This is the problem. The guide does not seem to specify when it makes a switch and what it is talking about in a given section.
#circuitpython-dev
timber mango
solar whale
I just ignore any references to the MODE switch.. IIRC, it is only referenced with respect to using a Serial UART which you are not doing with an SPI Friend. You can provide feedback to the guide.
timber mango
OK, I will do that, because this has confused me. I am a beginner when it comes to writing my own code to use Bluetooth.
Thank you.
lethal abyss
sorry wrong place, I'll remove and put it in #help-with-circuitpython
timber mango
@solar whale I have that library and am working on learning to use it.
I could not find any Circuitpython examples for the Bluefruit LE SPI Friend.
timber mango
@solar whale So, I cloned the repo for the libraries (Circuitpython and Arduino) and am creating my own Circuitpython scripts for the Bluefruit LE SPI Friend.
@tulip sleet Thank you. I just saw your link and can not imagine how I managed to overlook that. ๐ ๐
I think I need a nap.
crimson ferry
Has anyone seen instability on Teensy 4.0? I'm trying my app on Teensy (5.0.0, latest libs), and it hangs hard after just a couple of loops doing nothing but checking elapsed time to see if should do an action. I'll try to get a small reproducible code, but thought I'd sense check with those further on the curve.
slender iron
@crimson ferry I haven't used it long enough to know. It's all early so it wouldn't surprise me.
crimson ferry
OK, thanks, If I can isolate, I'll file an issue.
slender iron
thanks!
crimson ferry
Good news is... it makes it through all of the setup: SPI, I2C, Storage, RTC, ESP32SPI, HTTPS, etc. ๐
timber mango
@tulip sleet I had to change a few pins, but bluefruitspi_simpletest.py works! ๐ ๐
solar whale
@crimson ferry I have had Teensy 4.0 (CP5 latest master) running with a radio (rfm69) for several hours at a time - I have had a few occasions where the REPL hung (could not control-C out) but it kept running.. have not dug into that yet.
crimson ferry
@solar whale Thanks, I'm also not able to ctrl-C out, but it seems to hang up USB too and I have to power cycle. I'll work on narrowing it down.
onyx hinge
@slender iron When you pin the new meeting link, can you promote the new calendar file too?
manic glacierBOT
onyx hinge
set my phone to CET (mainland europe time) and the meeting times still look right. should have checked it sooner, glad it's correct
slender iron
@onyx hinge yup! I think @idle owl was going to make the notes doc for next week. We can add it there.
onyx hinge
awesome thanks both
idle owl
<@&356864093652516868> Here is the notes document for Mondayโs CircuitPython Weekly meeting. Everyone is encouraged to attend! Please add your hug reports and status updates even if youโll be attending the meeting - itโs super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and weโll read them off during the meeting. Hope to see you there! https://docs.google.com/document/d/1Ev0a-zNbTzmMbo__zPvqba9rX3kdfS6LahXAfJQveu0/edit
@onyx hinge Please add info to the top.
slender iron
thanks @idle owl
onyx hinge
done, re-word if it's too clumsy. Now we just need to remember to update this file if we ever vary the date, and to regenerate it in 2021
idle owl
Looks fine to me
onyx hinge
The meeting schedule is also in an โicalendar/icsโ format calendar, which works with many calendar packages. Copy and paste this URL into any compatible software: https://raw.githubusercontent.com/adafruit/adafruit-circuitpython-weekly-meeting/master/meeting.ical
slender iron
<@&356864093652516868> Here is the notes document for Mondayโs CircuitPython Wee...
lone axle
quick question on Github PR reviewing procedures. If I find a PR that has already been merged but I notice some things that might need to be changed. What is the proper procedure? Leaving a review on the (merged) PR? Or make a new PR with the changes that were noticed?
slender iron
@lone axle you can make a comment but a follow up PR is easier if you can do it
idle owl
Or file an issue if you're not in a position to create a new PR.
lone axle
Got it. Thank you both.
idle owl
And issue might be better because it's separate and remains visible to everyone who views the repo vs. only being on the merged PR.
Better than commenting on the PR I mean. A new PR is even better.
lone axle
Okay, yep I made a review on the PR first instinctually but then thought about a second and realized it's pretty much hidden.
lapis hemlock
@lapis hemlock : I'm experimenting with ulab since yesterday... it's amazing. I have a few questions (maybe someone else can answer too)
I can't findatan2()function so I useatan = ulab.array([math.atan2(y, row[i]) for i in range(width)])with row being a ulab.array and y an int. is there a faster way to do it ?
@cursive condoratan2is indeed missing. I have opened an issue here: https://github.com/v923z/micropython-ulab/issues/60. The implementation should not be too difficult, though, one might have to think about broadcasting (since the function involves two operands). Give me a couple of days.
cursive condor
thanks . I'll file a couple of issues likewise in the next days, and i'll keep an eye on this one too ๐
lapis hemlock
I don't think there's atan2() and I don't know about conversion to integer.
@onyx hinge @cursive condor You basically want to be able to initialise an integer array from a floating point one, right, like so ```python
a = array([1, 2, 3])
b = array(a, dtype=int8)
with floats in a, basically yes
lapis hemlock
@cursive condor a is automatically a float array.
cursive condor
yeah, ok
lapis hemlock
@onyx hinge It is a bit embarrassing: I have tried to add test modules for a couple of functions, but they all fail. I don't see an obvious error, though. Do you think you could cast a glance? It must be something really trivial, like wrong indentation or similar. https://github.com/v923z/micropython-ulab/tree/cholesky-test/tests
cursive condor
and I noticedabs() was missing, and round() too , but floor() or ceil() will do the job...
lapis hemlock
@cursive condor You can call abs directly (it is a unary operator) as a = array([1, 2, -3]); abs(a). We can add round.
lone axle
Is there a correct person or team to request review from for the circuitpython-org repo?
idle owl
@lone axle You can probably request from CircuitPythonLibrarians there as well.
cursive condor
nice for abs()
lapis hemlock
and I noticed
abs()was missing, andround()too , butfloor()orceil()will do the job...
@cursive condor Do you mean https://docs.scipy.org/doc/numpy/reference/generated/numpy.ndarray.round.html?
nice for abs()
@cursive condor I forgot to mention thatabsshould be fast (it is aware of the array type), it is implemented inulabitself.
lapis hemlock
@cursive condor In numpy, round is actually not a function on its own, but a method of the ndarray, and I think, that's why I missed it.
cursive condor
yes, this round()... I think it can be usefull
and you did an amazing job already
lapis hemlock
and you did an amazing job already
@cursive condor Thanks! One tries to please๐
@onyx hinge @cursive condor Could you comment on how this should be implemented? I have never used the function, but it is not simply the round function from math.h: it truncates on decimal places. My hunch is that we would have to multiply by some power of 10, call round from C, and then divice by the same power of 10. I wonder, whether, by doing so, we introduce truncation errors...
cursive condor
in cpython it's here : https://github.com/python/cpython/blob/master/Python/pymath.c#L72
or not...
lapis hemlock
I think this is a different function. That is part of python itself, and has nothing to do with numpy.
If you want to have round, then it must be a method of the ndarray, as in https://docs.scipy.org/doc/numpy/reference/generated/numpy.ndarray.round.html .
lapis hemlock
@onyx hinge Jeff, here is an example, where the speedup with the C code is very spectacular: https://micropython-ulab.readthedocs.io/en/latest/ulab.html#benchmarks . Factor of 50. True, the python code could probably be optimised...
like here : https://github.com/python/cpython/blob/master/Lib/_pydecimal.py#L1830
@cursive condor Do you want to try it yourself?
cursive condor
oh no, not yet. I'm not fluent in C.
I try to read code, and understand it, for now
lapis hemlock
Well, in that case, you might have to wait a couple of days.๐
@cursive condor As far as arctan2 is concerned, we might have to restrict it to ndarrays only. numpy allows general iterables and scalars, but treating everything might be too much hassle.
lapis hemlock
@cursive condor Instead of atan = ulab.array([math.atan2(y, row[i]) for i in range(width)]), you could do vector.atan(y/row), and shift it, if necessary: you have a single y value, and you know the sign of that. That should be much faster than iterating in python.
lone axle
@idle owl thank you. Took me a minute to get back to but I've done that now, found some unrelated fires at work that needed putting out shortly after I said that before.
onyx hinge
@lapis hemlock huh that is weird, I don't know what would change the spacing like that.
@lapis hemlock atan2() gives answers in 4 quadrants, so while you can sometimes substitute atan() of a division for it, you can't always
for round(), I guess I'd start with the case which does NOT specify precision, and converts to an integer while detecting overflow .. (which is not quite the same as round(val, 0), which rounds to 0 digits but returns a float. my oh my.)
lapis hemlock
@lapis hemlock atan2() gives answers in 4 quadrants, so while you can sometimes substitute atan() of a division for it, you can't always
@onyx hinge That's right, but the specific example has a constantyin it, so it is known at the outset, in which half plane you are. This is definitely not the general case, I get that.
@lapis hemlock huh that is weird, I don't know what would change the spacing like that.
@onyx hinge We can remove those tests, if you don't know what could be wrong.
onyx hinge
when you ./build.sh it should leave .out files, and renaming them to the exp file should ... work
lapis hemlock
for round(), I guess I'd start with the case which does NOT specify precision, and converts to an integer while detecting overflow .. (which is not quite the same as
round(val, 0), which rounds to 0 digits but returns a float. my oh my.)
@onyx hinge As far as I see,numpyalways returns a float, even if the result is an integer.
when you ./build.sh it should leave .out files, and renaming them to the exp file should ... work
@onyx hinge Oh, so that was the trick. I will try that.
