issue tracker this week about finding a way to generate list of compiler/linker flags for external build systems.
slender iron
I was debating pulling them from the build.ninja file
sour lynx
that might actually be not a bad approach!
slender iron
I think I need to union them though
or find the component I'm actually using and snag it from there
sour lynx
if you declare requirements of the main component on other components you use, then the include paths for the main.c will be correct
manic glacierBOT
sour lynx
regarding tools/.gitignore — was it needed because you set IDF_TOOLS_PATH=$IDF_PATH/tools?
slender iron
yes, should I not have? seemed like what it was going for
manic glacierBOT
sour lynx
usually if you don't set it, it defaults to $HOME/.espressif, or %USERPROFILE%\.espressif
manic glacierBOT
slender iron
ah ok
sour lynx
in docker containers, sometimes we put it elsewhere, like /opt/esp
slender iron
heh. I'm on fish so the install needed me to add it
sour lynx
oh, i see. fish versions of install and export were contributed, i don't think we have fish users on my team, so this part might have been untested.
with bash and zsh it works without setting IDF_TOOLS_PATH
slender iron
ok, I'll remove the .gitignore changes and look into updating the fish script too
@sour lynx will you want a rebase with it or is a separate commit ok?
sour lynx
separate commit is fine. I think we may have to amend your commit which fixes -Wundef after some bikeshedding, which ifdef/if defined(/if defined pattern we want to standardize on in our header files. And also will need to add a CI job to prevent it regressing in the future.
So likely the PR will be rebased and some commits might be squashed after we accept it.
slender iron
👍 I tried to match the file's use
sour lynx
yeah, i understand. Maybe it will pass as is. Will see.
slender iron
they should probably be #if defined(FOO) && FOO to be extra correct
in general I want us to move away from checking defined
sour lynx
for things which come from sdkconfig, if defined(CONFIG_FOO) should be sufficient, as Kconfig never defines to 0.
slender iron
ya, that was my first approach (to have Kconfig define 0)
sour lynx
it might have been nicer if it instead always defined CONFIG_FOO to either 0 or 1, but alas
slender iron
there was a lot of other issues to run down due to that
I like always defining because then you can verify the referencing code actually included the appropriate header
sour lynx
that's right, agree that defining 0s would have been a better option.
nowadays we have a CI script which checks whether we may have forgotten to include "sdkconfig.h" or misspelled the option name.
slender iron
I think I have the start of that work in a stash if you are interested
sour lynx
i'm afraid if we change Kconfig behavior to define 0 values now, we have a good chance of breaking a lot of our users' apps
slender iron
ya, it's a pretty big change
sour lynx
pretty sure we had a PR internally with the same change a while back (around the time of IDF 2.1) and even then when we had much less users it was decided that the change was too big.
slender iron
kk
sour lynx
maybe in IDF 5.0...
slender iron
luckily I realized the simpler way to fix the warnings 🙂
thanks @sour lynx I appreciate the help. will poke the linker on my stream in a 2.5 hours
spice crypt
will poke the linker on my stream in a 2.5 hours
@slender iron Where do you stream at?
slender iron
on Adafruit
spice crypt
youtube I'm assuming
spice crypt
aha
spice crypt
@slender iron Re: heap analyzer. I got as getting the analyzer dependencies setup for analyze_heap_dump.py, but now running into the issue of different named symbols in the linker. I'll continue poking to find the appropriate translations. rom_start has me scratching my head right now.
nRF: https://github.com/adafruit/circuitpython/blob/master/ports/nrf/boards/common.template.ld#L97
SAMD51: https://github.com/adafruit/asf4/blob/039b5f3bbc3f4ba4421e581db290560d59fef625/samd51/gcc/gcc/samd51g19a_sram.ld
I'll continue later on today
slender iron
what symbol are you looking for now?
I'll be around off and on. just exercising now
spice crypt
Looking for mapping of this:
rom_start = symbols["_sfixed"][0] KeyError: '_sfixed'
'rodata' in nrf seems similar
slender iron
ya, worth a shot. I'd have to look at the context
spice crypt
Tried, but that's not really a symbol
slender iron
you may need to add it into the linker script
or hard code it
I think it should be FLASH_FIRMWARE
er, whatever CIRCUITPY_FIRMWARE_START_ADDR is
k, bbl. exercise time
spice crypt
👍
Linker: _rom_start = ${CIRCUITPY_FIRMWARE_START_ADDR};
mkdir heapviz
analyze_heap_dump.py: rom_start = symbols["_rom_start"][0]
Getting somewhere!
slender iron
nice @spice crypt !
lucid solar
@slender iron I want to output the console on both uart and usb-cdc...thing is the serial.c file that does it over uart isn't included if usb.c is. So I'm wondering where is best to put the code. Any thoughts?
in some ways I feel serial.c should really be console.c etc, but that's something that needs bigger discussion I guess
slender iron
@lucid solar for now I'd just #if it into the usb one.
lucid solar
ok...that's what I just started to do 😄
slender iron
we can always reorganize it later
ideally it'd be any stream object
though the ctrl-c checking requires an interrupt
goes to lunch
solar whale
@tulip sleet upgraded to ubuntu 20.04 -- part of a a complete rebuild of my system due to a disk drive failure. Just finished a complete build of CP (en_US only) but no issues (other than fomu - I have not installed the risc-v tools yet) . I managed to get away with just replacing my disk drive with a new solid state drive ... so far so good.
raven canopy
i got the motd on my RPi last night (running 19.10). not real sure i'm ready to test the waters there. 😄
sour lynx
@slender iron i've narrowed it down to the presence of --gc-sections on the command line. Which, again, is odd, because we use this flag in IDF build system.
but once i remove --gc-sections, the linker starts actually complaining about missing symbols instead of crashing
oh, and you seem to be missing -mlongcalls from CFLAGS
slender iron
k, awesome!
remote dawn
Brand new Adafruit Metro M0 arrived today. The first time I plugged it in to my Windows 10 desktop, it seemed to connect OK, but several minutes later Windows said it was malfunctioning. Now Windows doesn't even recognize it. Nothing shows up in the Devices USB listing; there's no "beep" or any indication in Windows that anything got connected at all! But the Metro does receive power thru the USB cable. I can also plug in the 9V supply before connecting the USB; just nothing at all! What should I try?
slender iron
@remote dawn did it have circuitpython on it?
remote dawn
It was ordered as CircutPython. The Adafruit Metro M0 instructions are confusing because they told me to start by installing the Arduino IDE, which I did. I downloaded the proper set of board files, and selected the Metro M0 but it said it couldn't connect.
slender iron
what happens if you double click the reset button? that should get you into the bootloader
remote dawn
The light beside the reset button stops varying across the spectrum and turns solid green. Also the tiny LED labeled "L" turns on. But nothing shows up in Windows.
slender iron
ah, so the varying sounds like the arduino rainbow pattern.
remote dawn
In fact, now the LED turns red, not green
slender iron
red means that usb isn't detected by the bootloader
you may want to try a different cable and remove any intermediate hubs
remote dawn
Aha. I moved the cable to a different plug, without any intermediate hub. Now I have a green light AND Windows is giving me a METROBOOT hub. So now what do I do? Ignore the Arduino IDE completely and start up Mu?
