I hadn't thought about the number of items appended being the group size though.
#circuitpython-dev
idle owl
onyx hinge
(not sure why groups can't be enlarged after creation, must be some reason it can't be allowed)
idle owl
That's a place to start.
onyx hinge
a normal python list can be enlarged, so here's something you could do: gather up all your items in a list, get the len() of the list, and then create the group that big.
it would change your code a bit of course
I don't know if that's a sensible idea or a workaround, it's too early in the day to know such things
idle owl
hmm
onyx hinge
in my code in jeplayer I just always counted the number of things and made the group that big. but it was a small number like 3 or 8 ...
idle owl
yeah I think I'll pick an arbitrary number that is hopefully large enough and go from there.
This is 4 objects total, but the second one is its own group
er
fourth
onyx hinge
if nothing is built "conditionally" then I would expect it to run the same everytime
idle owl
one of the groups moves around the display based on accel data
onyx hinge
changing position by updating the x and y properties?
idle owl
yes
onyx hinge
that should not count against a group's max
cursive condor
but ... why is there a (arbitrary) limit to the number of elements in a group ?
idle owl
which btw, thank you @cursive condor for your maze demo. While that was way more complicated than what I'm doing, I was at least able to suss out how to make a group mobile.
onyx hinge
if you set a text label with too many characters I think you can get this. maybe a sensor goes from "3.14" to "-10.00" and exceeds a limit? Capture the stack trace next time and maybe that will give a clue.
a Label has a group inside, either the size of max_glyphs or len(text)
idle owl
I'm not sure what you mean by capturing the stack trace. is that the traceback? or something more complicated
onyx hinge
@cursive condor I am guessing there is a reason that the storage can't be enlarged after creation, I just don't know why that is. Tannewt would be able to tell you why he made it that way, but he doesn't seem to be around
cursive condor
TY @idle owl. if you have a monster m4sk you can check Chateau in my github, there is adding and removing of groups
onyx hinge
yeah the traceback. sorry, other language lingo creeping in
idle owl
@cursive condor If I hit a point where I need that, I will definitely take a look, thanks.
@onyx hinge no worries, simply making sure I understand. 🙂
onyx hinge
I have to go to this thing, but let me know what you discover!
idle owl
Ok. Thanks, @onyx hinge! Have a good one!
idle owl
Is there a way to have a single if statement that checks two things separately? So not "or" because that means if either is true. I am referring to I have foo and bar, and I want to do something to foo if foo is true, and something to bar if bar is true, but not to both if either is true.
I think I might need two separate sets of if statements.
tulip sleet
if foo:
foo = 2
if bar:
bar = 3
??
tulip sleet
are you checking for None or something?
idle owl
Yeah, basically
no
It's checking temperature against a temperature range
and humidity against a humidity range
and changing text color based on whether it's outside the range
tulip sleet
do you want to limit it to the range or just check that it's in the range
idle owl
check that it's in. Currently I have this for the min: python if temperature < min_temp or humidity < min_humidity: data[3].color = (0, 0, 255) data[3].text = "Temp: {:.1f} C".format(temperature) data[5].color = (0, 0, 255) data[5].text = "Humi: {:.1f} %".format(humidity) data.show()
problem is, it means if either is true, both things happen
stuck elbow
both?
tulip sleet
is the color different out of range?
stuck elbow
there is only one thing happening
idle owl
@tulip sleet yes
@stuck elbow that code is wrong, I know.
I need to change two separate lines of text based on two separate pieces of data
can't tell if I can do that in a single if statement
stuck elbow
if temperature < min_temp:
data[3].color = (0, 0, 255)
data[3].text = "Temp: {:.1f} C".format(temperature)
data.show()
if humidity < min_humidity:
data[5].color = (0, 0, 255)
data[5].text = "Humi: {:.1f} %".format(humidity)
data.show()
this?
idle owl
Yes, that is what I think I need, but wanted to make sure
tulip sleet
if temperature < min_temp:
data[3].color = ...
if humidity < min_numidity:
data[5].color = ...
data[3].text = ...
data[5].text = ...
That I wasn't missing some magical Python syntax
ok
yah not sure I can have .text outside the if, but I'll find out
stuck elbow
don't see why not
idle owl
anyway, was simply checking that I wasn't unaware of some kind of syntax to do weird things with a single if statement.
I am apparently not missing anything.
yah it works with the text outside the if
so that slims it a lot
tidal kiln
@idle owl w.r.t. group full and text labels:
https://github.com/adafruit/Adafruit_CircuitPython_Display_Text/issues/11
for group in general, i think it's because the memory allocation happens in the core C stuff, so size must be stated explicitly. or to put it another way, a group's size can not be dynamic.
idle owl
I'm not concerned with group size being dynamic, I assumed there was a reason for that, and that makes sense. I simply wanted to understand how to pick a group size because it seemed entirely arbitrary when working with text. Because text size changes. But I think that was fixed with max_glyphs. I don't know. Either way, tried to figure it out.
@stuck elbow @tulip sleet ```python
while True:
temperature = clue.temperature
humidity = clue.humidity
data[3].text = "Temp: {:.1f} C".format(temperature)
data[5].text = "Humi: {:.1f} %".format(humidity)
if temperature < min_temp:
data[3].color = (0, 0, 255)
data.show()
elif humidity < min_humidity:
data[5].color = (0, 0, 255)
data.show()
elif temperature > max_temp:
data[3].color = (255, 0, 0)
data.show()
elif humidity > max_humidity:
data[5].color = (255, 0, 0)
data.show()
else:
data[3].color = (255, 255, 255)
data[5].color = (255, 255, 255)
data.show()
you could put the data.show() at the end after the last else, since it's the same in all the clauses of the if
tidal kiln
can move data.show() out
idle owl
oh hmm ok
tulip sleet
what are we supposed to say, jynx 🙂
tidal kiln
jynx
manic glacierBOT
tidal kiln
could also have the (255,255,255) set be before the conditional and remove the else
stuck elbow
@idle owl that will only ever make one of the things red
I thought you wante the humidity and temperature independent
idle owl
I do.
tulip sleet
you need to group the temp checks together; they shouldn't all be elifs
tidal kiln
and have two if/else (think same as what @stuck elbow is saying)
stuck elbow
then don't do elif between them
idle owl
it works though...
stuck elbow
try with both low or both high
lone axle
@idle owl Got it, will do.
idle owl
oh i see. it gets weird.
tulip sleet
while True:
temperature = clue.temperature
humidity = clue.humidity
data[3].text = "Temp: {:.1f} C".format(temperature)
data[5].text = "Humi: {:.1f} %".format(humidity)
data[3].color = (255, 255, 255)
data[5].color = (255, 255, 255)
if temperature < min_temp:
data[3].color = (0, 0, 255)
elif temperature > max_temp:
data[3].color = (255, 0, 0)
if humidity < min_humidity:
data[5].color = (0, 0, 255)
elif humidity > max_humidity:
data[5].color = (255, 0, 0)
data.show()
ok. i won't copy/paste same thing this time.
idle owl
Ok, so I also need to add a buzzer. Which if I move the stop_tone outside of the else it pulses, it doesn't play the tone evenly on a temp/humidity event.
tidal kiln
what's the event? being above/below limits?
idle owl
yes, I was trying to type less.
tidal kiln
could set a flag and then trigger on that later?
idle owl
now that everything is moved around, it doesn't play the tone evenly anymore. not sure what change caused that, but even putting an else below each group with the stop_tone in it doesn't resolve it.
Also this doesn't work
when it's below the threshold, it flashes white then holds blue and repeats.