@onyx hinge But
./build.shcallsmicropythonitself, so one can't compare the results to those innumpy. Is that correct? I actually ran the scripts innumpy, and copied the results to the.expfiles.
onyx hinge
@lapis hemlock if there is no .exp file, then the result is compared to the result of running desktop python3
if there is a .exp file then the micropython output is compared to the .exp file
slender iron
@raven canopy or anyone else actively using FrequencyIn? I'm going to wean it off the ticks_ms value and wondering what the best approach will be.
lapis hemlock
@onyx hinge The trick with `./build.sh' eliminated the problem. Thanks a lot! I will just merge the branch. I haven't changed anything beyond test script, so no review is necessary, I believe.
onyx hinge
I think that the exact text of running with numpy and running with ulab are not likely to be the same
unless a lot of attention is paid to the problem anyway
lapis hemlock
That's true, especially with floating point numbers.
ionic elk
@onyx hinge I just merged master and now I can't compile anything D:
make: *** No rule to make target \"extmod/ulab/code/create.c\", needed by "build-meowbit_v121/genhdr/qstr.i.last". Stop.
onyx hinge
@ionic elk did you do the git submodules dance?
ionic elk
oh hmm k
onyx hinge
ulab is in a new-ish submodule and even more recently we made stm32 use it
ionic elk
ok it is fix
Sorry, didn't pick up that it was a submodule issue from that error message.
raven canopy
@slender iron hmm... let me look at your base change again for getting away from SysTick. and, refresh myself on my own code. ๐
slender iron
it looks like there is a reference timer but ticks is used for the actual period and computation
raven canopy
yeah. it uses ticks_ms for the time domain. iirc
orchid basinBOT
raven canopy
@slender iron so, would it switch to using port_get_raw_ticks, and then adjust the math from 1000 to 1024?
onyx hinge
@ionic elk great
cursive condor
@lapis hemlock thanks for the hint. it worked well with ย atan(row / y)ย so 1.5 secs become 0.15 secs
manic glacierBOT
lapis hemlock
@cursive condor Great!
cursive condor
atan(y/row) made an error, I can't divide something by an array
lapis hemlock
atan(y/row)made an error, I can't divide something by an array
@cursive condor I know about that issue, and it is related to the order of the operands. @onyx hinge mentioned that in his tutorial. I will fix that, in fact, there is already an open issue for that on github.
onyx hinge
row*(1/y) in the interim?
lapis hemlock
How does that work? I don't see that at the moment. He needs y/row, doesn't he?
raven canopy
@slender iron ahh. with the loss of current_us = current_ms / 1000 (basically), that's going to force more recalculation beyond 1000->1024. not sure how much that's going to affect accuracy. i'd have to sit down with the math and some raw data to re-orient myself to it.
lapis hemlock
But atan(row/y) is just a rotation of the coordinate system, so that should be OK. At least, for this special case.
cursive condor
@onyx hinge I think tan(row*(1/y)) should have the same result
and tan(row/y) was good only for 2 quarters of the image
lapis hemlock
row/y works, but y/row doesn't, because y is not an ndarray.
and
tan(row/y)was good only for 2 quarters of the image
@cursive condor Sinceyis fixed, it doesn't matter.
cursive condor
the resulting image was not as expected ... soooo ...
lapis hemlock
But if y > 0, then you know that you are on the upper half plane, so atan2 doesn't bring in any extra information.
You need atan2 only because there is an ambiguity in the sign, which you don't have.
cursive condor
well ... every intensive bits are now handled by ulab ... with the same parameters, worst version took 51 sec, best is now 6.6 sec
I don't know why it works one way and not the other equivalent way ...
y is a float, and it can be negative. it is fixed for one row, but change ...
onyx hinge
double atan2(double y, double x);
These functions calculate the principal value of the arc tangent of y/x
so maybe it's tan(row/y) to get the expected result ?
lapis hemlock
y is a float, and it can be negative. it is fixed for one row, but change ...
@cursive condor But in your example,ywas fix: you hadatan2(y, row), whererowwas anndarray. I understand that you might still have to iterate overy, but you can still remove the iteration overrow.
onyx hinge
oh what I said is nonsense, carry on
lapis hemlock
so maybe it's
tan(row/y)to get the expected result ?
@onyx hinge Thex->yis just a coordinate transformation. I can callyx, if I want to...
onyx hinge
@cursive condor if you are plotting stuff, it would be cool if you could share a photo of something you're plotting just for the sake of excitement ๐
cursive condor
yeah it's what i'm doing
onyx hinge
.. stepping away again ..
lapis hemlock
I am signing off, too.๐
solar whale
just a heads up -- needs more investigation, but I am seeing an issue where I need a hard reset to get an SPI device working after a soft reset. without the hard reset it appears to hang -- using current master commit d988af0 - observed on both Teensy4.0 and feather_m4_express -- I'll try to get a good test case and file an issue - probably not util tomorrow. Note - I am using board.SPI() if that could be relevant...
onyx hinge
neat!
thanks!
I don't know how many hours I spent with Fractint and Xaos way back in the 90s and early 00s
cursive condor
my pleasure. I'm cleaning the code to update the github repo
onyx hinge
I wonder if these are doable on CircuitPython. I did a sort of similar 1D simulation for neopixel, but going to 2D would be really neat https://en.wikipedia.org/wiki/Reactionโdiffusion_system
(well, mine is a 1D wave simulation, but they're all differential equations which can lead to interesting behavior over time)
I just had a weird problem. Multiple CircuitPython devices would show the CIRCUITPY device when plugged in, but serial programs couldn't receive characters from them. Re-plugging the USB hub I use fixed it.
I'm about to show this USB hub the door
ionic elk
@slender iron For the F7/H7 additions I'm guessing I should rename the port directory from stm32f4 to either stm32 or stm - do you have a preference?
slender iron
@ionic elk different from micropython
ionic elk
stm then?
ionic elk
Probably worth doing that as a PR by itself, I think - otherwise git won't highlight specific changes
Micropython is stm32
slender iron
@raven canopy I don't think we need ticks at all. the reference timer can give us a frequency to compare to
fierce girder
fyi - chat about the circuitpython powered nRF52840-based OHS20 wrist badge is over on the OHSummit discord: https://discordapp.com/channels/687040227272818724/687378975030247442
stuck elbow
I don't think this link works :(
prime flower
@fierce girder this link does not work for me
stuck elbow
"You find yourself in a strange place. You don't have access to any text channels, or there are none in this server."
in the distance: sirens
fierce girder
@prime flower darn... not sure how to link to a discord channel
stuck elbow
I think you need an invite link
manic glacierBOT
raven canopy
@slender iron with the reference timer triggering the handler based on the counter overflow, will it be consistent/accurate enough? the clock is set to 98.304MHz, which overflows 1500.nn times per second with a 16bit counter. if my equation is correct. (i haven't had to stretch these muscles in quite some time)
what happens when higher priority interrupts skew the handler?
slender iron
ticks_ms has the same problem
raven canopy
really? i thought SysTick was at the highest priority... hmm
slender iron
we could use the capture functionality of the counting TC to grab it faster
it is but interrupts still have lag
what frequencies are important?
raven canopy
capture
arg. that mode...brings back horrible memories. ๐ its blocking.
slender iron
what do you mean blocking?
raven canopy
using capture forces the interrupt handler to fire at each event, which is what was locking up pulsein above certain frequencies. blocking may have been the wrong term to use.
slender iron
I was thinking the reference tc triggers the capture
on overflow
the tc itself can also give you frequency
raven canopy
the reference triggers reading the actual counter. the others are just in count mode.
slender iron
using ppw capture
fierce girder
re: OHS nrf82540 badge, https://discord.gg/38C57Uf
Michael Welling started a "#badge-hacking" channelโ
โ
I guess I can't link directly to a channel
slender iron
ah, I'm thinking the TC could take two input events, which it can't
we could change the TC event from COUNT to PPW and it could measure anything slower than the base clock
raven canopy
trying to find all my old notes
ionic elk
@raven canopy you're the go to for the docs system, right?
raven canopy
hehe. i was chasing that notification!
"go to"...i wouldn't put myself at that level, but i know a thing or two. ๐
ionic elk
Lol no pressure. Just had a wee CI failure that I don't understand, couldn't remember who to loop in.
raven canopy
link me in, and i'll gladly take a look.
ionic elk
added you to review but it's here https://github.com/adafruit/circuitpython/pull/2699
thanks much
raven canopy
hmm. looks like it's pulling in all of the HAL files to parse.
slender iron
I think there is a blacklist for files in docs/ somewhere
raven canopy
yep. need to change these exclusions: https://github.com/adafruit/circuitpython/blob/master/conf.py#L143
ionic elk
seems like there's some junk in that file. want me to clean it up while I'm here?
slender iron
sure
@raven canopy did you try PPW at all?
seems like that could greatly simplify it
raven canopy
i remember trying it, but there was a reason it didn't work. that's what i'm trying to find notes on.
raven canopy
@ionic elk ๐ that's one down. incremental progress is the best, right? ๐
@slender iron i think this is the best i'll get, note-wise. https://github.com/adafruit/circuitpython/pull/1144#issuecomment-416008353
it was in response to that (PPW interrupt loop), that we switched to the reference timer setup. i took what seems to be horrible notes in the weekly meetings.
ionic elk
are you referring to me chipping away at CI here ๐คจ
ionic elk
At least it isn't failing just for giggles like on my PulseIO
That one's giving me big red github banners and stuff, "cannot request CI action at this time" etc
picking random new tests to fail that it passed last time โ
raven canopy
i think Dan has fought through those...
ionic elk
Well, it's a finicky beastie. I might just return to it tomorrow and find it working fine again.
Now I'm moving on to the fact that I can't install any STM32 tools on Catalina ๐
slender iron
@raven canopy I think if we use PPW we don't need an interrupt at all, we just read the register as needed
it won't average but the python code can do it
raven canopy
i vaguely remember the COPENx register not working with a pin (==1), which is what forced the use of EVSYS. does that require an interrupt (i've been out of that datasheet for too long)? its of course, not in the errata that it doesn't work. can definitely be revisited though. being combative has not been my purpose here. ๐
slender iron
hrm, will have to look into it
no worries, I appreciate anything you can remember
it's been two years so it may be in the errata now
raven canopy
its not. just pulled a fresh one. ๐
raven canopy
totally. it could very well have been my blundering.
onyx hinge
mmmmm chili
we just had fresh bread I made while frustrated about hard fault handler ๐
ivory yew
Hmmm... Would it be possible at some point to use AudioIO with a DAC over SPI/I2C/I2S?
onyx hinge
@ivory yew we have I2SOut on some MCUs, I2C is probably too slow for audio (400kHz vs 1.5Mbit/s for uncompressed CD quality audio). SPI would be interesting, but I don't know of any HW for it. https://circuitpython.readthedocs.io/en/5.0.x/shared-bindings/audiobusio/I2SOut.html
ivory yew
Ah cool I didn't know about audiobusio
Sol's dac is on spi, I wonder if I could port i2sout to spiout
It'd have to specific to a particular dac though. Hmm. No worries, just doing some research.
onyx hinge
yeah. First DAC I looked at, MCP492x, would probably work without too much trouble. Trigger a 32 bit transfer from a 44.1kHz timer. Tie /LDAC low. Sample updates will be offset by 22.7us. I think you have an AD5686xRUZ which is less likely to match up with the SPI hardware because you need to write 24 bits per channel you want to update
ivory yew
yeah
simple pulsar
FYI, the axoloti board which is designed for audio work uses an ADAU1961WBCPZ driven by an STM32F4something according to its schematic. It runs at 48k 24bit and notes say it's capable of 96k.