Windows just informed me that the device is set up.
I've downloaded the latest version of CircuitPython (a UF2 file) - should I try to install that?
Welp, I tried to open the INFO_UF2.txt file on the METROBOOT drive and now it's hung up. Window USB manager is not responding. I'm going to try resetting. Hmm... a simple single press of the Reset button closed out the METROBOOT and brought up a window called CIRCUITPY. So I guess I'm in business. Thanks, tannewt!
lucid solar
@slender iron I've kind of fixed some of the esp32s2 stuff
slender iron
@remote dawn no problem! glad you made progress
lucid solar
FileNotFoundError: [Errno 2] No such file or directory: 'boards/saola/saola-spi.ld' is my issue now
slender iron
@lucid solar igrr saw that you can remove gc-sections and it doesn't crash
weird! that might be an old version of my branch
lucid solar
ok...did you have that file then?
slender iron
ya, I think that is a litex remnant
but you wouldn't get that far without the newer makefile
lucid solar
@slender iron basically it seemed like xtensa-gcc was outputting the file, just not exiting correctly, so I tricked it into always reporting it exited ok. That seemed to fix some things, but not everything...so I think it's not a valid hack for now. I'll go to sleep now. Have a good weekend if we don't talk before Monday.
slender iron
Thanks you too! Happy birthday!
lucid solar
Thanks 🍰
idle owl
Here is the notes document for Monday’s CircuitPython Weekly meeting. 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/1hcf6M5Xfg-LgcnLCO06eExrDE44b1aTNV_eSqdDRjRE/edit
tulip sleet
@remote dawn If you are running certain antivirus or disk utility programs, they can interfere with seeing the BOOT drive (or CIRCUITPY). See some discussion here: https://learn.adafruit.com/welcome-to-circuitpython/troubleshooting#circuitpy-drive-issues
clever wren
I'm wondering if a library for micro:Bit micropython will work on circuitPython.
simple pulsar
@clever wren Which one?
clever wren
cube.bit accessary
4tronix Cube:Bit
Un a little unsure if it is the Python library to run the cube:bit from a raspberry Pi with Python or micropython for Micro:bit
lone axle
is the setup.py file included in the library repos "only" for installing the library via pip? Or does that serve a different purpose as well?
raven canopy
@lone axle only for PyPI/pip.
lone axle
Thank you 👍
simple pulsar
@clever wren From the 4tronix page on this, With Raspberry Pi, use any of the WS2812 code already available to drive your Cube:Bit eg. Pimoroni, Adafruit or 4tronix FireLed products. - The Adafruit NeoPixels are WS2812. The cube looks like it performs just like any other strip (or circle) of addressable LEDs. There are native Adafruit libraries to talk to these. https://circuitpython.readthedocs.io/projects/neopixel/en/latest/
spice crypt
@slender iron No wonder I have memory allocation issues all the time:
This is after initiation of all my code and then in "idle" state
I wish the GC just nullified all the "out of scope" blocks
ebon raptor
You could try using weakref more
ebon raptor
oh hm, does circuitpython even support weakref?
lone sandal
stuck elbow
@ebon raptor that could be hard to do, due to the way GC looks for pointers
ebon raptor
I'm more familiar with CPython than micro/circuit but weakref is more about triggering the refcount collector than the mark-and-sweep phase.
manic glacierBOT
onyx hinge
@tulip sleet @slender iron Looking at the changes on master since 5.3.0-rc.0, is the nRF firmware size calculation an important bugfix? I am pretty sure I don't want to go beyond that, where the SD version was changed. Just trying to decide whether to go to 5.3.0 final, bump slightly to the merge of #2800 (and get #2802, a nice little safe ulab update that might help PhilB), or what. I just happened to be checking now, will bring this to In The Weeds if chat doesn't produce a conclusion.
oh hmmm, #2802 is no good, it added docs but didn't actually pull in the ulab changes egg on face
tulip sleet
@onyx hinge The https://github.com/simmel-project/circuitpython/blob/689dd613791439b04b2cce6a45ebb9b99c95d098/ports/nrf/mpconfigport.mk change only updates the SD for nRF52833; it doesn't affect the nRF52840 ports
lucid solar
@slender iron when CPY is running a script...and it waits for CTRL+C...it seems to be running inside mp_builtin_input is that correct?
manic glacierBOT
tulip sleet
@lucid solar ctrl-C is detected in usb.c (mp_interrupt_char). It is not running inside mp_builtin_input. There's a check for ctrl-c there, but that's not the interrupting charater check
lucid solar
ok, thanks
tulip sleet
pyexec.c parse_compile_execute() is where code.py gets run (that is well down the call chain; it starts in main.c)
lucid solar
I think I found what I needed...I need to do mp_interrupt_char for serial also...nearly done...just having interrupt issues 🙂
lapis hemlock
@onyx hinge Jeff, do you want to add the equal and not_equal functions to the compare sub-module? As far as I remember, the == and != binary operators won't work properly in circuitpython, if the operands are ndarrays. The two functions would offer a workaround.
onyx hinge
I wouldn't mind seeing it added. 🙂 I have continuing work on the RGBMatrix stuff that takes precedence right now.
lapis hemlock
I have continuing work on the RGBMatrix stuff that takes precedence right now.
@onyx hinge My question was not an invitation to tango😉 I will see to adding the implementations.
manic glacierBOT
ionic elk
do we still have ure support or was that dropped?
tulip sleet
@ionic elk it's now re
ionic elk
how do I import it?
or is it something I might have missed on the stm32 port?
I can't get access to it with import re it seems
tulip sleet
check for MICROPY_URE or something like that; it hasn't been moved into the CIRCUITPY flags, I think
it's turned on for SAMD51 and nrf definitely
also same for json (was ujson)
ionic elk
Yep looks like it's something I just fixed with the docs change but wasn't on the specific feather I was using.
