#circuitpython-dev
manic glacierBOT
indigo wedge
I'm curious, how are the priorities gonna be with stm port coming in, is samd still flagship or will we slowly move to stm? https://www.instagram.com/p/B2NfdL4H_E6/?igshid=5kwa8ogetqn8
plucky flint
๐
timber mango
@onyx hinge you have written that sometimes your computer (linux) locks up?!
i have a eventually similar / related problem sometimes here:
- ItsyBitsy M4 Express (about newest UF2 bootloader)
- Kubuntu 19.04 64bit
- Laptop: Dell Latitude E6510 (~9Years Old..)
i currently mostly upload Arduino firmwares - this with the Bootloader and the UF2 drive thing.
sometimes my computer locks up - in a way i only can do a hard shutdown...
i think it has to do with the usb hardware/software and the multiple & fast enumerations....
and till now i did not had patience to investigate more time into this...
manic glacierBOT
onyx hinge
any deep reason why "thing plus" (samd51) isn't CP-supported? https://www.sparkfun.com/products/14713 no qspi, I guess, which might make it feel limited.
onyx hinge
hm, the schematic says it has "4Mb flash" (SPI connected). I wonder if that's 4Mb = 512kB, or 4MB.
main meteor
I'm guessing 4Mb (512kB), but since the part number isn't given and I can't read it on the board pictures, I suppose the next step would be to download the EAGLE files and see if the part is defined there. There is a zip file in the github, but I don't know if it contains a BOM or not.
onyx hinge
yeah no part number is visible on the schematic. I didn't bother trying to download the eagle file
"You have an order status of Exception. Don't worry! This order just requires a quick review by our Customer Service team before it can ship" blah! (sparkfun -- I decided to pick up one of those thing plus boards, plus another $180 of stuff I probably don't need ๐ฐ )
main meteor
I just had my shipping switched to FedEx on a Digikey order because something (I'm guessing flux remover) required special handling.
old smelt
@main meteor and @onyx hinge - I just happen to have a Thing Plus on my desk and a microscope within walking distance. The part number on the memory is 25SF041, and it's an Adesto part.
main meteor
That's a definitive answer, thanks! Sure enough, it's 4Mb = 512kB.
I love it when people "just happen" to have a microscope (and the object of interest) at hand.
onyx hinge
thanks @old smelt !
old smelt
You know... There's really no good "You're welcome" emoticon, is there?
lone sandalBOT
stuck elbow
@old smelt probably because it's not an emotion
tulip sleet
@indigo wedge There is no movement toward replacing SAMD. STM Arduino support is not great, and there are still plenty of Arduino users of our boards. Some STM chips have quirks, such as not enough USB endpoints, etc. The more chips we can support, the larger the ecosystem.
indigo wedge
Great to hear ๐
Just wanted to make sure cause I have boards in the pipeline that use CP and they are samd
manic glacierBOT
idle owl
@tulip sleet Question about cap touch on Bluefruit when/if you're around.
tulip sleet
@idle owl ok back
idle owl
@tulip sleet Ok. Does the same CircuitPython "calibration" concept apply to the Bluefruit as the Express, as in on board start-up, it calibrates based on what's connected to the touch pads?
tulip sleet
yes, but Nick picked a different threshold calculation method:
self->threshold = raw_reading * 1.05 + 100;
if that seems like its not consistent, then we can change it
idle owl
For the purposes of this, that isn't relevant. I wanted to make sure the bit in the guide that said to reboot the board if you're getting weird readings was still accurate or if I needed to add a note about the Bluefruit
Good to know, I'll keep that in mind
tulip sleet
the old one is just "not touched value" + 100
idle owl
Huh ok
tulip sleet
ok, sounds good
idle owl
What's the 1.05 do then?
5% added to it for a better buffer?
Or a more "relative" buffer anyway..?
slender iron
@onyx hinge @old smelt would love to support thing plus! just haven't gotten a pr from anyone to add it
tulip sleet
yeah, maybe, I'm not sure how he arrived it, and how much experimentation he did
slender iron
@tulip sleet you may want to chime in on https://forums.adafruit.com/viewtopic.php?f=60&t=156151
idle owl
It kind of makes sense because +100 could be minute at higher readings.
Anyway, good to know, I'll keep that in mind in the event we get feedback on it.
tulip sleet
I did a lot experimentation to get 100. It used to be a much higher multiplicative threshold. But i saw that the touched/not-touched values were not scaling when the wire got longer
idle owl
Interesting
I do remember it being terrible for a while.
The backflips I had to do to get the key limes to work on that first project were ridiculous.
tulip sleet
yes, it was way too insensitive
idle owl
They interfered with each other so badly.