mellow turret
Just received a STM32F405 Feather Express today. Thinking the USB A MALE / micro B MALE cable on my bench would fit. No go, no joy. So what is the USB port on this device, what am I looking for? Micro-B FEMALE? Is it a micro variation of USB C?
timber mango
@mellow turret It is USB-C
mellow turret
@timber mango Thanx, I'm on it, ordered a right A to C adapter. have plenty of USB A floating' around the lab. When the adapter arrives, it should be 'All the way with LBJ!' ๐
manic glacierBOT
solar whale
@tulip sleet I'm running into an odd issue -- where should i file this issue? When using my rfm69 on a feather_m4_express or teensy4.0 I find that the accessing SPI hangs after a soft reboot if I create the SPI interface with board.SPI() rather than busio.SPI(board.SCK.......). After a power cycle, either method works. Should I file an issue with CP , Bus_Devoce or RFM69? I am using the current master commit d988af0 ...
manic glacierBOT
tulip sleet
@solar whale are you using some RFM library that needs SPI? Do you pass it the SPI object or does it create it?
tulip sleet
that sounds like a circuitpython issue: we are not clearing the lock on soft-reload after an exception or ctrl-c
solar whale
that seems to be the issue..
tulip sleet
i can transfer the issue to circuitpython
solar whale
OK thanks
manic glacierBOT
manic glacierBOT
tulip sleet
@ionic elk the checkout failed again. maybe try merging from upstream and pushing again
ionic elk
@tulip sleet it's already up to date. I'll try force pushing
nvm it won't even let me do that
I pushed a minor comment change, we'll see what it does
tulip sleet
@ionic elk it's getting past the checkout problem, yay
ionic elk
@tulip sleet I'm also trying to figure out the best way to handle our submodule for the STM32 material https://github.com/adafruit/stm32f4 - it needs to be renamed and updated with the stm32F7 and H7 libraries.
tulip sleet
i would just call it stm32 if they are all going to be merged
the only issue with stm32 was the ports/stm32 name in micropython, but I got rid of that directory, right?
ionic elk
sure, the name doesn't matter. I was going to go with STM32_HAL_lib
tulip sleet
is that what it's called in the STM world? (
ionic elk
yeah @slender iron still wanted our directory name to be different so it's moving to stm, I've got a PR up for that now
@tulip sleet the actual HAL directories are all named things like STM32F4xx_HAL_Driver so I figured it'd be more descriptive than just calling it "stm32" and it's distinct from micropythons stm32lib
or stm32haldriver
In any case, though, if I change the submodule repo name that probably needs to be a coordinated process, right? Since it'll break things for a short period. We'll want to approve the PR and change the repo name at basically the same time.
@tulip sleet CI cancelled for no reason again. I know for a fact the F4 discovery passed before
tidal kiln
@slender iron any ideas on this?
https://forums.adafruit.com/viewtopic.php?f=19&t=163189
slender iron
looks
@tidal kiln very weird! can you have them copy the contents of their CIRCUITPY over to you?
seems like a software issue
tidal kiln
like a zip of the entire CIRCUITPY folder?
slender iron
ya, of the drive
tidal kiln
ok, requested. i'll ping you again when it's up.
slender iron
thanks! it kinda looks like the color processing used for eink
manic glacierBOT
ionic elk
@slender iron quick ping about my last couple PRs - PulseIn is having some trouble with CI but has passed every test at some point, and the stm namechange is ready to go but will probably incur merge conflicts with PulseIn regardless of whether it is merged before or after. Once the namechange PR is through I think we'll need to do a coordinated bit of work where we change the stm32f4/ submodule name and also submit quick PR to acknowledge the change.
That should allow future F7 and H7 submissions to go through more smoothly.
Does that sound ok by you?
lapis hemlock
@onyx hinge @cursive condor As requested, I have added the around function, and you can also convert dtypes between arrays. https://github.com/v923z/micropython-ulab/pull/63 Here is the documentation https://github.com/v923z/micropython-ulab/blob/round/docs/manual/source/ulab.rst#initialising-by-passing-arrays, https://github.com/v923z/micropython-ulab/blob/round/docs/manual/source/ulab.rst#around Jeff, if you think that it is OK, you can merge the branch.
slender iron
@ionic elk will look now. just went for run to get out of the house.
idle owl
@tulip sleet Can you take a look at this one? There's a ton of chatter but none of it is Adafruit folks. https://forums.adafruit.com/viewtopic.php?f=60&t=162938
timber mango
I wish I had a WiFi FeatherWing. I would add it to my FrankenCircuit that has Bluetooth (Bluefruit LE SPI Friend breakout board) and RFM69 radio (also breakout board). Bluetooth and RFM69 are both working. ๐ ๐ ```python
FrankenCircuit v0.0
Initializing Bluetooth
BLESPIFRIEND
nRF51822 QFACA10
A061ADEB7464CE53
0.8.1
0.8.1
Apr 10 2019
S110 8.0.0, 0.2
Initializing the RFM69 radio
RFM69 Radio Data
Temperature: 87.8ยฐF (31.0ยฐC)
Frequency: 915.0 MHz
Bit rate: 250.0 kbit/s
Frequency deviation: 250.0 kHz```
ionic elk
@slender iron ok, merges are in. Could you approve the additions to https://github.com/adafruit/stm32f4 so I can test them out? Then we can do the hotswap of the submodule
Oh wait, I guess I don't even need approval on this repo ๐
half sedge
I don't really like the suggestion of playing dropping egg with a CLUE. I believe a lot of kids or adults might destroy their CLUE doing that...
Even in the YouTube video one CLUE stopped to boot.
It's great, it's fun, but it is potentially destructive.
ionic elk
@slender iron you free to do this submodule swap? I don't have permission to rename adafruit/stm32f4
needs to be renamed to stm32drivers and then I'll quickly get a PR in to change it in Circuitpython
slender iron
are those the same drivers as : https://github.com/STMicroelectronics/STM32CubeF4
?
since we're switching maybe we should move to a world where we're based on st's repos or forks of them
there are repos for H7 and F7 too
ionic elk
Oh, did they even have that last time we checked?
ionic elk
it isn't very compact... lot of cruft that isn't the drivers
slender iron
we shouldn't do much in there though
I'd prefer those repos since they make it clear how to update them and where it came from
ionic elk
Where do you want it? Should I keep it in ports/stm? or add them all to Lib?
slender iron
maybe a subdir in ports/stm?
ionic elk
I wish they had a combo one... it's going to copy all the projects and CMSIS stuff for every chip family.
slender iron
looks for a way to do a subfolder
I don't see any way to do it. You are right that it is heavier weight but I think the history and ease of updating is worth it. It's one of my major gripes with HALs. Maybe @tulip sleet and @onyx hinge have other ideas.
๎ฑ git clone -o st git@github.com:STMicroelectronics/STM32CubeF7.git Thu Mar 12 13:14:51 2020
Cloning into 'STM32CubeF7'...
remote: Enumerating objects: 5109, done.
remote: Counting objects: 100% (5109/5109), done.
remote: Compressing objects: 100% (3128/3128), done.
remote: Total 30608 (delta 1911), reused 4179 (delta 1878), pack-reused 25499
Receiving objects: 100% (30608/30608), 454.34 MiB | 34.93 MiB/s, done.
Resolving deltas: 100% (15161/15161), done.
Updating files: 100% (31156/31156), done.
/V/GitRepos ๎ฐ
๎ฑ git clone -o adafruit git@github.com:adafruit/stm32f4.git 27.3s ๎ณ Thu Mar 12 13:15:46 2020
Cloning into 'stm32f4'...
remote: Enumerating objects: 3121, done.
remote: Counting objects: 100% (3121/3121), done.
remote: Compressing objects: 100% (942/942), done.
remote: Total 3121 (delta 2278), reused 2976 (delta 2138), pack-reused 0
Receiving objects: 100% (3121/3121), 30.44 MiB | 35.54 MiB/s, done.
Resolving deltas: 100% (2278/2278), done.
you/we could clone the submodules shallow and only do deep if we are going to alter them
slender iron
ah true
ionic elk
I must have to practice my git-funise - what is a shallow clone?
slender iron
didn't seem to help
๎ฑ git clone --depth 1 --single-branch -o st git@github.com:STMicroelectronics/STM32CubeF7.git Thu Mar 12 13:19:19 2020
Cloning into 'STM32CubeF7'...
remote: Enumerating objects: 27204, done.
remote: Counting objects: 100% (27204/27204), done.
remote: Compressing objects: 100% (12958/12958), done.
remote: Total 27204 (delta 14330), reused 25900 (delta 13925), pack-reused 0
Receiving objects: 100% (27204/27204), 449.36 MiB | 37.10 MiB/s, done.
Resolving deltas: 100% (14330/14330), done.
Updating files: 100% (31156/31156), done.
shallow is only cloning the current version of the repo instead of the whole history
tulip sleet
it's weird, it gets to 99%, fetching only ~~100MB and then it has to fetch a dozen or so huge objects
ionic elk
Doesn't seem to save much
tulip sleet
i wonder if they need to git gc
slender iron
@tulip sleet do you have a preference on which to use in circuitpython? the heavy st repos or our own
tulip sleet
@slender iron @ionic elk is the tinyusb ST repo good enough or is it incomplete? Given that we are using tinyusb, it has to include what we need
slender iron
not if we don't include the tinyusb submodules
ionic elk
I mean the contents of it are fine I just figured it wasn't a good idea to include another submodules submodule
tulip sleet
but i mean we can include that same submodule as a first-level submodule
doesn't tinyusb need its ST submodule to be able to compile??
ionic elk
well the Adafruit repo is just a copy of Tinyusb's repo right now
tulip sleet
i guess tinyusb is just looking on the -I path, so it needs to find them somewhere.
ionic elk
You were mentioning that it was an advantage to have our own clones of some submodules to reduce the chance of breaking changes
slender iron
right, generally we have a parallel copy of the submodule to use
tulip sleet
so tinyusb has its own trimmed ST HAL repo, I thought
isn't that what we are using now? I'm confused
ionic elk
we could nab Zephyr's nice tidy repo https://github.com/zephyrproject-rtos/hal_st/tree/v1.14-branch
tulip sleet
is it the same API?
ionic elk
yes
tulip sleet
the only problem is that the STMicroElectronics one is a pig, probably because of unnecessary history (like someone accidentally or deliberately checked in some binaries or zip files)
slender iron
yup
ionic elk
(a pig?)