Thanks!
does it have any extension in capability from what's on the docs, along with the rename?
tulip sleet
you mean for our version? Our version is the same as MicroPython.
ionic elk
ok
low sentinel
I'm working on a UI for a 3d printer using displayio and an m4 express feather:
MemoryError: memory allocation failed, allocating 1096 bytes
> in uart_port.readline()
I have to poll JSON from a UART to get the status of the printer. The JSON generates quite a bit of garbage (gc every couple of seconds), maybe fragments the memory.
I'd love to use a fixed buffer, and I'd just adore some kind of memory dump via USB. Any tips, dear friends?
tulip sleet
@low sentinel if you can get away without parsing the JSON, but just reading characters until you hit the status field, that should help a lot. I'm not sure what you mean by memory dump via USB.
There is regular expression support: import re
maybe do readline() from the UART, though it will return an incomplete line if it times out
low sentinel
I'm not sure what you mean by memory dump via USB
I mean I have a lot going on here between displayio, json parsing, instrumentation, displayio font glyph cache, icons... It is tough to understand exactly where my memory is and some kind of visualization would be wonderful.
maybe do readline() from the UART, though it will return an incomplete line if it times out
That's the line that is OOMing.
Regexes - I'm not really sure I'll be able to use them. There are indices and identifiers and stuff. It's not straightforward anyway - would need to dynamically generate the regexes.
(I am in the wrong channel, this is more appropriate for help-with, I'm sorry!)
tulip sleet
there isn't a detailed dump available in an ordinary way. We have some tools for doing this, but they are meant for development.
and it doesn't necessarily classify the memory as you wish
slender iron
@low sentinel here is a stream I did about memory analysis: https://www.youtube.com/watch?v=baa5ILZTRkQ&list=PLjF7R1fz_OOXBHlu9msoXq2jQN4JpCk8A&index=7&t=12s @spice crypt has been doing more recent digging and is using displayio as well
low sentinel
Yeah, these SVGs look right about perfect
spice crypt
I had to make a few tweaks for use on nrf (will make a PR) but otherwise the instructions are clear. You will need to uncomment the append line in the gdb script.
manic glacierBOT
ionic elk
@slender iron how do you tell what driver one of these waveshare epapers has onboard? I've been poring over their datasheets and they don't seem to list it anywhere
onyx hinge
this is weird. My ulab PR overflows espruino-pico, going from >60kB free to over-full. but .. the added code in ulab itself is negligible. and it only happens after being merged with origin/master, not on its own. (happens to be based at 5.2.0-rc.0)
hmmm
the linker script has
FLASH_TEXT (rx) : ORIGIN = 0x08010000, LENGTH = 320K /* sector 4 is 64K, sectors 5,6,7 are 128K */
but the successful build prints
68356 bytes free in flash firmware space out of 393216 bytes (384.0kB).
FLASH_FIRMWARE is the name that is supposed to be used for this region, the script gets it wrong and uses FLASH instead but that refers to the whole flash
so this port was really full before, but the script reported wrong and made it look like there was about 60kB free...
will note it on the PR
bright aspen
I'm back to see what's going on over here. I'm helping out with the Covid Watch project. Covid Watch is an anonymous exposure notification app to help contain Covid-19. I'm doing some BLE testing and measurements. I'm using CLUE this instant, but have some other boards. I wan't to squeeze out what I can from BLE in CircuitPython. I'm currently using _bleio. However, there seems to be a warning that it could change. How much of a risk is that?
tulip sleet
@onyx hinge It is possible that the double-precision float routines have been included due to something in the code. Try arm-none-eabi-nm --print-size --size-sort --reverse-sort --radix=d and look for large routines, and look for dadd, dsub, dmul, ddiv etc.
ah, I read your later message.
@bright aspen It probably won't change that much; if there's something that you need in adafruit_ble in Python that we don't supply, feel free to open an issue.
we just don't guarantee that it won't change
hence the underscore
onyx hinge
@tulip sleet still a helpful recipe, I'll file it away in my memory
so happy to see this added on github pull requests, even if it's a little hard to spot.
slender iron
@ionic elk I know how to do it when looking at good displays website. what screen do you have?
@bright aspen nice! I've got my eye on a couple contact tracing projects. I mean to make sure we can be compatible with the Google/Apple system with a dedicated device.
bright aspen
@tulip sleet @slender iron Thanks! Covid Watch will probably switch over to Google/Apple and I'm also in the non-phone discussions, so this sounds great. I'm currently working in improving distance estimations.
thorny jay
I'm using CLUE this instant, but have some other boards.
@bright aspen I suggested the use of CLUE for anonymous Covid-19 tracing. If one tracing protocol/app is 100% open, then it could be implemented on an anonymous device such as a CLUE. https://twitter.com/DavidGlaude/status/1254016084719607809?s=20
simple pulsar
@thorny jay I think getting decent distance estimates from BLE is going to be really tricky, I've been thinking about this. In a "normal" city environment I'm sure a basic system would be near useless. Apple might have some big data ML-magic from their "find my" service to improve this. There's also some basic techniqes with smartphones to try and work out if it's inside a pocket/bag or in-hand.
@thorny jay But it could certainly be an interesting educational tool for showing the protocol or allowing people to work out the limitations of the tx power/rssi approach. I'd like to persuade my godson to do some R&D here!
thorny jay
Yes, finding distance might be difficult... but maybe this is not what you really want. You need to build a risk estimation based on the signal strenght and duration. If you pass by someone, the risk is low. If the signal strenght is low, the risk is low. But once you have a long duration and a strong signal, then the risk increase.
I say that, but some friends in the city tell me they can see the Bluetooth signal from a lot of their neighborg, up/down/left/right. But since it's another appartment, the risk is pretty low. So false alarm might trigger if your neighborg is affected.
In preparation for the PyCon(?) or what was that conference where 3000 CLUE needed to be distributed as badge for participant, I worked (with @lone axle ?) on an application that show he colour of the badge from nearby users. That was just using the BLE colour advertisement from Adafruit. It would be super simple to show a spreading by having all the participant "green" and one with a "red" badge. Then see how long it take before everyone turn "red". And that would be very educational.
onyx hinge
oops, sorry about that ^^^ (branches deleted now, I guess that doesn't warrant a notification)
simple pulsar
@thorny jay I think I remember discussions about that app, where is it? I just stuck what I was reading in https://forums.adafruit.com/viewtopic.php?f=53&t=164851 And on the subject of city vs less crowded space, when I first got an Amazon Firestick (WiFi only) I plugged it in and it told me I didn't have enough bandwidth to use it. What is was actually telling me was the 2.4GHz wireless was way too busy where I live!! I run it now on 5GHz which happens to be less crowded.
tulip sleet
@thorny jay the nRF52833 and some other newer nRF chips have Bluetooth Direction Finding support, which is not present in nRF52840: https://www.nordicsemi.com/Products/Low-power-short-range-wireless/Direction-finding. I'm not sure how applicable it is to contact tracing, but we did just have an nRF52833 port added to CircuitPython. That chip has less RAM and flash so it's a bit of a squeeze for CircuitPython.
simple pulsar
@tulip sleet Agreed, but there might me some very clever stuff to refine the estimates if you have a smartwatch and a smartphone.