Presumably because it's running through the loop which sets it to white
tidal kiln
ah, yah. it would do that. sounds like there's more inside the forever loop?
idle owl
while True:
temperature = clue.temperature
humidity = clue.humidity
data[3].text = "Temp: {:.1f} C".format(temperature)
data[5].text = "Humi: {:.1f} %".format(humidity)
data[3].color = (255, 255, 255)
data[5].color = (255, 255, 255)
if temperature < min_temp:
data[3].color = (0, 0, 255)
if buzzer:
clue.start_tone(1800)
elif temperature > max_temp:
data[3].color = (255, 0, 0)
if buzzer:
clue.start_tone(1800)
else:
clue.stop_tone()
if humidity < min_humidity:
data[5].color = (0, 0, 255)
if buzzer:
clue.start_tone(1800)
elif humidity > max_humidity:
data[5].color = (255, 0, 0)
if buzzer:
clue.start_tone(1800)
else:
clue.stop_tone()
data.show()```is what I have currently
tulip sleet
(i have to drop out for now)
idle owl
@tulip sleet Thanks for your help.
tidal kiln
so you want the tone to play constantly until temp/humid gets back within the limits?
idle owl
yes. This works: ```python
while True:
temperature = clue.temperature
humidity = clue.humidity
data[3].text = "Temp: {:.1f} C".format(temperature)
data[5].text = "Humi: {:.1f} %".format(humidity)
if temperature < min_temp:
data[3].color = (0, 0, 255)
if buzzer:
clue.start_tone(1800)
elif temperature > max_temp:
data[3].color = (255, 0, 0)
if buzzer:
clue.start_tone(1800)
else:
data[3].color = (255, 255, 255)
clue.stop_tone()
if humidity < min_humidity:
data[5].color = (0, 0, 255)
if buzzer:
clue.start_tone(1800)
elif humidity > max_humidity:
data[5].color = (255, 0, 0)
if buzzer:
clue.start_tone(1800)
else:
data[5].color = (255, 255, 255)
clue.stop_tone()
data.show()
sort of
or at least it did work once. doesn't now.
still "flickers" the sound.
cursive condor
you better start the sound just once and set a var to True until you stop the sound it in the else: clause
tidal kiln
if one param is within limits, it will stop the alaram
idle owl
yeah ok
tidal kiln
while True:
alarm = False
temperature = clue.temperature
humidity = clue.humidity
data[3].text = "Temp: {:.1f} C".format(temperature)
data[5].text = "Humi: {:.1f} %".format(humidity)
if temperature > max_temp:
data[3].color = (255, 0, 0)
alarm = True
elif temperature < min_temp:
data[3].color = (0, 0, 255)
alarm = True
else:
data[3].color = (255, 255, 255)
if humidity > max_humidity:
data[5].color = (255, 0, 0)
alarm = True
elif humidity < min_humidity:
data[5].color = (0, 0, 255)
alarm = True
else:
data[5].color = (255, 255, 255)
data.show()
if alarm:
clue.start_tone(1800)
else:
clue.stop_tone()
is buzzer a global control for turning the alarm feature on/off?
idle owl
Yeah. set at the top.
@tidal kiln alright, that works.
tested with setting both off at the same time, and each one individually
results are what I wanted.
tidal kiln
looks like start_tone does a check for playing, so above seems safe-ish?
idle owl
yeah.
tidal kiln
ok to spam calls to start_tone
manic glacierBOT
onyx hinge
anybody know what a sphinx doc for special methods like "+" aka __add__ look like in a C file in the core?
do I just document the method called __add__
meager fog
@onyx hinge i know we have special things like setitem
i think'd be the same
@simple pulsar p0ke - plz tag me when yr around
onyx hinge
@meager fog aha that helps! shared-bindings/_pixelbuf/PixelBuf.c://| .. method:: __setitem__(index, value)
indigo wedge
@meager fog I emailed you some instructions and files, I'll be available in a few hours.
meager fog
@indigo wedge np i just know its hard to do over email - ping me when yr around
ill try what you sent as well 🙂
orchid basinBOT
obsidian compass
I’ve added side neopixel strip to lanyard of my cloudBadge : https://twitter.com/iayanpahwa/status/1225466296352956417?s=21
Here’s the video —> https://youtu.be/g_Ey5NY_z4Q
Used the stemma connector to attach strip
onyx hinge
I am stuck with this error while building the docs: <rst_epilog>:1: WARNING: more than one target found for cross-reference 'array': array.array.array, ulab.array
slender iron
@onyx hinge try replacing the array reference with ulab.array
so that it isn't ambiguous
onyx hinge
I don't have any references like ~array or `array` in my source, unless my editor's search is tricking me
//|
//| :param order: Whether to flatten by rows ('C') or columns ('F')
//|
//| Returns a new `ulab.array` object which is always 1 dimensional.
``` my references already look like that@cursive condor @onyx hinge @idle owl @tidal kiln Group size is fixed for now because it was the simple thing to do at the time. I was already doing a bunch of complex display transforms. If someone wants to change it to make it dynamic please do.
idle owl
@slender iron noted
I was trying to understand how to choose the number. I believe it was explained well enough.
slender iron
setting size up front is better for memory use too because dynamic algorithms tend to over allocate memory
idle owl
Becuase I kept getting group full and the number that worked was entirely arbitrary, so
slender iron
👍 it should be big enough to store anything you plan on adding later
idle owl
based on what you append?
slender iron
yup
idle owl
so each append is 1
slender iron
correct
idle owl
if I append 4 times, group size 4 should not fill up
slender iron
yes
idle owl
Ok. Thank you.
slender iron
np
tidal kiln
but if you append again, then you'd hit the limit
@slender iron it seems like there are good arguments for keeping it not dynamic
onyx hinge
aha ! There was a reference to array elsewhere; it had to be qualified to array.array!
(you have to find it, sphinx won't tell you)
idle owl
@onyx hinge welcome to Sphinx.
slender iron
@onyx hinge nice! I just realized I hadn't helped you find it. it makes sense that you'd find other previous references to array
onyx hinge
np, I wasn't stuck for that long
slender iron
@tidal kiln maybe, maybe not. it depends on how much extra would have been allocated
onyx hinge
anyway, learning experience 🙂
ionic elk
Whenever is good for you
slender iron
@tidal kiln looks like it doubles every time it runs out of space: https://github.com/adafruit/circuitpython/blob/master/py/objlist.c#L244
meager fog
im here now
slender iron
want to video, audio or text?
ionic elk
We could hop in amelia?
slender iron
k, lemme kick the cat off my lap and head upstairs
manic glacierBOT
onyx hinge
everything takes longer than you estimate, even when you account for having a cat on your lap
meager fog
i guess the big Q is do we want to do STM32FH7
or keep cleaning up the STM issue list
slender iron
I'd like to 🙂
then we get portenta and 32blit
I think we'll find similar uses to the imx rt
ionic elk
It shouldn't be too tough to add the F/H7, since Micropython has some good references as to what changes, but there's a number of modules I keep seeing people asking about
meager fog
also possibly the https://www.adafruit.com/product/4478
slender iron
to do that we'll need a camera module
meager fog
k
ionic elk
I also want to revisit the internal flash at some point, because it's still pretty slow, and a lot of boards use it.
slender iron
I think pulseio and rotaryio are good to do first
meager fog
i second pulseio and rotaryio
ionic elk
It doesn't seem like QSPI is very high priority though
Ok, cool. I can get those done pretty quick I think.
Do you still want me to do any audio stuff?
It's on my basecamp todo
ionic elk
oh right yes I can do that today
So, shortlist: PulseIO/RotaryIO + docs, Setup for F7H7, then maybe flash?
Is there a particular board you'd like to see first with the FH7?