slender iron
we could fork and delete stuff
tulip sleet
so we could use the tinyusb one or clone our own and throw out some old history and see how big it is
gmta
ionic elk
tulip sleet
a pig = huge
tulip sleet
@Hieropehect can you just expand on that?
slender iron
@ionic elk my issue with that repo is that it doesn't share history with the st ones
ionic elk
it's based off TinyUSB
slender iron
then it should be a fork
ionic elk
Is that the best option out of the three? 1) fork TinyUSB's officially 2) fork each of the STM32 ones and cut cruft 3) use raw STM32 ones and ignore ginormousness
slender iron
my vote would to be to use thach's repo directly
slender iron
that way we're in sync with tinyusb and I like the commit messages since they include version numbers
yup and we can always fork if we really need to
ionic elk
Ok. Cool. I will do this. Now I just gotta go back and remove all those ST submodules I just added ๐ฉ
indigo wedge
My i.MXRT port of CPY got me a Runner Up place in the Hackster NXP Challenge ๐ https://www.hackster.io/contests/nxpcrossover
slender iron
sorry for the indecision @ionic elk
ionic elk
is k
tidal kiln
@slender iron zip added:
https://forums.adafruit.com/viewtopic.php?f=19&t=163189#p803199
i still can't recreate it, even with their bmp's, lib's, and code.
slender iron
hrm interesting!
manic glacierBOT
slender iron
@tidal kiln it works for me too. it's probably cheaper/easier to just replace it
tidal kiln
ok. weird that the hardware checked out OK with arduino sketch.
slender iron
ya, two other ideas would be to get them to erase the fs
tidal kiln
could also give them a CP prog that uses shapes lib to just draw colors
slender iron
but any corruption would have been copied into the zip
could ask for a picture of the terminal
see if it is right or not
tidal kiln
ok, i'll try a few more things. will let you know if anything comes up.
slender iron
it's weird that it's only on bitmaps
you could also ask for exact plug/unplug and button press instructions
tidal kiln
yep, you'd think if it were a HW issue it wouldn't matter
BMPs vs. colored rectangles
but i was focusing on display HW - maybe it is a fs thing
slender iron
I would expect any fs issues to be copied over
timber mango
Trying to use older hardware really does not get one anything except frustrated. I can not get the information I need to set my Bluefruit LE SPI Friend to be a URI Beacon. Nothing I have tried works.
wraith tiger
timber mango
Can you rotate that image? It is hard to to tell what everything is. It will be a long time until I can afford to order more toys from Adafruit. ๐ฆ
lone axle
@timber mango did you try AT+BLEURIBEACON or AT+EDDYSTONEURL AT commands?
The simpletest in the bluefruitspi library appears to be writing AT commands to the BLE device. I don't see any example specifically for beacon. But looking at the list of AT commands these ones seem like the ones to use for making a beacon.
timber mango
@lone axle I tried the first but not the second. I do not have the fancy new nRF52840 hardware. I have the Bluefruit LE SPI Friend breakout.
lone axle
I found them here: https://learn.adafruit.com/introducing-the-adafruit-bluefruit-spi-breakout?view=all#version-0-dot-6-7-18-4
lone sandalBOT
timber mango
I have been through that tutorial.
timber mango
@lone axle I think I am just going to have to let this go because it has become way to frustrating for me to deal with. ๐ฆ ๐ฆ
orchid basinBOT
static jay
Greetings all, is this a good place to ask about adafruit_touchscreen running on a Pyportal? (running CircuitPython 5.0.0-rc.1)
solar basin
So, if I'm reading this right, _bleio scans the entire data section for each prefix, yeah?
slender iron
@solar basin advertising data? yup
each set of advertising data is length encoded so it looks at the prefix of each set
timber mango
Good night. I will be back in a few hours, or maybe sooner if I can not sleep. ๐
simple pulsar
@tidal kiln If you've run out of things to try, it might be interesting (if it's possible/easy) to drop the display's SPI bus speed substantially to see if that makes any difference for https://forums.adafruit.com/viewtopic.php?f=19&t=163189
simple pulsar
The soldered side of my connector looks a bit different to the misbehaving CLUE. I have black rather than green boards and with mine I can see the dark matte board between the soldered pins on left side. On the misbehaving one there's not much green on show but perhaps the light is just catching the lacquer?
umbral dagger
I just upgraded a CLUE to 5.0.0 and libs from today's bundle. Is there a known issue?
File "code.py", line 7, in <module>
File "adafruit_clue.py", line 917, in <module>
File "adafruit_clue.py", line 207, in __init__
File "adafruit_lsm6ds.py", line 247, in __init__
File "adafruit_register/i2c_struct.py", line 85, in __get__
File "adafruit_register/i2c_struct.py", line 85, in __get__
File "adafruit_bus_device/i2c_device.py", line 139, in write_then_readinto
OSError: [Errno 19] Unsupported operation```code.py is the slideshow code straight from the guide.
slender iron
not that I know of
@tidal kiln what markings does your display ribbon cable have on it?
mine doesn't match the user who is having trouble
umbral dagger
OK, when it first comes back from having the CP UF2 copied onto it:```Adafruit CircuitPython 5.0.0 on 2020-03-02; Adafruit CLUE nRF52840 Express with
nRF52840
Traceback (most recent call last):
File "code.py", line 7, in <module>
File "adafruit_clue.py", line 917, in <module>
File "adafruit_clue.py", line 207, in init
File "adafruit_lsm6ds.py", line 244, in init
File "adafruit_bus_device/i2c_device.py", line 68, in init
File "adafruit_bus_device/i2c_device.py", line 178, in __probe_for_device
File "adafruit_bus_device/i2c_device.py", line 176, in __probe_for_device
ValueError: No I2C device at address: 6a ```
tidal kiln
NFP133H-05A @slender iron
umbral dagger
This is after erasing and recopying the latest libraries again, then recopying the latest CP for the board.
slender iron
same for me
umbral dagger
If I CTRL-D then I get the unsupported operation error.
It's a green CLUE if that makes a difference
raven canopy
PSA: I will be running another library patch in about an hour. This one is a "manual" patch, so notifications will be sent with the new Pull Requests. Sorry for the inbox pain.
simple pulsar
FYI, my two (black-board) CLUEs both have NFP133H-05A in white writing on the display ribbon connector.
slender iron
@umbral dagger what happens if you do an i2c scan?
(can you also tell us what the display ribbon markings are?)
if they match what carter posted
solar whale
@umbral dagger -- tried 5.0.0 and it still is working OK for me.
slender iron
@tidal kiln the user's display marking matches the internally documented one
umbral dagger
@slender iron display cable markings the same as @opaque musk
slender iron
hrm, ok
umbral dagger
It seems better on a black CLUE
slender iron
i2c is better?
solar whale
hmm -- I'll try my green one ... just a sec
umbral dagger
Works fine on the black one (once I figured out where the images needed to be ๐ )
Hmmm.. soldering looks a little light handed on the green board around the LSM6DS
solar whale
@umbral dagger works on my green board as well
tidal kiln
@slender iron fyi - i passed that forum thread up the line
@simple pulsar thanks for the suggestion. i'm also wondering if it might be a marginal connection. maybe slower spi could help? would also be interesting to try a BMP display via arduino. but at this point sort of running out of amount of iterations the poster might be willing to cope with.
umbral dagger
@solar whale I'll try retouching that pin with an smt tip. There one that is noticeably missing the usuall bit of solder that's visible.
solar whale
good luck! those are small pins
simple pulsar
@tidal kiln I have seen displayport (or perhaps it was DVI) fail in strange unpredictable ways with overly long cable / splitter and that was a bit data dependent as to when your monitors would misbehave. In that case it was going beyond the spec but it reminded me of that. This is problaby happening more with (variable height) standing desks being in vogue.
raven canopy
patches starting.... ๐ฌ
slender iron
@tough flax I rebased my lower_power branch
onyx hinge
I noticed clue is back in stock at the moment, I was getting twitchy with just 1 on hand
solar whale
@raven canopy just a thought -- couldn't things like the Code of Conduct be a link so changes don't have to be updated in every repo...
or just change it to "Be kind" ๐
raven canopy
@solar whale i don't disagree with the thought, but that question is "too big" for me. :D
I could see a counter argument that it should be available when a repo is locally cloned (or created, via cookiecutter).
prospective in the weeds for monday?
solar whale
No problem -- new mail rules are working ๐
@idle owl Not that I want it changed, but shouldn't this have refered to community_moderators not helpers https://github.com/adafruit/Adafruit_CircuitPython_FeatherWing/pull/59/files#diff-a1ee87dafebc22cbd96979f1b2b7e837R77
line 77
solar whale
sorry -- I won't tell...
idle owl
@solar whale Nah, good catch. But it turns out a patch will work after we get them all standardised. They were too all over the board before for a patch to work.
That is much simpler than this.
tulip sleet
@slender iron I'm working on the Adafruit Services library. All the services provide measurement_period and service_version characteristics. Can the service class definitions share the same declarative object that defines those characteristics, or do I need to instantiate new Characteristic objects for each service class? I'm not remembering whether the declarative objects are tied to a service or not.
slender iron
@solar whale GitHub does some repo health checks and linking that'd we miss if we used a link I believe
@tulip sleet I'd subclass Characteristic to save uuid and such and then just instantiate it for every class
tulip sleet
but i do need a separate instance for each class? e.g. (code is not correct for the args)
_measurement_period = Int32Characteristic(uuid=VendorUUID(adafruit_uuid_str(0x0001)),
properties=(Characteristic.READ |
Characteristic.WRITE_NO_RESPONSE),
read_perm=Attribute.OPEN,
write_perm=Attribute.OPEN)
_service_version = UInt32Characteristic(uuid=VendorUUID("ADAF0002-C332-42A8-93BD-25E905756CB8"),
properties=(Characteristic.READ |
Characteristic.WRITE_NO_RESPONSE),
read_perm=Attribute.OPEN,
write_perm=Attribute.OPEN)
class TemperatureService(Service):
"""Temperature sensor
uuid = StandardUUID(adafruit_uuid_str(0x100))
temperature = FloatCharacteristic(uuid=VendorUUID(adafruit_uuid_str(0x100)),
properties=(Characteristic.READ |
Characteristic.Notify),
read_perm=Attribute.OPEN,
write_perm=Attribute.NO_ACCESS)
measurement_period = _measurement_period
service_version = _service_version
so this is wrong?
slender iron
yes you do
tulip sleet
ok, thanks!, will structure as appropriate
slender iron
you can have all those attributes wrapped in a subclass though
tulip sleet
yes, looking to make an AdafruitService superclass
slender iron
that would work too I think
tulip sleet
yeah, not sure about instantiation of the class vars
ivory yew
I'm getting an attribute error when trying to set the PWM frequency on a Trinket M0:
>>> pwm.frequency = 440
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError:
that doesn't seem normal
also it's weird that it has no details
onyx hinge
@ivory yew did you construct it with variable_frequency=True ?
onyx hinge
the "no details" thing is probaly still worth bug filing!
ivory yew
I'll see if I can send a PR once I get done with my experiments
onyx hinge
that would be awesome
ivory yew
ofc
manic glacierBOT
simple pulsar
I've just realised that a CircuitPython debugger would be very useful for some bugs I wrote over last few days, hence that comment on issue 298
tough flax
Thanks @slender iron
timber mango
@simple pulsar That thread looks very familiar. Is that the one I started?
timber mango
I have a script that is heading towards 1000 lines really fast. Sometimes, I wish for a debugger, but so far the print method is still working for me.
main meteor
A lot of people disapprove of print debugging, but it's quick and surprisingly versatile and effective
timber mango
If ARM MCUs keep going the way they are now, they WILL supplant Intel and AMD at some point. I remember when we drooled over 600 MHz Pentium systems...