I obviously assumed it was something I was doing. Until JP ran into the same thing and we fixed it. ๐
sterile bronze
@idle owl , @tulip sleet what I've done in my programming for cap touch and ldr connections is a run a calibration at boot up that reads the pin for 100 to 200ms and save the highest raw value read and then add 100 to the value. This usually works well for calibrating the input threshold.
tulip sleet
@sterile bronze do you mean in CircuitPython or using something else? We read the value when the TouchIn object is created, but just once, and then add 100 (for QTouch/FreeTouch on SAMD21)
sterile bronze
In CP I usually add it to my program but I have done this for other boards when the pin is assigned. I haven't written any bootloader files for CP.
idle owl
Is microcontroller.cpu.temp supposed to be in 0.25 degrees C increments?
tulip sleet
@idle owl not deliberately, but it might come out that way. I wouldn't write that down
idle owl
@tulip sleet Hmm ok. Thanks.
tulip sleet
it might be due to the granularity of the reading. There was just a formula to apply to the raw values
sterile bronze
The boot up time for the CP boards is very fast, which is very nice, so it may not be practical to add it when the pin assignment is done. Most of my previous work has been at the industrial end on dedicated machines.
idle owl
@tulip sleet Because it's definitely coming out that way on the Bluefruit.
tulip sleet
@sterile bronze since we assign pins dynamically, we wouldn't check until the pin is claimed for touchio
idle owl
My screenshots in the guide imply that it wasn't the case on whatever board I used for the screenshots.
tulip sleet
hmm, ok, you could just say the precision is 0.25 deg C. I'll look at the code.
idle owl
The guide page is meant to be general to apply to all of the boards. Except it refers to the ATSAMD21. So I'm assuming we wrote it before we had M4s.
idle owl
I guess I should test it on an M4.
tulip sleet
@idle owl hmm ok, data sheet says:
6.28 TEMP โ Temperature sensor
The temperature sensor measures die temperature over the temperature range of the device. Linearity
compensation can be implemented if required by the application.
Listed here are the main features for TEMP:
โข Temperature range is greater than or equal to operating temperature of the device
โข Resolution is 0.25 degrees
idle owl
Hmm
tulip sleet
so you can say that in the CPB guide. On the SAMD chips, there's a much more complicated formula to compute the value, involving multiple readings and calibration values
idle owl
We're not really doing separate pages for the CPB, so it'll have to be a note on the general page. And it's the same part on CPX, which I'm testing now.
oh wait
tulip sleet
are you using microcontroller.cpu.temperature? That's what I'm talking about
idle owl
I was thinking thermistor.
Yes that is what I'm talking about.
I conflated the two in my head just now.
So the nRF has 0.25 resolution.
is the important piece of information here. And the ATSAMD do not.
tulip sleet
yup. It actually returns an integer which we divide by 4
idle owl
Hmm ok
idle owl
Same for M4?
tulip sleet
but no reason to say that
idle owl
Right
tulip sleet
M0 and M4 are similarly complicated but not the same
idle owl
But the resolution on the M4 is like the M0?
Trying to figure out what I need to add to this page as notes.
tulip sleet
one thing about terminology: we casually say M0 and M4, but those are really ARM architecture designations. The SAMD21 is an M0, the SAMD51 is an M4, and the nRF chips are also M4.
idle owl
Good call.
tulip sleet
so I'd avoid using M0 and M4 to designate classes of boards, use SAMD21 and SAMD51 and nRF52840 instead
looking in SAMD51 datasheet
tulip sleet
there's no defined accuracy or precision for the temperature sensors on-chip on SAMD21 and SAMD51
tulip sleet
the raw values are taken from ADC readings, and then there's a formula using multiple calibration values to calculate the final value. The SAMD51 datasheet is actually incomplete about how to use the ADC to get the temp: I had to do trial and error to see what worked
onyx hinge
Should I chuck in an issue for samd51 thing plus in the u2f bootloader repo too?
er u2f must not be what I mean
uf2
or maybe it already has one
idle owl
@tulip sleet SAMD51 looks like 21 in terms of 4 decimal places returned for temp. Thanks!
onyx hinge
"In addition, this board comes flashed with the same convenient, UF2 bootloader as the RedBoard Turbo." yay
tulip sleet
@idle owl do you mean you see slight variations in the 4th digit? printing 4 digits is just the size of the floats.
and the accuracy is much worse than that
idle owl
No, I guess I didn't compare the two, I simply checked the resolution.
We explain in the guide it's not accurate or precise.