I can probably clean up remaining structural issues (package selection, etc) alongside the F7H7, since they'll require some structural thinking anyway
slender iron
I'd probably start with the stm32h743
tinyusb has the f767nucleo and h743nucleo
so those would be best to start with
meager fog
yep!
ionic elk
One modification I'd suggest though is that if the F7 nucleos don't have external flash, I really think that needs to be pushed up.
internal flash I mean. I don't think it should be too crazy, it just needs cleaning up. It never got a lot of focus after the feather got added.
slender iron
slow but working is ok
if you have specific clean ups in mind then we can talk about those. general clean up doesn't always have an end
ionic elk
I'd like to add caching and track down whatever flash erase error happens sporadically on startup
Could also add the CCRAM cache for the F405s
slender iron
what kind of caching?
a FS cache in CCRAM is a bad use of the ram
because the FS is only read at the start of the python vm
ionic elk
Having a better intermediary buffer so that fewer writes go to flash immediately
I remember that I had some specific optimizations that I was going to do but then the feather came out and we all switched gears
slender iron
I don't think write speed is that important though
and a buffer will only help large writes
ionic elk
Ok, I won't push it too hard if you don't think it's priority. I've just been using the Espruinos today so I've had the sluggishness on my mind. But I'm excited about the F/H7s too so I'm happy to just get started on those.
slender iron
I'd rather improve VM speed than flash write speed
do writes take more than, say, 4 seconds?
ionic elk
When you save a file it takes about 5 seconds or so yeah
slender iron
hrm
ionic elk
The f767nucleo and h743nucleo don't have external flash. So I could do this concurrently
slender iron
lets do pulseio and rotaryio and then f7/h7
if f7/h7 are slow too then we can look at refining the internal fs support
slender iron
the tricky bit with a cache is that you have a window where a crash, reset or unplug can lose data
ionic elk
True. I'd have to get back up to speed on how NRF and Micropython do it
Could also take a look at STM32Duino
Cool, thanks @meager fog @slender iron for the discussion. Anything else we want to cover?
Picking up nucleos now
slender iron
we already do it external_flash so it may be worth factoring out from there
it is a 4k cache though
ionic elk
Ok, I'll check it out - happy to do that after the F7 as you suggested.
slender iron
👍
its a bit complicated because the cache is allocated in the vm heap when available and outside of it if not
the stm32f/h7 could be a better option for large and fast circuitpython programs
assuming the internal flash is faster than the imx rt's external flash
ionic elk
hmmm gotcha
marble hornet
hey, to convert a bin file to uf2 do i just run it through the Microsoft uf2conv.py ? https://github.com/microsoft/uf2/blob/master/utils/uf2conv.py
tulip sleet
yes, you can find it in the circuitpython repo too (maybe as a submodule)
marble hornet
thank you. i am trying to convert a bin but it's not outputing a uf2 it just says No drive to deploy.
solar whale
@idle owl I was just trying the CLUE library and the display_sensor demo requires adafruit_display_text. (its not in the list in the readme) Nice demo!! -- moving on to the others
ooh -- the spirit level is great!
hmm -- actually, I wonder if the motion of the "bubble" should be swapped 180 degrees -- in a real level the bubble "rises" up
tidal kiln
@marble hornet FWIW, here's how it's done in the makefile:
https://github.com/adafruit/circuitpython/blob/master/ports/atmel-samd/Makefile#L338
idle owl
@solar whale Thanks!
solar whale
great demos !! also display_shapes is needed for some as well...
idle owl
Those were just added, forgot about the README
@solar whale Fixed the spirit level, it now rises.
solar whale
Great! Thanks!
idle owl
I'll push it. Holding off on releasing because I have been tweaking things.
serene warren
What should be the format of an SD card for the weather station? I'm getting SD card not found I placed my icons on one and installed it but get the same sd card not found
idle owl
serene warren
and it doesn't show in File Explorer
solar whale
ladyada beat me to it !! works great!
idle owl
Thanks for testing!
solar whale
Glad to -- Thanks for the demo!
@idle owl do you also want to add display_shapes to the readme... sorry
idle owl
No, it's for a demo only, not the driver itself. I'm putting together a guide page that will have all the libs for all the demos and library included.
That's why I added the note about demos maybe needing more libraries.
Didn't make sense to add under driver-specific dependencies since it's not a driver dependency. At least to me.
solar whale
ah -- ok -- thanks -- that makes sense
@idle owl That spirit_level demo is a great example of how much can be done with a few lines of code in CircuitPython. Very impressive!
idle owl
Thank you! I'm actually really proud of that... I figured it out on my own.
idle owl
@solar whale Do you happen to know if the lines on a spirit level have a name? Google seems to imply no.
timber mango
Don't know about name. They are division markings for precision. A 'Precision' level has more. A good machinist ( a lot better than me) with a precision level can check level and tram of a roll on par with a laser setup.
solar whale
@idle owl not that I know. I’ve also seen this type referred to as a “bulls-eye” level.
meager fog
@indigo wedge hihi
im at a breakpoint if you wanna have me look at this feather
maybe mine lives 😄
indigo wedge
sorry, wasn't able to find free time this evening and now it's midnight 😦
but if you connect the jlink and exec the commands i sent you'll know if it's alive or not
meager fog
np ok
ill do that now for fun
note im on windows
so i cant run linux execs (easily)
indigo wedge
oh
that's gonna be a problem, i do everything on linux so don't know what works and doesn't on windows
but for jlink you should have same commands on both
meager fog
yah ill try those now
@indigo wedge ok i just need to load the feather firmware.elf using jlink?
indigo wedge
yeah, that's the uf2 bootloader, if that works then you can upload CPY using uf2
meager fog
woops jlkink too old
indigo wedge
yeah it needs to update the fw in the jlink to add imxrt1011 support
meager fog
yay
indigo wedge
if it worked you should see the uf2 disk
meager fog
looks like a disk
indigo wedge
oh
meager fog
but nothing there
meager fog
the native ROM one?
meager fog
the firmware.elf you attached
indigo wedge
i mean qspi flash chip, sorry my bad
indigo wedge
aww
meager fog
so basically same thing u got i guess?
indigo wedge
quite annoying since it's a copy of my design with fairly minimal changes
meager fog
indigo wedge
right, that's the ROM one
indigo wedge
i removed a few caps but don't think that would affect it
meager fog
doubt it...
indigo wedge
like i had some stability issues first but it was just long cables
i'll sit down during the weekend and do a side by side comparison to see if i missed something
indigo wedge
sorry about the delay 😦
meager fog
np - hardware is like that
indigo wedge
but it was supposed to work on first try! 😄
slender iron
@tulip sleet is the i2c weirdness on both nrf and atmel?
tulip sleet
yes, was detected on clue, but i'm testing on itsy m0
slender iron
interesting
tulip sleet
random choice: I thought my itsy was an m4, and i may have fried my metro m4
so it's an issue in shared-bindings
i think
slender iron
ya, that's all that changed in that commit
tulip sleet
i have another pending ping from someone, so i'll let you know when I'm back with any more info
tulip sleet
aha maybe:
in writeto_then_readfrom:
writeto(self, args[ARG_address].u_int, args[ARG_out_buffer].u_obj, args[ARG_out_start].u_int,
args[ARG_out_end].u_int, false);
readfrom(self, args[ARG_address].u_int, args[ARG_in_buffer].u_obj, args[ARG_in_start].u_int,
args[ARG_in_end].u_int);
but writeto and readfrom both call normalize_buffer_bounds, which take the address of the start,end, and length args
so the first may be confusing the second
slender iron
the transaction length stays the same though right?
slender iron
👍
raven canopy
lol. what are the odds of simultaneous review?
slender iron
🙂
timber mango
I found two library issues I would like to work on. One is from @idle owl and the other is from @gilded cradle . I commented on both and am working from my geekguy-wy Github account now for all Adafruit stuff.
gilded cradle
I responded with some more thoughts on the issue, but you'll have to figure out the remaining on more specifically how you want it to work, which is part of the fun. Enjoy. 🙂
timber mango
@gilded cradle I just read them. Yes, this is what attracted me to this issue in the first place. 🙂 I already have some ideas of the things I would like to do with this and it does look FUN!
steep oasis
Are there any power consumption values for BLE? I’m interested in values newer nRF52 boards.
slender iron
@steep oasis check out https://devzone.nordicsemi.com/nordic/power/
with circuitpython you won't get these values though because we don't sleep at all
idle owl
@timber mango Thank you for looking into issues! Let us know if you need any assistance.