@main meteor It is still working well for me.
main meteor
It's getting pretty close, as efficient as ARM is: the only way forward for X86 is more cores at this point. I remember the days when a 66MHz pentium was considered fast and a 22MHz Sparc would outrun it easily.
timber mango
I remember that too! The price of an x86 system, even with an AMD CPU will not evade ARM for too long.
And now, look what NXP is doing... ๐ ๐
main meteor
I remember when a VAX 11/780 was the size of a refrigerator and gave you all of ... 1 MIPS. And we'd have dozens of users on it at once.
timber mango
Oh yes, the age of the dinosaurs!
@main meteor I think users were slower back then. ๐
main meteor
We just had the command line then, and were still impressed when we could ftp a 1MB tar file across the room in a few seconds.
timber mango
You must be as old as I am.
main meteor
Quite possibly.
timber mango
This is weird. I am getting a UnicodeError, but there is no other information about the problem except a line number.
I am only sending printable text strings back and forth over an RFM69 radio link and somehow a string seems to have become badly corrupted. Strange.
slender iron
@main meteor I'm a print-debugger first and foremost. It's the only debugging technique that works across language and platform in the same way. I was shocked to have to debug with an led blink on some micros.
timber mango
@slender iron It would just be really useful to be able to get inside code, step through it, maybe print some variables, and continue on. I am pretty sure nothing like that will exist though, so I march on with my trusty print().
@slender iron How do I work on more Circuitpython library stuff when I do not have the sensors, boards, or peripherals necessary to test stuff out? I was just lucky with the HT16K33 library stuff because I happen to have one here.
Print debugging is ALWAYS usable too, no matter what environment you find yourself in!
G'Night. I may be back later.
slender iron
@slender iron It would just be really useful to be able to get inside code, step through it, maybe print some variables, and continue on. I am pretty sure nothing like that will exist though, so I march on with my trusty print().
@timber mango I'm not prioritizing a debugger because it's not blocking anyone from doing anything. I'd rather spend time unlocking new projects like solar powered sensors or python badges with better battery life.
@slender iron How do I work on more Circuitpython library stuff when I do not have the sensors, boards, or peripherals necessary to test stuff out? I was just lucky with the HT16K33 library stuff because I happen to have one here.
@timber mango What do you have already? You could also start poking at the core of CircuitPython.
timber mango
@slender iron I do understand. I have become quite skilled in placing my print() statements over the years and knowing where problems are. I think knowing where a problem is is more than half the battle to hunting down a bug and squashing it.
@slender iron I have a couple displays (ht16k33 and charliewing), sht31d and htu21 temperature/humidity sensors, tcs34725 RGB, Fxas/Fxos NXP IMU, lsm303dlhc, lis3dh, vl53l0x, and apds9960. Of course, I have five various format rfm69 900 MHz radio modules to. I really like that little lis3dh - I got one because that is what is on the Circuit Playground Express (I have five of these now).
Poking at the core of Circuitpython sounds like poking a mama bear with cubs.
slender iron
The core isn't that bad. A debugger is helpful when doing it though. Do you have any of the esp32spi airlifts? You could also try to make a library for bluefruit spi that implements _bleio
timber mango
I do not have any of the airlifts or new bluetooth stuff. Do you mean write a _bleio compatible library for my Bluefruit LE SPI Friend?
slender iron
ya, for the ble spi friend
timber mango
That will be an interesting project. I am just learning how to use Bluetooth.
half sedge
Not that I have any BLE SPI Friend, but that sound interesting. I can tell that searching BLE things on Adafruit learn guide or product can be confusing. The confusion increase when you figure out that CircuitPython support the serial protocol and so the App that was done for the old BLE stuff. Of course it is great and directly provide a lof of feature and make lot's of project accessible. It is just that I was missing a starting point that say what are the BLE (or BlueFruit) stuff in Adafruit universe.
half sedge
@gilded cradle Any hit, piece of code, learning guide in preparation to control my Bit:Buggy?
gilded cradle
@half sedge, no learning guide in progress yet (I'm finishing a couple others), but hopefully soon. In the meantime, I started off with this as a base: https://learn.adafruit.com/circuitpython-ble-crickit-rover, used the neopixel and adafruit_motor libraries and found pixels are on board.D0 and left servo is on board.D1, and right servo is on board.D2. I don't want to give away too much, but that should get you started.
manic glacierBOT
granite crow
Hi, I was involved in the translation of cp into Spanish, but forgot the process of getting started with it, is it documented somewhere?
slender iron
I don't think it is actually. The easiest way to get started is to edit the .po file on GitHub. https://github.com/adafruit/circuitpython/blob/master/locale/es.po
granite crow
IIRC there was a step I should do after editing the po file, make translation?
lone axle
@granite crow there is some information and discussion here: https://github.com/adafruit/circuitpython/issues/1098. Not sure if it includes what you are looking for though for certain.
slender iron
@granite crow make translate is only needed when adding new strings in the code
granite crow
Thanks for the link @lone axle i will check it out
Ok @slender iron I guess I have all the information to get started
slender iron
๐
tough flax
@slender iron I can confirm that with the ESP32 connected, 5.0.0 boots and runs fine.. your low power version doesn't start CIRCUITPY
This is on an Feather Bluefruit Sense + Airlift Wing
slender iron
kk, will look this week at it. I think my approach is to change it all and then test and fix it up
digital elbow
anybody familiar with the 480 x 320 TFT LCD display with CircuitPython?
keep getting memory allocation errors
main meteor
Which CPU? That display does eat a lot of RAM.
digital elbow
Metro Express M0
configured for SPI and here is what I get
Traceback (most recent call last):
File "code.py", line 30, in <module>
MemoryError: memory allocation failed, allocating 19200 bytes
here is the line its trying to execute
color_bitmap = displayio.Bitmap(480, 320, 1)
lone axle
I think the M0 devices do not have enough RAM to to hold a bmp that size in memory along with everything else needed to work the display (and do whatever else your code does). On the Circuit Playground with Gizmo I believe I ended up needing to use OnDiskBitmap for any bitmaps that are the size of the screen. And it's even a bit smaller at 240x240.
Depending on what you are trying to do there is a "trick" where you can make a much smaller bitmap object than the screen and then put it inside of a Group that is scaled up enough to make it fit the screen. Then you can still edit the bitmap object which is now scaled to fit the screen
You lose the ability to make finer resolution things within the bitmap, but gain some efficiency. Totally depends on your use case though whether that technique could work for you.
digital elbow
thanks ..... where can I learn more about that
is there a way to place it on the SD card slot on the back?
lone axle
Hmm I am not sure if the technique is fully documented anywhere DavidGlaude is the one who showed me how to do it. The only thing I know of off the top of my head to point to is this CLUE BLE patchwork demo that he and I worked on: https://github.com/FoamyGuy/Adafruit_CircuitPython_CLUE_ble_patchwork/blob/master/examples/clue_ble_color_patchwork.py
the variable bitmap is created as 8x8. then put into patchwork_group which is scaled up 30x so the final size is 240x240 (same as the screen on CLUE)
Yes, I would think you could put a bmp file on an SDcard and then put it in the slot and use OnDiskBitmap to draw it to the screen.
digital elbow
thanks, maybe its time to get the m4
manic glacierBOT
tough flax
@slender iron No problem with the approach
manic glacierBOT
half sedge
@lone axle @digital elbow Yes get an M4 (that is what you have on PyPortal) or nRF52840. The scaling trick is only to speed up display if what you want is to display big square... This does not change the memory that displayio might need to "mirror" the screen content into buffer. The speedup come that scaling is done in C and not in Python.
displayio on M0 has only been made possible because of the Gizmo on Circuit Playground Express and HalloWing M0 Express. Both do require special "firmware" with frozen displayio inside.
manic glacierBOT
manic glacierBOT
lone axle
@half sedge Ah, I see. thank you for the clarifying.
manic glacierBOT
half sedge
@half sedge, no learning guide in progress yet (I'm finishing a couple others), but hopefully soon. In the meantime, I started off with this as a base: https://learn.adafruit.com/circuitpython-ble-crickit-rover, used the neopixel and adafruit_motor libraries and found pixels are on board.D0 and left servo is on board.D1, and right servo is on board.D2. I don't want to give away too much, but that should get you started.
@gilded cradle I have a working Bit:Buggy thanks to you: https://gist.github.com/dglaude/3338c5ebd48bbd0d9f8b6c221ba7d07c ... maybe I can adds the White_LEDs, detect proximity or diplay a changing image.
slender iron
@prime flower (when you have time) https://forums.adafruit.com/viewtopic.php?f=60&t=162631
prime flower
@slender iron I'm not feeling well, afk for a bit. could u email this to me so i remember when im back? thx
slender iron
Yup! Take it easy!
manic glacierBOT
manic glacierBOT
inland tusk
I was wondering if the teensey 4.0 verion of CP is available for testing?
solar whale
@inland tusk yes . https://circuitpython.org/board/teensy40/
You can use the teensy loader to upload the .hex file https://www.pjrc.com/teensy/loader_cli.html
onyx hinge
good morning!
timber mango
๐ฅฑ I woke up at 3:14 am.
main meteor
And had pie?
timber mango
Feeling a bit irrational...
timber mango
When I generate a PR, will it automatically be reviewed, or do I have to close it first?
manic glacierBOT
tidal kiln
don't close it. the review isn't automatic though, it's a manual process - someone has to actually look at the code. the only auto things are the CI stuff, which is up to you to get to a passing state.
timber mango
My PR has passed all checks. Is there a way to request a review?