Or even for the simpler case it could provide a small amount of extra information of correcting for attenuation from different angles. This is all guesswork, of course.
timber mango
I just stumbled onto this board which looks extremely interesting to me. I have sent a message to the seller to find out the exact M7 chip Toasty uses. I so want one of these if it can be supported by Circuitpython! https://www.tindie.com/products/webtronics/toasty-480mhz-stm32-usb-development-board/
ionic elk
I will add
it is an H7
STM32H743VIT6
already supported, stupid easy
I do need the flash chip tho
bright aspen
@simple pulsar @tulip sleet @thorny jay @slender iron Wow, lots of great ideas here. I should have come here much sooner. At Covid Watch we are talking about a Tile-like device for childret that depends on the parent's phone for memory and Internet. I too was thinking that CLUE or Circuit Playground Bluefruit could do that and mentors could help people get their devices started. That is very cool. I think that DP3T has a limited version in Python and that might be a place to start tinkering (I have not looked at it). A/G seems to be based on the DP3T work. I am on the Covid Watch team, and I lurk in the CoEpi, SafePaths and TCN Coalition Slack workspaces. (Covid Watch is part of TCN Coaliton). Besides the channel for making such gadgets, I'm also on the channel for distance measurements which is generalizing to exposure discernment. There we will be using CircuitPython for measurements. Well, two of us are getting up to speed on that. If anybody's workstyle allows, I'd encourage contacting a team that fits and jumping in. In the mean time, I'll be around.
onyx hinge
@ionic elk @timber mango the 2MB flash is just the figure for internal flash on STM32H743VI isn't it?
ionic elk
@onyx hinge on the Toasty? I wasn't actually going by the stats, I was going by the physical IC I see on the board that looks like a Nor flash chip
onyx hinge
I do see that 8-pin package near the USB connector too
ionic elk
I don't know what it could be, other than flash
the stm32duino port doesn't seem to have any information about it either
onyx hinge
agree
timber mango
I will add
@ionic elk Wow, cool! 🙂 I will see if I can find out what the flash chip is.
lucid solar
I designed a board recently, there is some ESD protection for the USB that was in an 8pin package...
I don't see why they would add flash to that board, 2mb on-chip and an SD card.
slender iron
@bright aspen I'm lurking on coepi. I'd suggest lurking on https://github.com/simmel-project/ too. It's bunnie and xobs making a trace device.
timber mango
I am hoping to order at least one Toasty when I get paid. I am hoping I can order two. I will just have to see how much money I have left after getting necessities.
tulip sleet
@timber mango don't forget to get a cheap multimeter re our convo weeks ago; will be useful for testing
slender iron
@ionic elk does the ribbon cable have a number on it? you can compare with the 2.9s on here: http://www.e-paper-display.com/products_list/pmcId=29&pageNo_FrontProducts_list01-004=2&pageSize_FrontProducts_list01-004=15.html
timber mango
don't forget to get a cheap multimeter re our convo weeks ago; will be useful for testing
@tulip sleet A decent DMM is very high on my aquisition list. I can not get by without one any longer. I had one of these before and it was excellent. https://www.digikey.com/product-detail/en/flir-extech/EX330/EX330-ND/6149613?utm_adgroup=Test %26 Measurement&utm_source=google&utm_medium=cpc&utm_campaign=Dynamic Search&utm_term=&utm_content=Test %26 Measurement&gclid=Cj0KCQjwhZr1BRCLARIsALjRVQOohqBI9gKi7cy2OuRuvg0m6cE3kV__00_JvHrk2xq-0_zin9e62bUaAr_lEALw_wcB
tulip sleet
@timber mango sounds good but if you are pinched a $20 one would be fine too. I managed to get a Triplett 1401 locally for a very good price. It has probe holders which I like a lot. I have the Extech also
timber mango
sounds good but if you are pinched a $20 one would be fine too. I managed to get a Triplett 1401 locally for a very good price. It has probe holders which I like a lot. I have the Extech also
@tulip sleet I will keep that in mind. If I can only get one thing, it will be a multimeter.
tulip sleet
Digi-Key has lots of 'em, including $20 (not autoranging, but that is usually OK with me): https://www.digikey.com/product-detail/en/flir-extech/MN35/MN35-ND/7322804
timber mango
@tulip sleet That does look like a decent little meter. If I can not get the Extech 330, I will go for the MN35. 🙂
timber mango
@ionic elk the 2MB flash is just the figure for internal flash on STM32H743VI isn't it?
@onyx hinge This chip has 1MB of on chip flash with the possibility of adding up to 1MB of external flash.
ionic elk
@timber mango ah yeah I missed that it's a VI, rather than a ZI
so does the Toasty come with the extra 1MB included? I don't know why it would be restricted to "up to 1MB" when could plop any old flash chip on there.
timber mango
Toasty has 2MB Flash, so has the 1MB external flash also, and 1MB SRAM which gives 864KB SRAM for users.
ionic elk
@slender iron wowzer that is a lot of almost completely identical 2.9" boards
I don't see any particular matches in the FPCB, seems like the silkscreen on it varies a lot even in the product photos. mine only has TP0-TP6 markings, no serial number.
marble hornet
How does the build script know what files to freeze into a firmware for some given directory? Does it do all py files except for setup.py?
manic glacierBOT
idle wharf
@thorny jay the nRF52833 and some other newer nRF chips have Bluetooth Direction Finding support, which is not present in nRF52840: https://www.nordicsemi.com/Products/Low-power-short-range-wireless/Direction-finding. I'm not sure how applicable it is to contact tracing, but we did just have an nRF52833 port added to CircuitPython. That chip has less RAM and flash so it's a bit of a squeeze for CircuitPython.
@tulip sleet Just a heads up, the Angle of Arrival (AoA) and Angle of Departure (AoD) that you can do in BLE 5.0 is not possible in phones or phone to phone. It requires a much larger antenna. Last I looked into it the antennas which could do it were more like ceiling tiles.
I was super excited about this approach until I learned that. 😕
tulip sleet
@marble hornet there's a list of what to freeze in the mpconfigboard.mk for each board (see CircuitPlayground Express for an example)
oh, sorry, the answer to your q is in tools/preprocess_frozen_modules.py
manic glacierBOT
marble hornet
Thanks! I will take a look and see where I can go from there.
manic glacierBOT
marble hornet
And I had some trouble figuring out: do you have to make clean every time you add or change what modules are frozen in?