timber mango
@idle owl I will. I have been looking for ones that I have the hardware to test with, which is not very many.
steep oasis
@slender iron Thanks. That’s what I was afraid of. Any plans to move to low power? Especially with BLE?
slender iron
@steep oasis I’d like to do basic low power sooner rather than later. I want to preserve ram though rather than completely shutting down. I’m not sure how big the power savings will be.
steep oasis
@slender iron I’m more on the electronics side of things. But let me know if you decide to work on it. I’ll help as much as I can. I see more and more work being done on the nRF52. I’m surprised nobody has tackled low power.
short phoenix
Hey sorry for the bad quality and (probably) stupid question, but my Feather M0 express running circuitpython 4 is not responding to code. I think I may have deleted something from the CIRCUITPY, or actual microcontroller but I don’t know what. I tried resetting and reinstalling circuitpython but it did not do anything. Any help?
raven canopy
@short phoenix do you get read-only errors if you try to save anything to the CIRCUITPY drive? it sounds like it may have gotten corrupted, which happens for a couple reasons.
short phoenix
Yeah i do
raven canopy
k. copy anything you want to save from the Feather to your PC, then run these commands:
upbeat plover
im doing some translations for zh_latn_pinyin, for "Already advertising." i can get close with somethat that translates to "Ads have been served." would this work?
raven canopy
import storage
storage.erase_filesystem()
run that in the REPL. it will wipe and reset the filesystem. then you should be able to put all of your relevant files back onto the board. make sure that you eject the board from your OS, before removing the USB cable. also, be wary of editing code files directly on the board; some editors do not complete writes before the board resets.
upbeat plover
hmm maybe something like "broadcasting" would work better
raven canopy
yeah. "Ads" doesn't quite equate to the context of BLE Advertising. well, i assume it doesn't; i don't speak Chinese. 😄
upbeat plover
yeah i went with "currently broadcasting" seems to work
short phoenix
@raven canopy so it will reset my device file system, and i will have to redownload circuitpython, install libraries, and then it should work?
raven canopy
@upbeat plover my feeble linguistics agree.
lone axle
you won't have to redownload circuitpython
raven canopy
yeah, it won't touch the firmware.
lone axle
but you will have to copy your libraries and code back to the device
raven canopy
yw!
and now, time for my late-night, RosiePi, small accomplishment post. the push notification subscription program had its first few successful flights:
(.venv) ~/Dev/physa_sub/physaci_subscriber:$> subscribe_to_registrar
Initiating physaCI registrar subscription...
Generating new node signature key...
Sending subscription request...
Successfully subscribed.
WooT!
raven canopy
the node-not-expired collision detection even worked with 4 month old, pre-planned code. can't have duplicates in the queue! 🎉
upbeat plover
"Cannot have scan responses for extended, connectable advertisements." another hard one
upbeat plover
hmm does "Cannot compile MP3" work for "Failed to parse MP3 file"? idk "parse" in chinese
manic glacierBOT
crimson ferry
The Adafruit_CircuitPython_Bundle February 07, 2020 auto-release is still transmogrifying from last night?
raven canopy
@crimson ferry the CI failed to upload the release. My quick-look-on-phone points to an upstream tool that is suffering from a recent GitHub API authentication scheme deprecation.
@idle owl I'll work this as soon as I can, unless someone else wants to. Problem is here, I think: https://github.com/adafruit/Adafruit_CircuitPython_Bundle/blob/master/.github/workflows/release.yml#L39
That action is still using the old token scheme that we just fixed with adabot.
idle owl
Hmm ok. @raven canopy I'll leave it to you as I haven't touched the actions stuff yet and don't need to break the bundle.
Or break it more anyway.
lone axle
Is there a known issue with the Github CI currently? I just had it fail in a way I haven't seen before. During "Install Deps" I ended up with these errors:
E: Failed to fetch http://azure.archive.ubuntu.com/ubuntu/pool/main/s/systemd/libudev-dev_237-3ubuntu10.33_amd64.deb 404 Not Found [IP: 52.177.174.250 80]
Fetched 230 kB in 0s (4477 kB/s)
E: Unable to fetch some archives, maybe run apt-get update or try with --fix-missing?
##[error]Process completed with exit code 100.
@lone axle those errors are usually transient and are due to some problem in the github actions cloud
lone axle
@tulip sleet I see, thank you. will it re-attempt itself automatically later? Or should I just make another push after a while to get it to try again?
tulip sleet
if you go to the actions page and click on the action that broke, on that page there is a dropdown on the right at the top of the list that lets you re-run the checks.
you don't need to push to make it happen
you have to click on one of the runs on the left first: example:
lone axle
Ah cool. I'll wait a bit and have it retry. Thank you
tulip sleet
you can try right away, it's usually quite transient
or wait until all the current jobs are done if you want
lone axle
Same error on the retry
I'm not sure if it's expected to be visible from outside of the Github / Azure cloud but I'm also getting 404 response from the URL for the .deb file referenced in the error when I try from my local PC. I'll give it another try here in a bit.
tulip sleet
ok, that may be a problem with their archives.
lone axle
Nice! if you are looking for testers, I can try it on a standard PyPortal later.
onyx hinge
@lone axle cool thanks!
as for the problem installing that package .. there was recently a security update to the package (feb 4): http://changelogs.ubuntu.com/changelogs/pool/main/s/systemd/systemd_237-3ubuntu10.38/changelog
this causes the old version of the package to be removed from the package servers. However, the package metadata on the ubuntu image hasn't been updated to know about the new package yet.
IMO the ideal case would be that github makes sure the metadata is up to date, but if my understanding is right it can be fixed/worked around by putting an "apt-get update" step before the first "apt-get install".
(some package management systems in linux download metadata every time, but debian's apt-get does not)
lone axle
I see. Is that something we have control of in the Github CI system? I'm not very familiar with how much of that is in the users control vs. up to Github to make the changes
onyx hinge
yes .github/actions/*.yml files within individual projects are where this apt-get line would be added
lone axle
I don't see any apt-get install commands in build.yml or release.yml but it looks like maybe it's executing a separate script that would be issuing the install commands:
- name: Install deps
run: |
source actions-ci/install.sh
what repo are you/we looking at?
onyx hinge
seems that it is referring to this file https://github.com/adafruit/actions-ci-circuitpython-libs/blob/master/install.sh
lone axle
Ah, so I could add a section like this before it:
- name: apt-get update
run: |
apt-get update
would that need to go in build.yml, release.yml, or both?
onyx hinge
I think that if you added it to that install.sh then it would be automatically picked up by ALL repos that use actions-ci-circuitpython-libs
@lone axle remind me your github so I can mention you? I'll PR this change I mean...
lone axle
Oh, that does sound like a better option.
onyx hinge
weird there are LOTS of audio glitches on pyportal pynt that I wasn't hearing on pygamer 😦
lone axle
I am FoamyGuy on github
onyx hinge
okay
lone axle
I've still got it on my list of stuff to give some more time to testing but I was getting pretty different results between PyGamer and NeoTrellis M4 when playing back audio
On the NeoTrellis I though it was related to Mixer voice 0 vs voice 1, but started to make a self contained example and testing on the PyGamer and wound up with different results. I'm going do some more testing and try to get my findings documented this weekend.
onyx hinge
is neotrellis also an M4 based board?
onyx hinge
these audio glitches are ones I think means the board isn't quite keeping up with mp3 decoding
maybe the screen having about 2x as many pixels is enough, it feels pretty "marginal" on pygamer
meager fog
@onyx hinge 0ortibg 🙂
lone axle
In my case they were wave files instead of mp3.
onyx hinge
wave files on flash storage should be solid, but I've been wrong before.
on sd, I don't expect they work very well
lone axle
Flash storage in my case
manic glacierBOT
slender iron
@tulip sleet I'm through almost everything. need anything else?
tulip sleet
that's great! thanks
orchid basinBOT
tidal kiln
@slender iron good luck at pycascades this weekend!