I did not see any way to do that.
lone axle
@timber mango check the right hand column:
if you see a "reviewers" section with a gear you can click the gear and choose the appropriate team or person to request review. I'm not 100% sure if requesting review requires write access or not though. So possibly it may not let you.
tidal kiln
@timber mango can you link to the PR?
timber mango
@tidal kiln I see the reviewers section, but there is no gear there. My PR is https://github.com/adafruit/Adafruit_CircuitPython_RFM69/pull/27
tidal kiln
added reviewers. i think you need permissions to do it.
timber mango
Thank you.
idle owl
<@&356864093652516868> Meeting in just under 1.5 hours. Please remember to add your Hug Reports and Status updates to the notes doc even if you'll be attending the meeting - it's super helpful! Thanks! https://docs.google.com/document/d/1Ev0a-zNbTzmMbo__zPvqba9rX3kdfS6LahXAfJQveu0/edit
turbid radish
Are you subscribed to the weekly email CircuitPython newsletter? You can easily with no spam or selling your email to a list - see this blog post with all the details https://blog.adafruit.com/2020/03/16/the-informative-python-on-microcontrollers-newsletter-subscribe-now-circuitpython-python-circuitpython-micropython-thepsf-2/
onyx hinge
@fierce girder thanks !
turbid radish
Lurking, doing the ICYMI MakeCode newsletter for JP
tiny oriole
I will be absent. Group hug to all!
graceful heart
lurking
tiny oriole
overlapping meeting ๐
turbid radish
Group hug to everyone
modern wing
Quite so, that is how you do a virtual conference. I think folks will use that as a template for smaller communities that want to put on an event but don't have the resources for one.
raven canopy
I'll be a few minutes late. It's meetings meetings meetings... ๐ซ
sterile bronze
lurking today
lone axle
I am absent today
tidal kiln
lurking
timber mango
lurking today
river quest
as we get going folks, feel free to put any questions you may have for adafruit, what we are doing, how we are doing, anything at all, i can address and answer them during the community section. consider it an AMA "ask me anything" - pt
timber mango
Just mostly lurking today.
timber mango
I woke up at 3:14 am this morning (PST).
onyx hinge
@inland tusk will you have hug reports and status updates?
inland tusk
I am lurking
onyx hinge
@maiden chasm will you have hug reports or status updates for us?
@inland tusk thanks, ๐
river quest
We are 17,000 people! Post going up later!
https://discord.gg/adafruit
maiden chasm
lurking today
slender iron
wow 2,000+ online
river quest
The biggest worldwide online Show and Tell begins Wednesday, March 18th, 2020 at 7pm ET
http://adafruit.com/showandtell
https://youtu.be/VmVkX_7PkLs
slender iron
๐
river quest
micro:mag issue 7, CLUE 10 of 10!
http://go.micromag.cc/issue7
graceful heart
๐ฅณ
modern wing
10/10 --> ๐ฏ
river quest
hackster.io gets a CLUE!
https://youtu.be/E7NJklkyCBA
CircuitPython is fast for prototypers
https://twitter.com/infosnek/status/1239334447423426561
Twitter
What Is Adafruit IO?
https://youtu.be/bYhRhDvTP5c
Video is posted: IoT Design Week with Microchipโs โWizard of Makeโ Bob Martin and Adafruit
https://youtu.be/fu37Uf-CXNE
March 12, 2020: added USB device support (tinyusb stack), clarified that 2MB PSRAM is supported, not 4MB:
https://esp32.com/viewtopic.php?f=10&t=14532
Open Hardware Summit video, lots of CircuitPython:
https://www.youtube.com/watch?v=Ifin8nPZpGA
Free teaching resources for CircuitPython - Google Docs:
https://docs.google.com/document/d/1IFW0hVlbdIL8TCOloOHHzwv5pYj0YG106IjXT6b7Qq8/edit
Preview/Draft here, thank you everyone for the @ # and more!
https://github.com/adafruit/circuitpython-weekly-newsletter/blob/gh-pages/_drafts/2020-03-17-draft.md
modern wing
Thanks PT ๐
gilded cradle
Just for Pi day
half sedge
lurking
idle owl
modern wing
๐
tidal kiln
totes agree. huge hug to pt.
slender iron
simple pulsar
A late bit of lurking
neat folio
merely lurking today
onyx hinge
maiden chasm
Simon is totaly ok
onyx hinge
I don't know how far you can bend it ... you can find out, but then you need a new panel
and yeah cutting eye holes will wreck it all
@steady osprey do you have any status updates or hug reports?
modern wing
I got one of the 32blits myself, but haven't had any time to explore it yet.
simple pulsar
@modern wing Mine sits a few feet away largely unused ๐
ionic elk
I can't play Doom Eternal in my COVID exile so I guess I'll have to program classic instead ยฏ_(ใ)_/ยฏ
onyx hinge
tulip sleet
@onyx hinge re RGB LED: @brazen cedar is working on that now, so you'll want to coordinate
simple pulsar
[For Adafruit IO data logging] The size of the binary or the size of memory on popular boards when it's running?
onyx hinge
@tulip sleet awesome, I am happy to step out of their way on this one ๐
raven canopy
if only
brazen cedar
@tulip sleet awesome, I am happy to step out of their way on this one ๐
@onyx hinge Is this bug a really gnarly one? Its my first attempt at a contribution.
slender iron
@timber mango you have an rfm69 right?
onyx hinge
@brazen cedar I hope not, but you never know. Remember we're here to talk about it if you're feeling stuck
timber mango
Yes, several. ๐
slender iron
would you mind testing for jerry?
timber mango
Sure. I do not have his latest code though.
solar whale
its in pr24 of the repo
brazen cedar
@onyx hinge I'm currently reading through rgb_led_status.c and trying to wrap my head around the functions. I'm coming from a C# background so I might not use the right terminology for a C/C++ project.
onyx hinge
@brazen cedar we can help with terminology too, and also we don't snark at you when you get it wrong ๐
onyx hinge
3 players and a ... judge? we have the word "arbiter" in english but it doesn't seem the right one.
cursive condor
yeah a judge
timber mango
@solar whale Where in PR24 do I look for it? Is it an archive or separate scripts?
ivory yew
nope
timber mango
Oh, wait, I have to checkout your branch, right?
solar whale
@geeguy -- waht I do is - in my fork of the repo - checkout and update master then do git fetch upstream pull/24/head:pr_24 this creates a branch called pr_24. Then git checkout pr_24 and it is all there.
idle owl
@maiden chasm We have a guide on the recommended workflow for CircuitPython on Git and GitHub, if you were not aware of it.
maiden chasm
yes I've gone through, it was great point to start
idle owl
@maiden chasm Excellent ๐ That's the goal.
solar whale
or you can clone my fork of repo and then checkout jerryn_ack but I think it it is better to work from the PR
timber mango
@solar whale That command did not work for me.
maiden chasm
@slender iron you mean https://github.com/joedevivo/vscode-circuitpython
solar whale
do you have upstream defined or is it just origin. If it is not a fork use origin not upstream
idle owl
I need to drop off. Thank you everyone for participating today. Stay safe and remember we're here to help you in the ways we are able during this time - if nothing else, we can help you while keeping busy with projects. Remember to reach out if you need contact. We're all in this together. Cheers, all.
maiden chasm
yes it is only 0.0.1 version released
solar whale
git fetch origin pull/24/head:pr_24
ivory yew
it's been quite a while since we pulled from upstream micropython?
timber mango
@solar whale I do not know how to work from the PR. There is much I still have to learn. ๐
ivory yew
yeah we have some changes in py that deviate from upstream. ๐ฆ
cursive condor
who can I ask for pythonistas role ?
onyx hinge
@cursive condor I can add you
cursive condor
yes please ๐
onyx hinge
@cursive condor here you go
cursive condor
thanks ๐
solar whale
@timber mango you can clone it from here https://github.com/jerryneedell/Adafruit_CircuitPython_RFM69 and then git checkout jerryn_ack
timber mango
fatal: Couldn't find remote ref pull/24/head
solar whale
type the command you used
are you in your fork or just a clone of the adafruit repo?
gilded cradle
I'm going to drop off.
timber mango
OK, I have it now. I cloned your repo and then got your branch.
onyx hinge
@tulip sleet can you mute?
raven canopy
Gotta drop too. ๐
timber mango
I will start looking at it this afternoon hopefully. I am a little under the weather right now.
solar whale
No rush -- hope you fell better -- I'll test/approve your PR later this afternoon
timber mango
Ok, sounds good!
slender iron
timber mango
@slender iron I am actually in Spring Valley, CA right now, but hoping to move to WY before the end of summer.
slender iron
ionic elk
Oh I'm excited now I gotta get my projects ready
graceful heart
I think an hour long is a great length, the half hour time has been feeling a bit crunched for time.
modern wing
It's a great problem to have ๐
ionic elk
Gotta burn the slogans into everyones brains
onyx hinge
but 90 minutes begins to feel like a slog sometimes ๐ฆ
but not you all, you're great ๐ผ
ionic elk
We could also timecap status updates :/
onyx hinge
In my former life we were all about timeboxes, but they always felt a bit unfriendly. Not sure how it fits with the community nature of the meeting.
onyx hinge
afk, time to take out the laundry and then go on a walk even though it's cold
ionic elk
It's feasible we could split off the library stuff from the core
modern wing
I agree, it's nice having it recorded -- I can't always join so being able to spin it up later is my only option.
maiden chasm
@slender iron your stream about logic analyser was great, I enjoy to see how programming on low level looks, please continue streaming about this project
modern wing
My neighbors must think I'm bonkers, seemingly talking to myself all the time.
solar whale
Have a great week -- dog walk time...๐ถ ๐
modern wing
My 2001 outback had a similar tape deck, no AUX input. So i got one of those tape adapters to connect my phone to the car's speakers.