Or can you just re-build?
idle wharf
@thorny jay and @simple pulsar Do you know if the circuitpython ble library can control the output power of the BLE radio? Since you want to track shorter distance interactions you can turn this down. I worked on a project where we had hundreds of BLE Beacons in a building and one issue was the beacons were transmitting at a power that you could see beacons from so far away, and it was too noisy\ hard to determine location.
slender iron
@idle wharf I don't think we can now but it could be added
thorny jay
@idle wharf sorry, I am not an expert... that would be for @tulip sleet but he is in a meeting now.
idle owl
CircuitPython Weekly in ~45 minutes. Everyone is encouraged to attend - you can always attend and listen in if you don't want to participate. Hope to see you there! Here is the notes doc. Please add your hug reports and status updates even if you'll be attending - it's super helpful! Thanks! https://docs.google.com/document/d/1hcf6M5Xfg-LgcnLCO06eExrDE44b1aTNV_eSqdDRjRE/edit <@&356864093652516868>
lone axle
Do we have any existing library (or builtin?) for parsing XML?
tulip sleet
no, sorry
@idle wharf there is skeletal power support but it's not really implemented. It would not be hard
idle wharf
Thanks @tulip sleet if I get far enough down this road, I can bring a C dev to the table to help.
tulip sleet
it's only a few lines of code
idle wharf
ha! You're not supposed to admit that. 😉
I'm pretty deep into looking in this (contract tracing with BLE) for a client and specifically the suitable Adafruit boards.
I'm choosing circuitpython for the proof of concept because the MicroPython boards I've used have incomplete libraries, lack of inventory or the development seems to have died off. So huge kudos to this community for the awesome momentum around CircuitPython!
lone axle
I wonder if there is any interest in having one created? I got to thinking about some way to define interface layouts within their own file more like how mobile development handles them. I thought about possibly using JSON for it for a bit over the weekend but I think XML may be able to express them in a more clean and easier to understand way.
tulip sleet
i would say that json and xml are equivalently functional.
for both it's harder to write an incremental parser
which would save memory
lone axle
Definitely both functional but the hierarchical / tree like nature of UI Layouts lend itself to XML a bit better I think. Maybe I'll go ahead and try mocking up some ideas for possible layout code with JSON. Perhaps it will work out nicer than I am thinking.
simple pulsar
There's also YAML ...
lucid solar
and TOML 😉
lone axle
This blogpost: https://engineering.instawork.com/when-xml-beats-json-ui-layouts-53c7f1d3fdb7 has some thought put into JSON vs. XML for use with interfaces. I'm not familiar with the "Hyperview" platform but they do raise some pretty good points about the differences between the formats and pro/cons of each for layout building.
bright aspen
@idle wharf What is the nRF52833 board? Concerning power, I think that is possible, thre is a soft device interface. Unless you beat me to it, I'll add an issue for that.
lone axle
YAML I am less familiar with. I should read up on that one some.
idle wharf
@bright aspen the different Nordic chips just have different specs. Some add NFC as well the Floating Point Module
https://www.nordicsemi.com/Products/Low-power-short-range-wireless
https://developer.arm.com/docs/ddi0439/latest/floating-point-unit/about-the-fpu
simple pulsar
I've used it for fairly simple configuration files in the past.
onyx hinge
both xml and json can be processed incrementally, it's the SAX model vs the DOM model. However, when writing to the SAX model you often end up with assumptions about the order of data encoded in the reader and writer (when the data formats may even specify that e.g., mappings in json can be arbitrarily re-ordered without changing meaning) that hinder interoperation with other generators/consumers
bright aspen
@idle wharf That is a cool table! Thanks! But, I asked my question poorly. You mentioned "but we did just have an nRF52833 port added to CircuitPython". Is that something available?
idle wharf
Ah.. gotcha. Sorry. I was quoting someone else who mentioned the nRF52833. But no, on Adafruit there no boards with the nRF52833. I do see the 832 come up though. https://www.adafruit.com/?q=nRF52833
The original post was about BLE AoA and AoD for precise distance measurement and as long as the chip\library support BLE5 you could use that part of the BLE Protocol, but again, I think the antennas are pretty big.
tulip sleet
@half geyser added support for nrf52833. We don't have a board right now, but he does (that is, he is working on a project with one)
bright aspen
I see nRF52832 and nRF52840. Ah, @half geyser . I'll look.
tulip sleet
we no longer support the '832 on CircuitPython, because it doesn't have native USB
is what xobs is working on. https://simmel.betrusted.io/
so you might want to get in touch with him 🙂
idle wharf
I also see that if you search for "direction" on the Nordic page that some chips support it and some do not.
https://www.nordicsemi.com/Products/Low-power-short-range-wireless
bright aspen
Do any of the non Nordic boards support BLE?
modern wing
Happily lurking today 🙂
...or is it lurking happily? Same difference.
onyx hinge
hello, I'll be taking notes today. Slow getting set up.
tulip sleet
@bright aspen no
modern wing
Loud and clear
onyx hinge
thanks @serene warren it's a big help
serene warren
Went fishing yesterday, got beat up pretty bad by wind and waves. Sore as can be today.
tulip sleet
@idle wharf we are starting our weekly audio meeting if you want to join. See the top pinned message for a link to the notes
@bright aspen too
serene warren
I find shaking the crumbs out occasionally helps with my keyboard
idle wharf
@tulip sleet Thanks for the invite, I can't today... just about to head away from my keyboard.
tidal kiln
lurking
sterile bronze
Hugs only today
timber mango
Text only
onyx hinge
@slender iron I'll get the links during community news
though I guess the links aren't in the doc, so it's anybody's guess:)
uncut nexus
lurking today
onyx hinge
Twitter
https://github.com/google/cyanobyte/tree/master/test/sampleData/micropython https://github.com/google/cyanobyte
idle owl
Stanford course expected 1k people - ended up with 80k people applying.
marble hornet
And flexible e-paper displays
onyx hinge
there's lots more in the newsletter that Scott isn't covering right this second
slender iron
lucid solar
k0d is swedish for code....but spelt differently...maybe I should try to change my github username 😉
lucid solar
I'll probably be lurking most of these meetings, as it's a time when my wife is home watching TV so I can't talk.
idle owl
@lucid solar You can also be text-only.