slender iron
Thanks @tidal kiln ! Headed out now
tidal kiln
anyone else seeing the library download buttons broken on the webpage?
hmmm. looks like the release assets aren't there for latest.
manic glacierBOT
idle owl
@tidal kiln Yeah known issue.
tidal kiln
okie doke
tidal kiln
@idle owl do you know if theres a good basic hello world example for receiving text from BLE Connect App?
idle owl
@tidal kiln No idea. I haven't had much to do with that end of things.
tidal kiln
ok. no worries.
idle owl
The issue with the bundle affects all releases on all libraries. If you release a lib, the actions job will fail to release and upload assets. This is currently being looked into.
tidal kiln
@tulip sleet ping
short phoenix
Running some example code for my featherwing OLED and Feather M0 Express from https://learn.adafruit.com/adafruit-oled-featherwing/circuitpython-and-python-setup, and keep getting the error "AttributeError: 'module' object has no attribute 'I2CDisplay'" in the serial display. Any help?
tidal kiln
@short phoenix what version of CP are you using?
tidal kiln
yep. i think it's 5.x only.
short phoenix
Yeah it is 
tulip sleet
@tidal kiln are you looking for a BLE Connect text sample app?
tulip sleet
https://github.com/adafruit/Adafruit_CircuitPython_BLE/blob/master/examples/ble_uart_echo_test.py. But we're always looking for a packet on the receive side, and there are no "text" packets.
so you'd have to read raw stuff and then parse it into a color packet if it looked like one.
tidal kiln
ok. thought that might be the case.
tulip sleet
https://github.com/adafruit/Adafruit_CircuitPython_BLE/pull/14 someone else wanted kind of the same thing
tidal kiln
so just read everything raw, and can borrow code from color packet
look for !C at start, etc?
tulip sleet
yes, you can use Packet.from_bytes() to read a packet and generate the proper object. You need to register the packet class by importing it first
tidal kiln
oh. i was just going to uart.read and then just work with that.
tulip sleet
yeah, you would do that, and then once you recognize the color packet, you can call from_bytes() on the accumulated bytes (if you want)
passing in the correct number of bytes
are you sending from the official app, or from something else?
tidal kiln
official app
tulip sleet
it's a little bit too bad we don't have a text packet type, but we'd have to add support in the app (as opposed to just the raw UART)
tidal kiln
has anyone ever suggested a "text" something for the control pad?
tulip sleet
that PR above kind of implies it
I originally thought there was such a thing, and had to be deconfused by limor
apparently the typical use case was sending text back to the phone, not sending text to the device
tidal kiln
from what i can tell, the packets are nothing but special text strings
tulip sleet
that's right, and the original idea of the app seems to be to -send- stuff like accelerometer and color packets, not receive them
tulip sleet
that shows whatever text is sent back from the device (not the app)
that's what I was confused about
it's just raw text
oops
tidal kiln
does this look correct in terms of doing the raw read?
while True:
ble.start_advertising(advertisement)
while not ble.connected:
pass
ble.stop_advertising()
while ble.connected:
if uart.in_waiting:
msg = uart.read(uart.in_waiting)
print(msg)
looks good to me, but it's possible (not sure) a packet might get split
i have to go out, sorry
tidal kiln
ok. no worries. thanks for the help. mainly wanted to make sure i wasn't overlooking something obvious.
short phoenix
okay one more thing @tidal kiln, my text is not displaying when using the example code in CP 5 latest. I think I am missing a library maybe?
tidal kiln
you should get an error if you are missing a library
short phoenix
I am not getting any errors, it just shows the outline as mentioned but there is no text
tidal kiln
are you trying the example code here?
https://learn.adafruit.com/adafruit-oled-featherwing/python-usage#example-code-8-14
short phoenix
yes
tidal kiln
hang on a sec and let me give it a try...just happen to have a Feather M0 and OLED wing sitting here...
short phoenix
there is one line of pixels of text in the top left corner
but i tried shifting x and y values to get it to center but it would only shift the border
I am pretty sure the text is just sitting in the top left by default and hasn't been moved to center
tidal kiln
looks like that example needs to be updated
not sure when, but x and y can't be passed in anymore
need to set them after
try this:
text_area = label.Label(terminalio.FONT, text=text, color=0xFFFF00)
text_area.x = 28
text_area.y = 15
golden sail
hello, i am using itsybitsy m4 express microcontroller and I am trying to connect it to esp8266 wifi chip. I noticed it says that the ESP 8266 is not supported by circuitpython and that in order to use it, it has to be used as a wireless co-processor. Can someone explain how this works please :)?
short phoenix
Thank you cater
golden sail
and would it still be able to receive data from the itsy bitsy m4, and send the data over wifi?
tidal kiln
@short phoenix np. there's also some fancier anchor options that were recently added. worth checking out:
https://github.com/adafruit/Adafruit_CircuitPython_Display_Text/blob/master/examples/display_text_anchored_position.py
@golden sail it works via special firmware that runs on the ESP8266. you then talk to the ESP8266 using protocols defined by that firmware. i think this guide covers most of it:
https://learn.adafruit.com/adding-a-wifi-co-processor-to-circuitpython-esp8266-esp32/
short phoenix
Can an Airlift act as a wifi coprocessor?
tidal kiln
do you mean this?
https://www.adafruit.com/product/4201
if so, yes. that's basically an ESP32 with the special firmware preloaded.
i think it was found that the ESP32/SPI combo worked better than the ESP8266/AT combo, which is why everything has been using that, like PyPortal, etc.
from that guide:
If you can choose - use ESP32 with SPI Commands - it's the best experience by far.
do you have any hardware already?
golden sail
using itsybitsy m4 express, Adafruit Ultimate GPS Breakout, and mpu6050
gps and mpu650 will send data to itsybitsy m4 express
and now im looking for a way to send that data to a smartphone over wifi.
golden sail
i have an esp8266
lone axle
I believe there is no library for the esp8266 atm, only esp32.
tidal kiln
good point. in terms of CP support. only the ESP32 one:
https://github.com/adafruit/Adafruit_CircuitPython_ESP32SPI
i think given what you currently have, the airlift should work.
GPS on UART, MPU on I2C, ESP32 Airlift on SPI
sercom party!
@tulip sleet fyi - i got something working. thanks again for your help. pointed me in the right direction.
short phoenix
I am planning to make a game on the Featherwing OLED and Feather M0 Express (basically the chrome dinosaur game) and use the A B and C buttons on the OLED to control it. It will obviously be fast moving and I will need to control it using buttons. What libraries will I need?
meager fog
@tidal kiln i think the MCP2221 fix ya came up with worked out?
you could merge in a 'final' fix with values
tidal kiln
yep. i hope so. seems to be working.
what's your thoughts on value? seems like most were able to get by with under 0.5. but could just crank it up 1?
manic glacierBOT
lone axle
@short phoenix this page will get you started on which libraries you need in order to draw to the screen: https://learn.adafruit.com/adafruit-oled-featherwing/circuitpython-and-python-setup
It looks like the 3 buttons connect to 3 digital pins, you don't strictly need a library to use them. But you could use the Debouncer library to make it a bit easier if you wanted.
idle owl
<@&356864093652516868> Due to a recent deprecation of a GitHub API authentication procedure, many services that relied upon it are not working. Release mechanisms for both the Adafruit and Community CircuitPython bundles, and each library are broken. Any releases will not have the proper release assets. For now, please point anyone looking to download the CircuitPython Library Bundle to the last successful release here: https://github.com/adafruit/Adafruit_CircuitPython_Bundle/releases/tag/20200206
timber mango
@gilded cradle DId you create the image that shows the 16-bit format of the segments for the alpanumeric display?
gilded cradle
@timber mango, yes
onyx hinge
@idle owl ouch. any meaningful way we can help?
idle owl
@onyx hinge Not at the moment. Sommersoft is looking into it. I'll let you know if there's anything you can do.
timber mango
@gilded cradle Wonderful! Could you please add the decimal values for each bit under each segment letter? Like for segment A would have "1" on a slanted line under the box, up to segment N which would have "16384" It will make it easier for people to add the bits to make a bitmask.
gilded cradle
I'm not quite sure what you mean, but now that I look at it, I noticed that it shouldn't start with 0x as that implies hex, when it's actually binary.
onyx hinge
yay! my LED demo went from 15fps -> 30fps with pixelbuf, and now it went to 60fps with ulab
it's "just" 96 pixels but it's still pretty cool to go 60fps with circuitpython
lone axle
that is amazing
timber mango
Yes, the way it is shown in the image is actually binary, so should be prefaced by 0b instead of 0x. I mean to add the decimal values of each bit under the box for each segment. 1 for segment A, 2 for segment B, 4 for segment C ... up to 8192 for segment M , and 16384 for segment N.