Allrighty, I'm out. See y'all later
slender iron
Hi all! Next week's community meeting will be at the normal time on Monday at 11 am Pacific / 2 pm Eastern here on Discord. If you are outside the US please double check the time locally because the US has already shifted for daylight savings time. All are welcome to attend. The notes doc for next week is: https://docs.google.com/document/d/1XThAvWXJ-9gQJQ9y4ME6Tz7oRQsSzMzG_aE_cGeWzt8/edit?usp=sharing <@&356864093652516868>
brazen cedar
Anyone have time for some C questions about the CP source files?
slender iron
I've gotta grab lunch but can answer after that
brazen cedar
ok ttyl
steady osprey
@onyx hinge No updates as of yet. I've tried the solution for the feather, which was unsuccessful. It seems like it isn't really fixable without a J-Link.
ivory yew
@brazen cedar always feel free to ask, someone else might be able to answer. ๐
brazen cedar
@ivory yew ok thanks! I'm digging in to the code for the RGB status LED on the Xenon. There's two things wrong (show the correct status colors, and allow the user to take control of status LED). There's a bifurcation of logic between the Xenon and Argon/Boron code. Which set of code is the one I should build out more or bugfix?
User control of the status LED seems to work on the Argon/Boron but not on the Xenon
the file I'm looking at is /supervisor/shared/rgb_led_status.c
slender iron
I'm back now too. good point @ivory yew
raven canopy
@brazen cedar in a quick look from my phone, there are differences in ports/boards/nrf/particle_[argon, boron, xenon]/mpconfigboard.h. The xenon has odd-man-out names: CP_RGB_STATUS_ vs MICROPY_HW_RGB_LED_ for argon & boron.
lapis hemlock
I have just got home, completely missed the weekly meeting. Sorry.
brazen cedar
@raven canopy I noticed that too, that's why I asked which is the correct name/methodology to build further upon and bugfix
the status led/user control should be that same across all the particle series I would imagine
raven canopy
I'd have to look deeper; heading home from work. But, if argon and boron work, that'd be an easy starting point to test.
brazen cedar
@raven canopy ok I look forward to your analysis
slender iron
I'd suggest doing the one that works. ๐
brazen cedar
@slender iron There's a todo in the code on the Xenon side that made me think that method was developed later but not completed so...idk. lol I'll give the Argon/Boron stuff a deeper look
slender iron
ya, sounds like it could lead to that
brazen cedar
Is a functioning status led and user control of the same led mutually exclusive?
slender iron
consistency is good ๐
no, you should be able to do both. there are some tricks to do in the pin_reset and claim_pin though
@brazen cedar https://github.com/adafruit/circuitpython/blob/master/ports/nrf/common-hal/microcontroller/Pin.c#L76
lapis hemlock
@slender iron Scott, @cursive condor requested two functions, around, and arctan2 to ulab. I have implemented those as part of the vector sub-module. However, these two functions add around 2 kB to the firmware. In the long term, would you like to keep them in vector, or should they be moved to extras?
brazen cedar
my C chops are so rusty haha this should be interesting
slender iron
@lapis hemlock I think vector is ok if they fit in existing builds. do we have other arctan implementations already that we could share with?
@brazen cedar ๐ we're happy to help get the rust off
lapis hemlock
@lapis hemlock I think vector is ok if they fit in existing builds. do we have other arctan implementations already that we could share with?
@slender iron I don't think so.arctan2is special in the sense that it requires two arguments with all the hassle that comes with it. It is significantly more complicated (i.e., adds more to the firmware size) than a vectorised single-argument function. Nowhere else isarctan2used/implemented.
solar whale
@brazen cedar C is like riding a bicycle ๐
slender iron
kk, all good then @lapis hemlock
lapis hemlock
@lapis hemlock I think vector is ok if they fit in existing builds.
@slender iron Actually, that was my question. I have no oversight of the tightness of the firmware on all your hardware.
slender iron
@lapis hemlock I think it'll be fine since ulab is only in larger builds. The easiest way to test is to update the submodule and see if everything builds
lapis hemlock
OK. I've got to log off now.
slender iron
thanks @lapis hemlock
lapis hemlock
@slender iron As soon as @onyx hinge can review the code, it can be merged.
slender iron
๐
onyx hinge
you can merge it without my oversight, I am sure it is fine
lapis hemlock
@slender iron I don't mind, if you review it๐ I just don't want to push into master without a second opinion.
you can merge it without my oversight, I am sure it is fine
@onyx hinge If it is not, someone will complain about it, and it can then be fixed๐ OK, I will do that, then.
slender iron
I don't see a CircuitPython ulab PR yet
cursive condor
I will not complain ๐
onyx hinge
@lapis hemlock is there a way for atan2 to leverage the code that binary operators like '+' use to reduce code space? Or perhaps you did that and it still cost 2kB..
@slender iron let me make sure my code is pushed for Protomatter before we chat
okay, not tidied, but pushed
lapis hemlock
@lapis hemlock is there a way for atan2 to leverage the code that binary operators like '+' use to reduce code space? Or perhaps you did that and it still cost 2kB..
@onyx hinge I wouldn't do that: the binary operators are type aware, whilearctan2doesn't have to be, because the result is always afloat, so we already save a lot of space there. I don't see an obvious way of reducing the code size further. And the 2 kB wasarctan2, andaroundtogether.arctan2alone is around 1.3 kB, if I remember correctly.
onyx hinge
I see, thanks
slender iron
๐
gonna run and get water then we can video
gonna run and get water then we can video
raven canopy
@brazen cedar just read the issue thread; sorry to jump in late and point out what has already been pointed out. ๐
simple pulsar
What boards is ulab available for/on, btw?
I'm just flicking through https://learn.adafruit.com/ulab-crunch-numbers-fast-with-circuitpython - first page doesn't tell me what boards are supported but I'll carry on flicking through it...
brazen cedar
@raven canopy no worries, we're all communicating asynchronously so it happens
slender iron
@simple pulsar I think it's M4s and up
good feedback for the guide so please file it in Learn too
simple pulsar
I stuck some feedback in.
slender iron
thanks!
raven canopy
hrm. its a corner-case for the support matrix too...
raven canopy
ahh. the MICROPY prefix explains it:
# ulab numerics library
ifndef MICROPY_PY_ULAB
MICROPY_PY_ULAB = $(CIRCUITPY_FULL_BUILD)
endif
What's the "support matrix"?
timber mango
My neighbors must think I'm bonkers, seemingly talking to myself all the time.
@modern wing There is a cup I want that says "Of course I talk to myself. Sometimes I need expert advice." ๐
onyx hinge
@raven canopy if I should change it on my end, can you send me a GitHub issue for it? (Ulab gating define)
raven canopy
@onyx hinge i'm not sure that changing solely for support matrix is warranted. but, it is the only entry in circuitpy_mpconfig.mk that has the micropy prefix. is there a reason for that?
onyx hinge
@raven canopy there's not a strong reason for it, no. I think it's related to wanting to enable it in the UNIX port for testing purposes
but I'm not sure that I can't rename it
raven canopy
ahh. that's a very valid reason, if so.
i understand the difficult bridge with ulab. the support matrix builder is a better candidate to adjust, imo.
timber mango
I might have a bad Feather M4 Express. ๐ฆ All indications so far are pointing in that direction.
granite ingot
Hi. I'm trying to port CircuitPython to a custom STM32F12Cx board and everything seems to work, except USB. (doesn't mount drive or serial device). Does anyone have any troubleshooting suggestions. Here's a more thorough explanation of what I'm running into: https://forums.adafruit.com/viewtopic.php?f=60&t=163516&p=803857
half sedge
Today I will "officially" use discord for business purpose. Thank you to the CP community, thanks to you I am an expert already! Also I look less old/stupid when talking to my son. I don't know who started this. For me it started with Pimoroni and also Gadgetoid own server. But this is where I use it the most. I also know Phil is promoting that usage for communities. You are awesome.
fierce girder
A tiny, CircuitPython-compatible ARM Cortex-M4 module
https://www.crowdsupply.com/null-byte-labs/circuitbrains-deluxe
fyi - my friend Kevin is launching this soon on CrowdSupply
solar whale
@idle owl do we need to update our local pylint installations to be consistent with the new changes. If so, what do we have to do? will the guide be updated?
@idle owl @trim elm sorry -- I merged a PR to the RFM69 library that created a conflict with your pending PR. I did not realize you were changing the adafruit_rfm69.py file - I thought your changes were all in the support files. This will also - I also have a pending PR that will requires a lot of updating to resolve conflicts with your PR. Thats not a problem. I just was not expecting it. Go ahead and merge your PR is you are ready an I'll make the necessary changes to mine to resolve the conflicts. I see you also are doing the same to the RFM9X library which I have been working on as well. I guess I was not paying attention to the pending changes....
trim elm
@solar whale thatโs fine. Thanks for the heads up
solar whale
@trim elm do you know if I need to change my local pylint setup to be consistent so I can check my PR's before pushing them.
trim elm
@solar whale using the latest version of pylint and running black should do it
solar whale
I have no idea what black is
lone axle
I learned yesterday about it a bit. It's a python code formatter
makes changes to whitespace and some other non-functional changes like " in favor of ' for strings and things.
trim elm
@solar whale https://github.com/psf/black install this and run it by using โblack .โ In the repository you want it to reformat
solar whale
Do I also have to update pylint?
ther eis a now outdated guide to using pylint https://learn.adafruit.com/improve-your-code-with-pylint
lone axle
I am not 100% certain but I think you would need to update your local one if what you had is 1.9.2
solar whale
Is running black now a requirement on all commits?
lone axle
In the build file it's changed from pylint==1.9.2 to just pylint. If I understand correctly that means it should be using the newest available version which looks like 2.4.4 at the moment.
not sure about whether black needs to be run locally or not though sorry.
solar whale
I guess, what I am confused about, is has there been a change in what we all need to be doing. I guess i missed the memo ...
lone axle
Well on the pylint front I believe actions will be using the new version due to the change in the build file. So if your local version was different the only risk is that there would be some things flagged differently if you run local vs. when it runs automatically after a commit.
solar whale
I every repo being revised to conform to black? i just had no idea so now need to correct a lot of conflicts in things I have been working on. Just caught by surprise
lone axle
The guide does probably need updating to not specify the 1.9.2. I hadn't seen this guide before actually. I need to read through it to get a better idea of how to use pylint locally to be honest. I've been relying on the default behavior of PyCharm + the github actions pylint for everything.
I believe every repo is being updated (all of the libraries certainly). If I understand correctly there was an automated script that pulled each repo ran black formatting on it and then pushed the results and made a PR.
The current process is reviewing those PRs to make sure that no functional changes slipped in, and that it didn't mangle any formatting that was intentional for code clarity.