Where you type your notes into the chat channel, or better yet, add them to the notes, and we'll read them off.
lucid solar
@idle owl yeah, I've added them to the notes.
idle owl
@lucid solar Excellent, thanks!
onyx hinge
@timber mango will you be participating or are you lurking?
marble hornet
Need a minute before talking
raven canopy
i still maintain that Git is a strange, and magical realm we've only recently discovered. 🦄 😄
marble hornet
Back and can talk
slender iron
k, will do you last
onyx hinge
@sterile bronze drop us the link again in case anybody missed it 🙂
timber mango
@raven canopy Certainly strange. Unsure about magical.
sterile bronze
www.youtube.com/kingernorth and go to my Micro-controllers with kingernorth playlist
turbid radish
lurking
onyx hinge
thanks @idle owl I got distracted
idle owl
@onyx hinge No worries. 😄 I understand entirely.
turbid radish
Folks, please if you have any last minute news, projects, events, etc. in the Python/CircuitPython/MicroPython community, please email them to me at anneb@adafruit.com as soon as possible, in the next two hours if possible, for tomorrow's newsletter.
Thanks and sorry for the hit and run post, something personal came up in the last 45 mins and I'm sorting that out also.
slender iron
Twitter
tidal kiln
yes
onyx hinge
@slender iron sorry, cater got out of order in the document
slender iron
np
marble hornet
onyx hinge
@timber mango any notes for status updates?
crazy clock is a fun one https://www.tindie.com/products/nsayer/crazy-clock/
thorny jay
Yes that one.
slender iron
@lone axle have you looked at the 32blit at all? I was starting to think about games for it
onyx hinge
the circuitpython FAT filesystem is case-insensitive, which is why 'import label' and 'import Label' both wored
lone axle
Whoa neat, nope I hadn't seen that yet. I'll look into it. Definitely looks like a cool device.
thorny jay
I would love to know/have the electronic to control a clock from GPIO. It mean driving a solenoid with 1.5V, so to get from 3.3V and some protection for return current(?). It would make it possible to control the second hand from any microprocessor.
lone axle
Ah, Thank you Jeff. That makes sense.
slender iron
@lone axle they are doing some cool game asset stuff too
modern wing
@onyx hinge That's a pretty nifty novelty clock
timber mango
@onyx hinge I have nothing. I do not know where my brain is going to take me this week.
onyx hinge
slender iron
@ionic elk mute please
ionic elk
sorry!
slender iron
np, just heard you mousing. not disruptive
onyx hinge
@old smelt before next time, you may want to take a look in your discord "user settings", you can tweak the input sensitivity or use "push to talk" to prevent your audio from cutting out. When you are talking, watch for a green circle around your avatar -- if it's not steady while you're talking, then you are cutting in and out for us.
idle owl
@old smelt You still cut out a little, but it was a lot better this time.
old smelt
Oh sure. Okay. Using my studio setup, so that's odd.... I'll check it out.
Thanks!
Sorry... 😦
old smelt
I appreciate it.
idle owl
It's likely a Discord issue, not your hardware.
onyx hinge
modern wing
+1 for Jeff's suggestion, I've had that issue with each new mic I've attached to my system. And sometimes it randomly resets back to "auto-adjust" for no good reason 🙂
old smelt
Oh, meant to share this image.
onyx hinge
@old smelt I think discord is pretty aggressive at its "voice activity" sensing, which is why you might want to experiment with the options
old smelt
Will do.
solar whale
modern wing
And there's the new voice compression (on MacOS it's next to share screen and hangup), indicated by a vertical bars of various lengths. That might be causing an issue too.
onyx hinge
@idle owl we can look at that together in the weeds or some other time soon, if you like
raven canopy
@solar whale that's a good looking frame!
old smelt
@modern wing - I've had seemingly unending audio trouble when I run the mac app. I usually end up on Chrome, like today. But, I'm game to try again.
slender iron
@idle owl it'd be good to use the same board names in the CP library as you'd get from os
idle owl
@slender iron That's what it uses.
slender iron
perfect!
onyx hinge
https://wiki.radxa.com/RockpiS ROCK Pi S is so cute and small
old smelt
I have a hard stop (meeting) at 2:00. Gotta go. Have a great week! Thanks all...
idle owl
@onyx hinge Thanks, we can deal with it offline.
old smelt
My pleasure! Always fun.
modern wing
Thanks for joining us Jason!
onyx hinge
Isn't this what setup.py.disabled is for?
idle owl
@onyx hinge Yes.
onyx hinge
👋
serene warren
A Good time was had by ...me... Thanks everybody.
modern wing
Thanks!
ionic elk
👋
modern wing
@onyx hinge Quick question -- what brought up that crazy clock link? A possible CircuitPython project?
onyx hinge
@idle owl OK, we can pick up the topic of audio in here or in DM. Don't stay stuck for long.
@modern wing @thorny jay mentioned in his "other stuff" section that he was re-flashing the clock to turn it into a "week clock"
modern wing
Ahh, nice -- thanks 🙂
thorny jay
@modern wing It's my non CP project of the week.
onyx hinge
I had just heard about it long ago due to the Discworld "Vetinari clock" tie-in
thorny jay
But I would like to be able to control the second hand of an analog clock from a GPIO.
idle owl
@onyx hinge Let me at least get the rest of the code functioning on CPB and see if I manage to fix it in the process.
onyx hinge
@idle owl sure thing
thorny jay
Right now I try to flash an attiny45 from my Raspberry Pi. I did that years ago, so I know I can do it... but I forgotto document it.
onyx hinge
avrdude and the (equivalent of) 6-pin header?
thorny jay
Yes, I just want to make sure I connect the right pin to avoid burning it.
onyx hinge
yeah no kidding
marble hornet
Can you hear me?
tulip sleet
nope
you could make sd first, and then make flash to load the bootloader after the softdevice is loaded
slender iron
@marble hornet see this? https://github.com/daniel-thompson/wasp-os
@lucid solar see https://cyanobyte.dev/ ?
@pastel panther would be interested in it too
lucid solar
Nice!
lucid solar
Not sure why you shared that
tulip sleet
- hpa:
- sum:
- hpa
- division:
- sum:
- rawComp1
- rawComp2
- valueDP7
- 16.0
- hpa:
- division:
- hpa
- 100.0
return: hpa
i think a grimace emoji is appropriate here
i.e.
hpa = hpa + (rawComp1 + rawComp2 + valueDP7) / 16
hpa = hpa / 100.0
return hpa
When I detect CTRL+C inside of python on the serial, I call mp_keyboard_interrupt() as that's what happens on the USB-CDC also. But it seems to trigger a hard fault...any ideas why that might happen?
tulip sleet
hmm, noo. Set breakpoints at reset_into_safe_mode() and HardFault_Handler() to catch this and see what the backtrace is
marble hornet
@slender iron Yes! I did for the pine time Looks neat
slender iron
Here is the notes document for Monday’s CircuitPython Weekly meeting. 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/1RFV1kKlyBJfJkLivY_r_a28FFKXT83av-Wr1b4RMtu8/edit?usp=sharing
raven canopy
that cyanobyte spec schema is going to have to be immense...
i figure I2C is the "easy" one...
lucid solar
Yeah, I just wonder if any manufacturer will actually write schema themselves.
lucid solar
@tulip sleet thanks for the pointers...I know why it's not working now.
lucid solar
@ionic elk thanks...now I need to code faster as I have no outstanding PRs 🙂
lapis hemlock
@onyx hinge Jeff, would it be OK, if we moved spectrogram to either the fft module, or something else? It was a bit silly of me to put it into the user module.