Then, people just add up the values for the bits of the segments they want to light up and that is the bitmask for that character or animation frame.
gilded cradle
I think putting all those numbers will make it look a bit more cluttered than I would like it
It's more meant to be a reference for which segment is controlled by which bit than a binary to decimal converter.
timber mango
It is not really a binary to decimal converter though. It is a bitmask constructor. Without this information, I think a lot of people might not understand what a bitmask is, especially those who are completely new to this.
How about this: A wheel diagram with 15 segments. The display segment letter (A, B, .. M, N) on the outside edge and the bit values below that. Think of an umbrella.
The Wheel of Fortune wheel is a much better example of what I am going for.
onyx hinge
timber mango
@onyx hinge Impressive!! 😃 👍
onyx hinge
😊
short phoenix
Sorry for asking so many questions (I'm new to this as you may be able to tell)... but I am trying to use this guide https://learn.adafruit.com/circuitpython-oled-knob-sketcher/drawing-pixels in the latest circuitpython 5 build with an OLED Featherwing and Feather M0 Express. This guide may be made for Arduino or outdated, so my question is what would be an updated version of how to draw even just one pixel in circuitpython?
onyx hinge
I tried to design an FIR band pass filter using scipy and all I got was this weird moray mouth
timber mango
lol
onyx hinge
hey that looks like a band pass filter
too bad a 97 pole filter is probably infeasible to run even with ulab supporting it
tidal kiln
@short phoenix responding to your questions over in #help-with-circuitpython , which is a better channel for general questions. this channel is more for developer questions about the circuitpython firmware
short phoenix
Gotcha, thanks for everything 🙏
tidal kiln
sure. no prob. i'm outta here. have fun with your project!
steady coral
can anyone help I feel like I've lost my mind lol
was going to try downloading the latest circuitpython library for 5.0.0
I go here: https://circuitpython.org/libraries
click latest releases
steady coral
ah thanks!
lone axle
If you scroll up a bit there is note from Kattni explaining a bit more
steady coral
ah ok I missed that thanks for the heads-up
woven nova
Is there a support for the CANbus protocol on CircuitPython?
stuck elbow
no
lunar crown
@onyx hinge What program are you using to design/plot your DSP filters?
onyx hinge
@lunar crown those were from scipy.signal.firls.
And the plot is also by matplotlib iirc, based on a scipy sample
Parameters poorly chosen by me
@idle owl maybe pin that message about GitHub problems?
lunar crown
@onyx hinge Yes, balancing the parameters is almost an art. Thanks.
onyx hinge
the background is, we are in the process of adding "ulab" to accelerate numeric calculations in circuitpython.
ladyada asked me to look into doing low pass / high pass signal filtering with it so I'm literally just learning since yesterday.
raven canopy
@onyx hinge i've got the release issue cause pinned down. i'll put up another message, and pin both.
onyx hinge
@raven canopy thanks for burning the midnight / weekend oil on this!
raven canopy
its the only oil i have available, generally. 😄
also, i used this when i was trying to figure out filters: https://fiiir.com/
lunar crown
@onyx hinge For real time filtering, probably better to make some C-Extensions to CP and expose selected functions from the ARM filter library. I am trying to learn how to do a C-extension on the Teensy4 / M7F.
onyx hinge
ulab is in C but does not use CMSIS functions, it uses portable C.
lunar crown
I understand the tradeoffs. More general solution vs. higher performance.
onyx hinge
cmsis arm_fir_m32 looks quite optimized compared to ulab.convolve, but will have bigger code size
raven canopy
<@&356864093652516868> Update on release asset uploads: we've narrowed down the cause of release asset uploads being broken. The GitHub API Authentication was actually a red herring; thanks to @ruby atlas and @idle owl for guiding me away from that. The actual cause turned out to be a change to the 3rd-party release asset upload action we're using. Issue has been filed to that repository, and the maintainer is engaged on a fix PR. I'll be putting the PR in shortly, so hopefully this will be fixed rather soon. Thanks for your patience! https://github.com/csexton/release-asset-action/issues/6
lunar crown
@onyx hinge Where is ulab being developed. I didn't see it in the master branch.
raven canopy
@lunar crown its currently in this pull request: https://github.com/adafruit/circuitpython/pull/2583. ulab is external. i think jepler is working on adafruit's fork, and may be sending things upstream when possible. https://github.com/adafruit/circuitpython-ulab
lunar crown
@raven canopy Thanks
half sedge
@onyx hinge any "Bilinear interpolation" in ulab? I would love to double the number of pixel when displaying thermal camera image...
bright aspen
What is the status on I2S on Feather STM32405 Express? (I need to choose a small board such as that or Feather M4 Express.)
idle owl
@raven canopy Thank you for you work on this!
meager fog
@raven canopy yay good 🕵️ work
raven canopy
it was a team effort!
onyx hinge
@half sedge I didn't see any interpolation function like that.
@half sedge the upstream was receptive to the first function we wanted to add (convolve), so maybe look at how you would do bilinear interpolation in numpy and maybe we can suggest it or even pull request it
half sedge
@onyx hinge Maybe this is too much math for me. 😉 I was thinking that the Raspberry Pi simple example was using numpy for that... but it use PIL: https://github.com/adafruit/Adafruit_CircuitPython_MLX90640/blob/master/examples/mlx90640_camtest.py
I am pretty sure I saw somewhere an example of MLX90640 that was inventing pixel values to double the resolution. Maybe that was AMG8833 but that is Arduino code: https://learn.adafruit.com/thermal-camera-with-display?view=all
tidal kiln
@meager fog pong. ok can for a bit.
tidal kiln
@half sedge yah, i think it was AMG88xx. the guide above used Arduino:
https://github.com/adafruit/Adafruit_AMG88xx/tree/master/examples/thermal_cam_interpolate
theres a CP example that uses numpy on rpi:
https://github.com/adafruit/Adafruit_CircuitPython_AMG88xx/blob/master/examples/amg88xx_rpi_thermal_cam.py
which probably also could have used PIL
@meager fog sry. gotta scoot. can chat l8r.
twilit geyser
How do you guys edit code? I’ve been using VSCode directly on the CPy device, but occasionally it hangs and I don’t like that.
lone axle
Mu when working with code directly on a device, PyCharm for code stored on the PC
meager fog
@tidal kiln all good - sent mail
half sedge
Thanks @tidal kiln , it seems there is more than one way to kill a cat, now I need to find a Circuit Python way...
river quest
happy saturday folks - here's the DRAFT of the upcoming “Python for microcontrollers” newsletter that ships out tuesday at 11am ET via adafruitdaily.com - please send any links, news, events and more via an issue/PR, or @ us or really any thing 🙂 we'll get it added!
tulip sleet
<@&356864093652516868> I am deleting or marking as pre-releases the bad library bundle releases, so that https://github.com/adafruit/Adafruit_CircuitPython_Bundle/releases/latest will point to the last good release. I'll try to keep it that way manually until we fix the bundle building problem.
timber mango
Greetings everyone.
In a library, if I want to have some constants available to the outside world, is there a preferred way of doing that?