But I'm not sure what the process is moving forward after all of these PRs are merged.
solar whale
There was a real push awhile back to get everyone to install and use pylint locally before pushing changes. I just want to keep up to date. I'll stop grumbling and update my pylint.. I just wanted to be sure it would still be consistent with what I get from from the updated repos.
lone axle
I think 2.4.4 would get it to match what the github actions are doing at this point. Though worth noting that with it not specifying a version in the build file I believe it's always going to take the latest version which is 2.4.4 now, but will continue to increase as time goes on.
raven canopy
You could setup a couple venvs, and keep things happy that way.
onyx hinge
@solar whale In theory, if you "black"-formatted your code before merging, it will reduce (rather than increase) the amount of conflicts. In practice, I don't know how it works out and I don't know it well enough to talk you through doing it
@raven canopy looks like I can rename the ulab-define just fine
@raven canopy for the support matrix
raven canopy
But does Unix port stay happy?
onyx hinge
yes
raven canopy
๐
solar whale
anyone know what this is trying to tell me Traceback (most recent call last): File "/home/pi/.local/bin/black", line 10, in <module> sys.exit(main()) File "/home/pi/.local/lib/python3.7/site-packages/black.py", line 5, in main raise NotImplementedError("This is still being implemented.") NotImplementedError: This is still being implemented.
I think it means black does not work with python 3.7 yet -- is that correct?
onyx hinge
What black version did you get? I got $ python3 -mpip show black | grep Version Version: 19.10b0 and line 5 is an import, not a raise.
solar whale
hmm pi@gjnpi4ram4g:~/junk $ python3 -mpip show black | grep Version Version: 0.0
i just did pip3 install black
timber mango
@fierce girder I looked at the specifications on that, and it does look very interesting. I hope it gets fully funded!
onyx hinge
that's sure wrong! python3 -mpip install --upgrade black?
timber mango
I do not use the "-m" - I just do pip3 install --upgrade black
solar whale
on my Mac I get [Jerry-desktop-mini:circuitpython/ports/atmel-samd] jerryneedell% python3 -mpip show black | grep Version Version: 19.10b0 the previous was on a Raspberry Pi .... will try update
onyx hinge
raspbian buster?
solar whale
yes
timber mango
Same here.
solar whale
upgrade does not help ```pi@gjnpi4ram4g:~/junk $ python3 -mpip install --upgrade black
Looking in indexes: https://pypi.org/simple, https://www.piwheels.org/simple
Requirement already up-to-date: black in /home/pi/.local/lib/python3.7/site-packages (0.0)
Requirement already satisfied, skipping upgrade: click in /usr/lib/python3/dist-packages (from black) (7.0)
Requirement already satisfied, skipping upgrade: attrs in /home/pi/.local/lib/python3.7/site-packages (from black) (19.3.0)
pi@gjnpi4ram4g:~/junk $ python3 -mpip show black | grep Version
Version: 0.0
pi@gjnpi4ram4g:~/junk $
one of my computers tells me to run it as python3 -mpip but maybe it's my own mistake. ```$ pip3
WARNING: pip is being invoked by an old script wrapper. This will fail in a future version of pip.
Please see https://github.com/pypa/pip/issues/5599 for advice on fixing the underlying issue.
To avoid this problem you can invoke Python with '-m pip' instead of running pip directly.
same result either way -- pip3 or python3 -mpip
on my mac I also have python 3.6.5 not 3.7 as on the RPi
onyx hinge
it looks like something has requested black version 0.0 to be installed. I don't know how to clear out that 0.0 to get the latest.
solar whale
black is happy on the Mac with 3.6.5
onyx hinge
oh that's interesting, I get exactly the same thing as you on my pi4 with raspbian buster
:smh:
simple pulsar
@solar whale Is it easy to install different versions (at the same time) of Python on Macs? I've not used them since a brief dabble in the early 90s, things have probably changed since then...
solar whale
@simple pulsar - probably, but I am not willing to do that....
main meteor
Yeah, I use virtual environments on my Mac all the time, I'm probably running 5 different versions of Python in dozens of different environments
solar whale
The Mac is not the problem
Sorry -- I'm just frustrated by the flurry of changes in the necessary build environment. Old dog...
ionic elk
@solar whale I really, really would recommend using virtual environments, no matter the experience level. It avoids so, so much awful garbage.
timber mango
@main meteor I am just starting to try virtual environments here.
onyx hinge
@ionic elk but right now the problem is, a bad version of black is installed by pip; virtual environments don't help address that afaik
it's not "I need version 9 and version 11 of black", it's "I keep getting version 0.0 of black, but nobody wants it"
timber mango
@onyx hinge Have you tried uninstalling black and reinstalling it?
simple pulsar
For Macs, are you guys talking about Python's virtualenv or something different?
ionic elk
I misunderstood then sorry
onyx hinge
np @ionic elk
@timber mango on this pi4 it was not previously installed. as jerryn said, upgrade and force-reinstall don't make a difference.
solar whale
I understand and appreciate the suggestion. i'll work on that. I would also find it helpful if I knew what environment I needed to setup ๐
ionic elk
Are you sure a virtualenv wouldn't help, though, in isolating the issue?
timber mango
@solar whale What are you trying to do?
main meteor
Yes, python's virtualenv. I like to use the one built in to python3 like ```python3 -m venv myvirtualenv
source myvirtualenv/bin/activate
I am trying to make sure my development environment is consistent with the github repo so when I submit a PR, it does not fail the build process.
main meteor
Thereafter, running "python" and "pip" automatically use the ones in the virtual environment, and store things inside it, avoiding system-wide incompatibilities.
onyx hinge
the broken black is coming from these people, who presumably mean well https://www.piwheels.org/
somehow, pip3 determines that a version 0.0 of black uploaded to pywheels is better than 19.10b0
this is enabled on raspbian but not on normal debian
workaround 1: Specify the version of black you want, to avoid getting 0.0. python3 -mpip install --user black==19.10b0
onyx hinge
workaround 2: disable piwheels, permanently or temporarily; but I don't know how. "python3 -mpip config unset global.extra-index-url` doesn't do it
I am filing an issue with piwheels
solar whale
one new trick for this old dog...
onyx hinge
workaround 3: install --pre to install prereleases might work, not sure. In PyPI all the releases are prereleases and there is no 0.0, so I guess it installs a prerelease anyway. but on piwheels, there's a version 0.0 which is a released version so without --pre the versions like 19.10b0 are not considered for installation.
๐ฆ
well now I know more about python packaging than before ๐
issue is https://github.com/piwheels/packages/issues/97 if you want to track it @solar whale
solar whale
@onyx hinge thanks for the followup and explanations.
onyx hinge
@solar whale of course. I always want an excuse to learn something
idle owl
@solar whale The guide needs to be updated, yes. We need to document somewhere that we're using Black as well. You can figure out what versions are necessary by looking at the build.yml file in any library repo - it has the install line necessary to install the proper bits. We recently moved that out of hiding so it was easier to find and replicate. If you'd rather a quick audio chat sometime today, I'm happy to go through what changes we've made and what they mean to you. Otherwise, I'm also happy to discuss it here.
I wouldn't have been able to help on Raspbian though.
solar whale
@idle owl i think the main thing I need to know is what version of pylint to install
idle owl
Also there is something called pyenv that lets you SUPER easily have different versions of Python for different things on Mac at least (I'm assuming on Linux as well), you can specify it per folder even if you want a particular repo to remain on a different version of Python.
The latest should be 2.4.4 - pip3 install pylint is what you should, in theory, run. We're running it without a version number, so if you install it without a version number, you'll be on par with us.
solar whale
OK -- great -- that should get me going ok. I'll let you know if I have other questions. I have black installed now -- on the Mac and Pi !! So I should be able to use them to keep my mods "compliant"
idle owl
Excellent.
I feel like, at least for me, it's something I'll eventually get used to and format "properly" myself without Black running, but it'll be a little time.
Or rather, Black will run and find less and less over time.
timber mango
First post: my Trinket MO flashed with the latest 5.0.0 firmware only have 45K left on the CIRCUITPY drive; is it normal? Is there a way to get more space? I also asked on the forum: https://forums.adafruit.com/viewtopic.php?f=52&t=163572
solar whale
I have been perusing build.yml -- it is very helpful
idle owl
It was less helpful not long ago. We fixed that.
@solar whale Apologies for not being clearer or more vocal about the changes. I guess I thought we had been, but thinking back on it, we were certainly not.
lone axle
@timber mango I don't know the correct number for sure off the top of my head but that does sound correct to me. The Trinket M0 (and any other non-express devices) have very little storage.
solar whale
@timber mango I'm afraid that is normal and a limit of the Trinket flash size.
@idle owl I think I was "out of the loop" at a bad time. trying to catch up.
idle owl
@solar whale You shouldn't have to be in the loop to know about things like this. We should have documented it better and announced it here a few times and so on.
In other words, it should have been more obvious even if you're out of the loop. Easier to find when you returned.
solar whale
Well -- in an ideal world ... Clearly, you have heard my concerns today and i appreciate that .
idle owl
Your concerns are entirely valid. It's something I will take into account moving forward and try to ensure we're more communicative about things like this before and as they're happening.
As long as we learn from mistakes, that's what matters. Making them in the first place is unavoidable.
@solar whale Thank you for voicing your concerns.
solar whale
It's nice to be part of a community where I feel comfortable voicing them.
@idle owl I do have a question about all the repo updates. Since the refomatting makes many changes to the actual library code, (for example - rfm69) are you planning to test them all on hardware before merging? or just looking over the code?
idle owl
Most of it is whitespace and changing ' to ". If it changes functionality, yes, we're testing. But 99% of the changes are not functional changes.
Or changing hex addresses to capital letters.
So mostly looking over the code.
timber mango
@solar whale I agree! I have always been hesitant to voice my opinions everywhere, because doing that usually leads to bad things happening. I do not feel that way here though!
solar whale
It's that "if it changes functionality" part that can be tricky .... it's not supposed to, but... I have not seen any issues in the ones I have looked at (rfm69, rfm9x) -- feel free to ping me to test ones I have hardware for if you need any help.
idle owl
@solar whale Agreed. I've erred on the safe side, in my opinion. But that doesn't mean I won't miss something. I'll ping you if I think something needs testing to see if you have hardware. Thanks!
solar whale
I come from a background where even if you only change a comment, it has to be fully tested. Not advocating for that -- just explaining my initial reaction to the seeing the changes...
idle owl
I understand. It made me a little itchy at first as well, but I read up more on Black and felt a bit better. You were also in a far more specialised situation.... you break things, you lose things in space. We break things, we submit a PR to fix it and it's good to go.
solar whale
move fast -- break things..
idle owl
I wish I was moving faster. This is taking much longer than I though it would, though if I had given it more consideration, I probably would have expected it. Going through the entire PR for each of them. ๐
solar whale
I'll leave you to the fun -- good luck!
idle owl
Thanks!