Actually, something like scipy could make sense, and then we could dump all functions that come from scipy into that module.
marble hornet
or sometime that works for you? about uploading the softdevice via uf2
tulip sleet
@marble hornet you can't upload the softdevice via uf2. You can only upload it via DFU using the bootloader (if the bootloader is there), or using a J-Link.
(@gentle bronze will be working on a self-updater, but we don't have it yet)
marble hornet
@tulip sleet i tried make BOARD=<watch_name> SERIAL=<device location> dfu-flash and dfu-sd (just incase) and other variations/commands.
the dfu-flash seemed to work, ie update, but circuipython still crashed when trying to use the bluetooth. is there a command
how does one upload a new sd via DFU w/out wiping out the bootloader?
tulip sleet
i'm not sure you can. So there may be some other reason why bluetooth is not working. I am about to go afk for a few hours
dinner and a movie
marble hornet
thank you, i'll keep on trying / researching. have a great evening!
tulip sleet
good luck!
marble hornet
thanks
manic glacierBOT
onyx hinge
@lapis hemlock it's unfortunate to move it, since users are already using the current name.
lapis hemlock
Well, they have to bite the bullet, then. It is partly my fault, I admit, but it is still a mistake.
Another option could be to add an empty user module. That wouldn't upset the current scheme.
The problem is, if user functions and core functions are in the same module, then updating code from git is difficult.
onyx hinge
If you want to revisit it, let's talk to @slender iron about it. Maybe you can lay out what you see as the trade-offs in an issue or PR and mention us ? we can make a decision there
lapis hemlock
The last observation is the sticking point, I think.
onyx hinge
that's what "extra" was supposed to be, a place for your users to add their own functions and not get conflicts with changes in ulab?
lapis hemlock
That was my idea, and then I added the spectrogram. .
tidal kiln
just checking - there's no I2S in support in CP yet? only out?
lapis hemlock
If you want to revisit it, let's talk to @slender iron about it. Maybe you can lay out what you see as the trade-offs in an issue or PR and mention us ? we can make a decision there
@onyx hinge Done: https://github.com/v923z/micropython-ulab/issues/111
slender iron
fyi @idle owl we can't release pypixelbuf in its current state because it'll break neopixel spi
I downgraded librarians from write to triage to reduce the number of folks who can release but you still have admin permissions that could do it.
idle owl
@slender iron I haven't been releasing it. I told Dylan not to as well. Did it get released?
idle owl
Ok. Was already on my radar.
slender iron
epic! you are always ahead of me. thanks!
marble hornet
how could one check to see if there is a softdevice on the nrf52840 ?
if they aren't sure it is on there
timber mango
Hi, there. I'm using TSL2591 and it's been working but today I started getting "RuntimeError: Overflow reading light channels!" and that section of code is not all that clear. Anyone have knowledge of that sensor and can help me understand what the overflow condition is?
Looks like the constant is set to 100ms but not sure why that is and if there's any problem making it longer?
lucid solar
I've started to create a google sheet with potential CPY ports and their statuses. You can view/edit it here https://docs.google.com/spreadsheets/d/1vxaY6Xf1FIRPTwSZc-hnc2M3lZS9EWK31rR-7ULQS-M/edit#gid=0
I'll be adding more boards/info over the coming days.
It could also help if we add who has what boards...so if we need to test something, we know who to ask.
@slender iron @ionic elk
timber mango
Hi, there. I'm using TSL2591 and it's been working but today I started getting "RuntimeError: Overflow reading light channels!" and that section of code is not all that clear. Anyone have knowledge of that sensor and can help me understand what the overflow condition is?
@timber mango Answered in #help-with-circuitpython channel.
slender iron
@half geyser do you plan on improving circuitpython so it can work with the google/apple ble exposure notification stuff?
@tulip sleet when you have a chance can you test m0 master for flakiness? I think I'm able to reproduce it with master
ionic elk
@lucid solar Nice! Note we've already got the F412 and F411 discovery boards implemented. There are definitely some other no-brainers to implement on there, anything that's already got a supported SoC is a shoe-in. I'm not 100% on stuff like the F429 yet, just since they sometimes have breaking gotchas that can take time to resolve.
lucid solar
It's nice to have some idea of the scale of things, how many boards exist etc, and to see if any are interesting to port to...
ionic elk
We're also not prioritizing anything with less than 128K of ram, so there aren't any immediate plans to add support for the F3 series and below
Oh yeah it's a fantastic list, I was thinking I needed to put one together the other day so this is great
lucid solar
The reason my name is grey on some boards, is because they're in my basket ready for the next order 😉
I will add the Adafruit boards on tomorrow.
crimson ferry
@lucid solar I'd suggest a "user RAM" column if other boards are like the STM32F405, which has 256, but only 128 is accessible to user code
lucid solar
@crimson ferry maybe...but this is more from a porting view....and that's maybe not interesting?
lucid solar
https://github.com/adafruit/circuitpython-org/tree/master/_board This isn't autogenerated right? I should add info about the STM32 boards I've ported? Of course there are other files I see need data also...
slender iron
@lucid solar it isn't autogenerated
lucid solar
Thanks...now I sleep, have a great rest of the day everyone.
half geyser
@spice crypt those graphs look cool! Do you have any information on how you generated them?
@slender iron The goal with simmel is to provide a platform to develop on. The choice of what protocol to use seems very divisive, so the goal is to support them all. It's partly why I have the AES256 module for circuitpython.
Having said that, the Google/Apple protocol assumes you have an always-online device, which is incompatible with Simmel, which is designed to run up to 6 months disconnected on a pair of AAA batteries.