But that it's close
tulip sleet
@onyx hinge we thought you might be a good choice for working on #2142 above; added to Basecamp
onyx hinge
I saw some signs that j-link can do detailed instruction level coverage reporting, which would be useful in answering just how many cycles background tasks slurp up when they are called. I will have to see whether I can get that to work.
tulip sleet
@onyx hinge I think you can just set the USB background on a 1ms intervl and comment out most or all of the other background tasks, and then put them back one at a time. I think that will still lead to a functioning CPy, and you'll be able to measure each one, and how improvement to each task's overhead contribute
as I mentioned in #1933, I was able to get CPy 3.x to run with no background tasks for a little while, and so got an upper bound on improvements
old smelt
Is STEMMA QT intended only for sensor boards/small devices? The documentation kind of leans that direction. The QT connector is nice and compact + QWIIC compatible which is dandy.
It looks like the new Feather STM32F405 uses the QT connector. So, apparently that's cool? Just verifying...
meager fog
yep will have a stemma QT
its back compatible with qwiic - only diff is it will work with 5V or 3V
QT is for..cutie ๐ smaller connector! great for lil breakouts
old smelt
Great, thanks!
old smelt
We have something of a requirement for QWIIC, but like the STEMMA a lot. QT solves the problem.
meager fog
yep works just fine with 3V logic
more sensors coming soon
lemme know if you have any requests
old smelt
Will do!
idle owl
@pastel panther Is MSA301 not in the bundle for a reason?
pastel panther
Likely I forgot?
idle owl
That's fair. Wasn't sure if it was WIP or something.
Looks like the only important one missing from the bundle.
pastel panther
ah poop, gimmie a minute
idle owl
No rush, simply ran the check to verify.
pastel panther
ya, It'll just take a sec and then I won't have t remember again
idle owl
Right on
pastel panther
indigo wedge
speaking of stm: https://www.st.com/en/microcontrollers-microprocessors/stm32l552zc.html
Cortexยฎ-M33 @ 110 MHz, 512 Kbytes of Flash memory and 256 Kbytes of SRAM
- Four I2Cs
- Three SPIs
- Three USARTs, two UARTs and one low-power UART
- Two SAIs
- One SDMMC
- One FDCAN
- USB device FS
- USB Type-C / USB power delivery controller
๐
idle owl
@pastel panther I requested changes, in case it didn't notify you.
pastel panther
@idle owl Didn't check my email yet. I responded
pastel panther
@idle owl Bitte
idle owl
@tulip sleet Other than removing the 2.x code and 3.x references, any other changes you'd want to make to this code? It still works great, but I wanted to make sure there weren't improvements to be made. https://learn.adafruit.com/adafruit-circuit-playground-express/playground-sound-meter
And since I'm refactoring a lot of the CPX code right now, now's the time to do it if we want to.
tulip sleet
@idle owl, no I think it's fine
idle owl
Right on, thanks
onyx hinge
obsessively removes vias from a board design in eagle
pastel panther
๐
@idle owl or other mac users: Do you have any idea how to keep OSX from pooping ._adfruit_foo.mpy files into my lib when I try copying over the bundle?
Or am I strange?
ionic elk
@slender iron let me know if you could take a quick look at my comments regarding the bit packing, as they'll impact the wrap-up of analogio as well.
slender iron
@ionic elk I'll look tomorrow. In the ble weeds at the moment
idle owl
For circuitpython.org are the names of the boards on the downloads page pulled from the circuitpython-org GitHub or are they pulled from CircuitPython itself?
@gilded cradle Do you know? ^^
slender iron
@idle owl the human readable names are from the -org repo
idle owl
@slender iron Ok thanks. One of them is wrong. Wasn't sure where to fix it.
gilded cradle
@idle owl sounds like scott answered your question already
idle owl
Yep, sorry to bug you! Scott seemed busy so I figured I'd ask you since you dealt with it a bunch.
Thanks for the merge!
gilded cradle
yw
@idle owl, I'm not sure if you saw the circuitpython.org guide I did, but it might help: https://learn.adafruit.com/how-to-add-a-new-board-to-the-circuitpython-org-website
idle owl
@gilded cradle Yeah I wasn't sure whether it applied to that name, and if it didn't, figured it wouldn't be in the guide. So I asked instead.
gilded cradle
Makes sense ๐
idle owl
Ok off to assemble a bunk bed. Later!
gilded cradle
Have fun
manic glacierBOT
slender iron
@tulip sleet you around?
meager fog
@ionic elk hiya if ya give me a bin for '412 ill test it now!
ionic elk
It's a draft! Let's start the testing process in full after Scott gets a chance to review the bit packing stuff tomorrow and I integrate it in there.
onyx hinge
hmmm what to do tonight
๐ผ . o O ( Work on something assigned ) ๐น . o O ( Work on something fun )
๐ผ . o O ( Tidy your working area )
onyx hinge
hmmmm
so this can be made to "work", except it wouldn't survive from boot.py, contrary to micropython: ```>>> import busio, board
u = busio.UART(board.TX, board.RX)
u.mirror_console = True
@meager fog to work on something fun of course
onyx hinge
๐
meager fog
still count it towards your hours if its circuitpython
since your idea of fun is still, umm, work
onyx hinge
I have a project to hook some kind of python-on-hardware board to a Commodore 64 like a good old fashioned serial modem
I will keep that in mind, thanks.
meager fog
hmm over the uart? you should be able to, we talk to cell modules over AT
did you end up grabbing any nrf feathers?
onyx hinge
Yeah, I can talk to it via UART, but what I would LIKE is for the code.py print() to talk to it, for the repl >>> to talk to it, etc.