@tulip sleet Did Github cause this problem?
tulip sleet
It was third party library. See pinned messages in this channel for details
timber mango
Ah, OK, I had missed that part.
I have been absorbed in working on animations for the HT16K33 and alphanumeric display.
timber mango
@gilded cradle I have something to show you.
lime trellis
@meager fog sweet! it's very minor, but it works well.
timber mango
Snakes on a satellite!
meager fog
@lime trellis do you have an example library you're using it in?
lime trellis
@meager fog not yet- but I'll be having students use register for the BMX160 IMU library they're building. I confirmed it works, tho
I can send you the quickie code I used to check it on the ADS124S08 and SX1280 if that'd be helpful? I use it exactly like I use i2c_bits in my https://github.com/maholli/CircuitPython_BQ25883 library
meager fog
@lime trellis thing i want to try to get working is dual/use SPI or I2C
we have a lot of dual use sensors -
maybe BMX is too?
lime trellis
(you should totally sell a breakout btw, it's a fun little IMU so far)
meager fog
its the chip inside the BNO055 right?
so like https://github.com/adafruit/Adafruit_CircuitPython_DPS310/blob/master/adafruit_dps310.py#L49
maybe instead of adafruit_register.i2c_struct we'd have adafruit_register.i2cspi_struct
lime trellis
holy cow!! if the BNO055 just a BMX160 with an M0 in it? I need to look into this further. Might give the students a great starting point for the library if they have the same calls.
@meager fog agreed on the adafruit_register.i2cspi_struct. I didn't port over any struct stuff since I wasn't using it.
Would it still be helpful for me to send over a library with spi_bits implemented?
lime trellis
I'm just not sure if there are SPI devices that don't use the register-access scheme I described in the PR? I couldn't think of any offhand
meager fog
it isnt the classic 8-bit-for-RW?
lime trellis
As far as I know. I also implemented reading more than 1 register at once in the spi_bits, but that's not necessary if you wanted to keep it simple.
For SPI:
meager fog
oh yuk
tahts different lol
every company re-invents!
still, does BMI have the same register map for i2c and spi?
lime trellis
Yes- only one register map that I know of
meager fog
@lime trellis well we have various spi modes in busio
for arduino
see....
if you're up for implementing and adding yours as a third option (for cirucitpy) it would be a mega mitzvah
but the idea is that it would allow us to use i2c or spi in a tidy manner
lime trellis
@meager fog oh that seems super doable! It won't be able to be that tidy I don't think, but I'll whip it together when I get a chance and you can be the judge
meager fog
id be in your eternal debt 😄
cause i was going to implement it - but got so much on my plate 🙂
did you try out the ulab PR btw?
lime trellis
happy to help :)
and no I haven't touched-base with the Andrew lately on ulab implementation. Enough on my plate that I'm trying to not get involved 😃
meager fog
we did a PR for circuitpy
you could try it out hehe
no involvement needed just...distractions!
lime trellis
omg they will quite literally shout for joy! 😆
slacking them now. Darn now I'm going to have to come up with something else to grade them on this quarter! 🤣
meager fog
oh they can add plenty o things to ulab
tough flax
Hi folks, is anyone working on making CP work as a keyboard in "BIOS mode"? I think that means HID with no other profiles & no other USB descriptors... but I'm not sure. I know that a user's having trouble getting our custom keyboard interface boards to work through a KVM and suspect that this would solve it... hoping not to start from scratch https://www.facebook.com/groups/ATMakers/permalink/815702142174560/
upbeat plover
not sure why but when i looked at https://github.com/adafruit/circuitpython/pull/2600/conflicts its like at line 785, but when i look in the file its line 682
timber mango
With the HT16K33 alphanumeric display animation I am working on it may be possible to make simple games. I have not tested this out though because I am not a game maker. Is any game maker interested in trying this out?
I suggest using this with an M4 or faster micro.
timber mango
When making changes to something I want to contribute back to master, should I be working on the original or a cloned fork of the original?
solar whale
@timber mango use a fork then create a branch <https://learn.adafruit.com/contribute-to-circuitpython-with-git-and-github
timber mango
OK, that is what I thought and what I am doing. How can I keep in sync with the Adafruit master?
Reading the learn guide now.
solar whale
it's all in the guide -- fetch/merge from the remote repo
timber mango
I want to have a class C with a class variable N, but I want N to be immutable. If I have x=C(), I could also have x.N to access the variable. Could the instance modify class variable N. I do NOT want to allow x.N = 2.
onyx hinge
I don't know how you make a member of a class not assignable. You could give all instances of a class an immutable member with an @property, but then you can't get that value given only the class.
timber mango
I just want certain members of the class to be immutable. Just a few variables.
onyx hinge
"all members of a class" -> "all instances of a class"
# Give all instances of C the immutable property N with value 3
@property
def N(self): return 3
but then C.N is a property object, not the number 3
onyx hinge
the only way I can find involves a python feature not in circuitpython (metaclasses)
@property
def N(self): return 3
class C(metaclass=CMeta):
@property
def N(self): return C.N
print(C.N)
x = C()
print(x.N)
### Either of these lines will cause an AttributeError
#x.N = 9
#C.N = 9
but it works in desktop python3
timber mango
I could use a dict, but that seems awfully clunky for this purpose. I do not want to make a user do C.dict["N"] to get the correct value.
raven canopy
i would go the @property route, as well.
timber mango
Hmmph.
onyx hinge
I wouldn't worry about, "what if somebody goes assigning C.N or x.N in contravention of the docs". In general this is going to make things breaky-breaky
(my snippet based on answers at https://stackoverflow.com/questions/5189699/how-to-make-a-class-property which maybe has an idea that would work in cpy)
timber mango
That looks awfully MESSY to me and over complicated for what I want to do. I just want a CONST N = 2 such that N can not be modified.
onyx hinge
don't worry, 2 can't be modified no matter what you do 🙂
raven canopy
you can use from micropython import const; N = const(2)...
didn't realize you were wanting a const-like thing.
even though immutable variables kind of go against CPython's ethos. 😄
timber mango
@raven canopy That does not make N immutable though.
@raven canopy I know. I am trying to combine an FP feature with a non FP language.
raven canopy
i mean const() is available. use it. we're in the gray area a lot around here, since we have the constraints that PCs don't. hehe
and yeah, 'immutable' was not the word i wanted. 'static' is more what i wanted...
half sedge
Hi @idle owl , I know you are working on a blog post for mlx90640 camera. I have improved the mlx90640 demo for the PyGamer: https://github.com/dglaude/CircuitPython_MLX90640_PyGamer_Plus It is not yet at the level of the "DANIU HY-18", but I believe it is possible to make a very similar UI and match the feature. That one use STM32, likely not using CircuitPython and has a 320*240 screen. Maybe on the PyGamer we will not have interpolation, but we have SD card support that could be great for saving some picture.
timber mango
My brain is breaking, so I need a nap. I will be back later.
raven canopy
hmm. you're right though. i guess i've misunderstood const() this whole time. my bad!
onyx hinge
I think what X = const(123) does is, allow replacement of X in an expression with 123; and also in some circumstances it CAN signal an exception when X is assigned (but not in general). It works for module level variables, and I think the protection against assignment is only when evaluating the module's code, not something that imports it.
tulip sleet
viscid pollen
I tried to use the BLE libraries to have one central CPB connect to two CPB's that act as peripherals. Is it possible? I just want to be able to send messages to both of the peripherals from the one central CPB. Thanks in advance.
tidal kiln
@tulip sleet fyi, here's what i ultimately came up with:
https://github.com/adafruit/Adafruit_Learning_System_Guides/blob/master/TFT_Gizmo_Candy_Hearts/candy_heart_ble.py#L88
tulip sleet
@tidal kiln that seems pretty simple. It's slightly possible that raw_bytes will be an incomplete packet, but probably not in practice.
tidal kiln
it looks like Packet.from_bytes() at least has some checks? or do you think more is needed in the user code to handle a potential incomplete packet?
half sedge
@tidal kiln are you trying to transfer images over BLE?
tidal kiln
nope. just some basic stuff, from the BLE connect app.
tulip sleet
if it's not a disaster if there's a lost packet, no issue
half sedge
Ok.
tidal kiln
yah. it's not.