As a workaround, you could set up a server to do the polling and notification, which /is/ compatible.
slender iron
how does simmel check to see if it's been exposed to someone who reports having covid?
tulip sleet
@slender iron I'm going to stop working now for the evening, but in the morning I'd like to test MIDI with the new packet length changes. Are you talking MIDI to/from an iOS device, or are you using something else? THanks.
slender iron
@tulip sleet ya, using an iPad. can you test master with usb as well?
tulip sleet
sure, you made the interrupt change?
slender iron
my branch should have current master merged in
tulip sleet
I am just testing on my very vanilla Dell desktop, USB2, but I think it was bad on USB3 also. Were you using Garageband or something else?
slender iron
I think it was a mix of synth one and xynthesizer iirc. check with @split ocean. he has better apps to test with I think
tulip sleet
i got the Moog synth app for free when they were offering it. Are the examples in the repo a sufficient test, you think?
split ocean
Moog synth app is great for testing.
slender iron
@half geyser it'd be nice to have an aes module that is a subset of the one in pycrypto: https://www.dlitz.net/software/pycrypto/api/current/
split ocean
I'd recommend the free MIDI Wrench to see realtime messages as well as to set up virtual MIDI ports if needed. Some apps won't connect directly to BLE MIDI devices so MIDI Wrench allows virtual wiring of all that.
I've also connected BLE MIDI to mac os machine using the Audio MIDI setup settings in the OS. I gather that it may be painful to do this in Windows.
tulip sleet
thanks, that's very helpful. I have a 25-key controller and an older digital piano but they're all DIN MIDI or USB
i could see about Linux too. I have a Bluetooth dongle
split ocean
With MIDI I feel like the age of the connector standard is proportional to the ease of use and robustness.
DIN5 MIDI, bulletproof. USB, hrm. BLE MIDI, yikes.
WiFi MIDI, RUN!
tulip sleet
i have a simple cheap USB MIDI interface. I had an older more expensive one but it had proprietary Windows drivers (circa Win98) and I think it eventually stopped being usable all together.
split ocean
I've had good luck with a cheap USB MIDI interface cable as well.
tulip sleet
I'll certainly ask if I run into difficulties. tx much
split ocean
Happy to help.
half geyser
@slender iron thanks, I didn't know what the current preferred cryptography library was.
spice crypt
@spice crypt those graphs look cool! Do you have any information on how you generated them?
@half geyser These are using @slender iron 's gc activity monitor code.
Video: https://www.youtube.com/watch?v=baa5ILZTRkQ
Git readme: https://github.com/adafruit/circuitpython/blob/master/tools/gc_activity.md
There are other files related to that in the tools folder
It's setup in the repo for SAMD, needs some tweaks to make work for NRF
half geyser
Ah, I see. gc_activity.html in tools reads allocation_history.json and parses it into a pretty graph. That's super fancy.
slender iron
The visualization in the video isn't using gc_activity.html. It uses graphviz
spice crypt
Here are my personal notes on how to get this to run for NRF. https://gist.github.com/k3wals/d48f49050c1d1914f8f0d3fb0314c851
marble hornet
@slender iron does MICROPY_ENABLE_GC turn in all gc or just long lived?
marble hornet
okay, i'll leave it be then.
slender iron
the bugs we've found with it usually cause crashes
marble hornet
@slender iron, i am running your low_power branch on the watch after implementing a screen sleep, i'll report on the battery life within the next few days
half geyser
Is there a way to do something like MP_ROM_LIST()? I'm working on the crypto module, and I'd like a static list of the available sizes.
Most instances of mp_obj_list_t seem constructed at runtime.
slender iron
You could do a static tuple since they are immutable
No fancy macro for it though that I know of
half geyser
I think that's probably what pycrypto uses.
Are there any examples of statically allocating static tuples?
I guess it's just a len, a base, and a list of items...
Ah, MP_DEFINE_ATTRTUPLE looks like exactly what I need.
marble hornet
@slender iron scratch that, whenever it soft reloads the watch hangs on trying to init hardware
double checking
yeah, the non - low_power verson of the nrf52840 master hangs after a soft-reload and needs to be reset before it can re-init hardware (spi lock )
soft reload via ctrl-c, ctrl-d . do you want me to register and issue on your fork tomorrow?
half geyser
@marble hornet brownout?
full epoch
Hi, this might be a dumb question but: I plan on buying a Featherwing prop maker. But is it compatible with WS2812 and WS2812B led strip as they both use RGBW?
marble hornet
@half geyser it only happens with low power, and it is pugged into 5v
half geyser
Ah, okay. I ran into a problem with Simmel, but that just has to do with how I'm measuring current. Also, it's not a good idea to power it off of a pair of 1.5V Li-Ion AA batteries. The power briefly dips below 2.7V.
(Also, I discovered that the ARM debug core doesn't step into interrupts when you do stepi)
manic glacierBOT
lucid solar
Hi, this might be a dumb question but: I plan on buying a Featherwing prop maker. But is it compatible with WS2812 and WS2812B led strip as they both use RGBW?
@full epoch
@idle owl Maybe you know?
tulip sleet
@lucid solar WS2812 or WS2812B should be fine. The difference is internal and it doesn't appear the timings differ. WS2813B currently doesn't work but I have a sample and will try to make it work.
But I thought WS2812(B) were straight RGB, not RGBW. A lot of our RGBW strips are SK6812RGBW. So maybe your RGBW are not really WS2812x? There is a lot of casual use of "WS2812" when the underlying pixels are not really that type.
lucid solar
@tulip sleet that wasn't really for me, it was @full epoch
full epoch
@tulip sleet thanks for your answer. Keep me I formed please
bright aspen
BLE: Does the softdevice API allow obtaining the channel that an advertisment was received on? Folks over at Covid Watch have expressed an interest in this datum.
tulip sleet
@bright aspen The channel index from the last received advertisement is available, though we don't currently pass it on to Python. https://infocenter.nordicsemi.com/topic/com.nordic.infocenter.s140.api.v6.1.1/structble__gap__evt__adv__report__t.html#a7620e6f82c248bc997b212037f6ce472
You could add an issue if you want this reported up.
idle wharf
BLE: Does the softdevice API allow obtaining the channel that an advertisment was received on? Folks over at Covid Watch have expressed an interest in this datum.
@bright aspen I'm curious how this information would be helpful in the solution. Is that something you can share?
slender iron
@tidal kiln you around today?
tidal kiln
yep
slender iron
can you review and test https://github.com/adafruit/Adafruit_CircuitPython_NeoPixel_SPI/pull/15 ?
we can roll out pixelbuf once it's updated
@lucid solar I haven't had time for esp yet this week. lower_power is getting in my way
tidal kiln
sure. looks mildy urgent? things are blocking on that?
lucid solar
@slender iron I'm trying another approach to how you did it...adding CPY(makefiles) to CMake instead of the other way round.
slender iron
@tidal kiln if we release pypixelbuf then we'll break neopixel_spi