@half sedge the main thing was being able to receive generic text from the UART interface in addition to the other "special" packets that get sent by the Controller interface (like the Color Picker)
in the example above, i only cared about the ColorPacket
but the if instance(): could be expanded for others
half sedge
Maybe some "Text Message" should be added to here: https://github.com/adafruit/Adafruit_CircuitPython_BluefruitConnect
I don't know if it exist in the Arduino version and in the App however.
It seems possible to register new Packet Type to enhance the "over UART" protocol.
tidal kiln
you could, but there is no special "text" packet in the BLE Connect App side - the app that runs on the phone
there's only a simple UART that that sends "raw" text
so the above example is one way to handle that
onyx hinge
good morning
tulip sleet
@idle owl @raven canopy et al Each time the library bundle generates a bad release, I am editing it by putting a warning at the top and marking it as a pre-release. Making it a pre-release means that https://github.com/adafruit/Adafruit_CircuitPython_Bundle/releases/latest skips those and points to the last good release. I was deleting the releases but this is a bit better because it preserves the release notes.
maybe we should fork and fix https://github.com/csexton/release-asset-action at least temporarily?
timber mango
Is (total memory of micro) - gc.mem_free() an accurate measure of how much memory a script needs after it loads?
tulip sleet
@timber mango gc.mem_free() is reporting how much heap storage is left for creating objects after doing imports. The total RAM memory is not too interesting: there are static RAM areas and the stack in there, so if you consider those you are just considering overhead. On the M0 boards, the stack is about as small as it can be (4-5kB). On M4 boards we allocate about 24kB for stack. You can adjust the stack size on restart: https://circuitpython.readthedocs.io/en/latest/shared-bindings/supervisor/__init__.html#supervisor.set_next_stack_limit.
timber mango
I am just curious how much RAM a script requires after all imports and instantiations are done.
tulip sleet
you can call gc.mem_free() (or gc.mem_alloc)()) just after all the imports, and then call it various places. But your program is probably creating objects while it's running. When it runs out of heap, it will garbage collect. There is something called gc.mem_info() which gives more details but we have turned it off to save space.
timber mango
Ah, OK.
tulip sleet
if you are worried about space the best thing to do is to figure out how not to allocate new objects that live a long time (e.g. complicated lists and tuples). Allocate buffers once and then reuse them. If you have a significantly-sized static data structure, considering encoding it in a bytes() or reading it from a file rather than coding it as a nested set of tuples or lists or whatever.
timber mango
I am not worried about space. It was just a curiosity that popped into my head when I was sleeping. Odd things like that happen when I sleep.
tulip sleet
when we do we a build for a particular board, there is some stuff printed that is interesting. Here's part of the output from a Trinket M0 build:
2512 bytes free in flash firmware space out of 188160 bytes (183.75kB).
23836 bytes free in ram for stack and heap out of 32768 bytes (32.0kB).
Space is too vast to worry about.
tulip sleet
so about 4kB of RAM is used for stuff other than heap and stack. The stack is about 4k, so the rest is heap
timber mango
Interesting. So, that shows the maximums available for scripts and stuff?
tulip sleet
yes; if you just started a REPL and didn't import anything, gc.mem_free() would tell you total available heap. But if you do any I/O, there's a 4kB buffer that's allocated pretty immediately. The command history in the REPL also uses up some up. (You'll see it shrink as you type more commands.)
Adafruit CircuitPython 5.0.0-beta.5-4-ga63f49cb8 on 2020-02-06; Adafruit Metro M4 Express with samd51j19
>>>
>>> import gc
>>> gc.mem_free()
159264
>>> gc.mem_free()
159168
>>> gc.mem_free()
159104
>>> gc.mem_free()
159040
>>> gc.mem_free()
158976
>>> gc.mem_free()
158912
>>> gc.mem_free()
158848
Yes, I did see that. I have a script that is printing gc.mem_free() at the start of each loop. I can see when gc is done.
On another note, I am experiencing a long delay from when I save a script to flash to when it actually starts running. I get ```main.py output:
soft reboot
Auto-reload is on. Simply save files over USB to run them or enter REPL to disable.
main.py output:
OSError: [Errno 5] Input/output error
Press any key to enter the REPL. Use CTRL-D to reload.``` right after I save a script.
tulip sleet
@timber mango what editor and what OS are you using?
meager fog
that's often because the file is not 'flushed' by the editor
@tulip sleet we might wanna elucidate that error message
it comes up for me on win10 a lot
timber mango
Editor is Geany and I run Raspbian Buster on a Raspberry Pi 4.
tulip sleet
i think it might be because the FAT metadata is in a bad state.
Geany probably doesn't flush to disk immediately on write. See https://learn.adafruit.com/welcome-to-circuitpython/creating-and-editing-code#1-use-an-editor-that-writes-out-the-file-completely-when-you-save-it-7-13
@meager fog we could add "... make sure your file is completely written", but it's a pretty general error that could happen in other cases (like filesystem corruption)
ah, I see I even tested geany and it doesn't flush to disk
it's in the not-recommended list
you could lobby the geany maintainers to fix that, or if they have plugins you could write a simple plugin that does a sync.
timber mango
Hmmmmmm. I have not found any other editor I really like.
tulip sleet
if you type sync in a terminal window after writing the file that will speed up the flush
but it's best if the editor does it
timber mango
Github question: I went to push changes to fix an issue and Github wants me to login every time I try to push. This is on my second account. This does not happen on my main account.
@tulip sleet Ah, OK.
timber mango
OK, reading, thanks.
tulip sleet
raven canopy
@tulip sleet we could point to a fork, but there is an easier way by providing the step with an extra variable (the breaking change). With the maintainer's quick response on the issue, I thought the PR would move quickly (I F5'd it all weekend 😄)
timber mango
@tulip sleet OK, thanks. I will have to mess with this later.
onyx hinge
I'm grumpy today. grumpy at my own nrf pwm audio out code, which is not well structured and resistant to adding ramp up/down code.
this looks good, right? Well, there's a hum during the ramp time and still clicks at the start and end 😦 😦
meager fog
@onyx hinge theres no way to avoid it realy with PWM because by definition you'll get an initial square wave when the first pwm starts
square waves are like juggernauts 😄
onyx hinge
should I just move on? users are grumpy on the forums, and they should be -- it's a BIG pop with the PAM8302A amplifier which is on crickit (I'm using a breakout tho)
I would love for it to be better than it is
meager fog
whats the link to the forums?
onyx hinge
I think there are a lot of things I don't understand about how PWM audio works out in practice. Theory says, you can't hear a 62.5kHz carrier frequency. Practice says, I hear something just fine.
meager fog
@onyx hinge yes! square waves have harmonics
meager fog
they leak below 20khz and we cant effectively filter them, you really want the PWM freq to be 2MHz (100x highest freq)
pastel panther
not sure if I won or lost the CLUE lottery but I got a black one when I was hoping for a green one. If anyone wanted black and got green, I'd be happy to trade
ionic elk
@slender iron quick PR report: core temp should be ready now, F407 is ready for your re-review, and I have a question on your espruino wifi request.
slender iron
@ionic elk thanks for the update. I'm a bit behind on everything since I was at pycascades all weekend. Will get to everything by the end of today. I have meetings this morning
ionic elk
@slender iron np take your time! just mentioning it since you said it was good to ping you
slender iron
yup, thanks!
idle owl
<@&356864093652516868> Here's the notes doc for the Weekly meeting in under an hour. Please add your status updates and hug reports even if you're attending. Thanks! https://docs.google.com/document/d/1CTN9esAIa35AYV7EVIBUsa9nZsf_vzcJTJIvZqtiPgc/edit