Yeah, danh alerted me and I think I got an order in on the last one in stock
so as far as concrete learning goes, I now know what the heck it means when it is written: MP_STATE_VM(serial_mirror_obj) = (mp_obj_t)self; -- I still couldn't explain it though.
onyx hinge
micropython has a form of repl redirection, but I couldn't get it to work 100%, and it is not at all configured to work with how circuitpython works
I hear scott is motivated to work on this too, because he wants to do a bluetooth keyboard + displayio screen as standalone
@meager fog btw thanks for taking good care of me. I look forward to having fully half my time to spend on CP stuff in November. I know you're busy, but it would be nice to have an actual conversation with you sometime as well.
onyx hinge
That is a good plan to have
old smelt
About a month ago, I opened a bug for displayio that has been added to 5.x.x-Bug Fixes. I presume that if I wanted to hack away at a fix for this, I would need to create an environment where I could do full build of CP, right? I'm not a C programmer, so this could be an interesting exercise.
onyx hinge
@old smelt Yeah that would be a first step. What environment are you usually in -- mac, linux, windows?
meager fog
@onyx hinge yah repl redirection is cool - i think...poke at it for a bit. but if it looks like more than a day's effort lets keep cuttin' out those audio bugs so we have super awesome audio. another thing to check btw is that you can do displayio without audio cutting out (display will be slower of course). as is, audio irq starves
old smelt
Mac
old smelt
...generally, though I flexed some old Linux muscles last weekend and put Ubuntu on an old Windows laptop
onyx hinge
@old smelt you can totally do it on mac
meager fog
im on win10 but most others are on macos
old smelt
A-ha! Thanks for the link.
onyx hinge
@meager fog yeah I will check out what I can see on my pyportal with display.
old smelt
I recall seeing that before, but I couldn't recall where I had seen it.
meager fog
jep, i am psyched about all the bugs you're finding & closing - epic work there!
onyx hinge
@meager fog so far on nRF I approached it by putting in more BACKGROUND calls, which is .. not ideal. Do you see a different route? (and yeah ful repl redirection/duplication looks like much more than one night of a half-handful of hours)
meager fog
i dont know - it may not be totally solvable without preemptive handling
onyx hinge
I will also try to ask scott what he says and check back over what he has written on previous issues I can find.
meager fog
which is like "man why did i build an rtos in python"
like right now 3 things happen in the bg
usb, audio and display
onyx hinge
yeah, only "audio" might now be 3 or 4 different kinds of audio, etc.
onyx hinge
I sure have imagined setting up some kind of deadline system
on nRF, unfortunately not. Their "easydma" means that each individual peripheral manages its DMA through unique registers within the periph
they even do double buffering at least two different ways
(PWM is one way, PDM and I2S are similar)
onyx hinge
there are interrupts that could be enabled of course, and the EVENT registers are quick to check
onyx hinge
also with audio you don't really know how many ms of audio one "chunk" is, it depends on lots of details. I think WaveFile is tuned to try to read about 1 block from flash per call but I'm not 100% sure of that assessment. But 512 bytes can be anywhere from 64ms (8kHz, 8bit mono) to 3ms (44.1kHz, 16bit, stereo)
meager fog
right
onyx hinge
so .. interrupts to set flags, quick checking of event-done registers, having some kind of deadline scheduler in the background stuff, ... I dunno what all to do. but it still all depends on anything long-running cooperating, too
meager fog
ok well, dont need to do it - just think on it as you do other stuff
if it comes to you in a flash, gr8
if not, it works as is
onyx hinge
haha okay
I will also be thinking awful hard while I'm on vacation. Anything I should pick up for my work while I'm in japan for 3 weeks? ๐
meager fog
enjoy your break!
onyx hinge
I am sure I will. It's a new destination for me.
meager fog
if you have a hankerin' you can always take a pygamer/pybadge or circuit playground bluefruit with ya
they're small
onyx hinge
haha haven't been able to find them in stock yet
well I have a cpb
I should take that, and finally learn anything about bluetooth
it's bedtime here! ttyl and thanks again
this is a blast
plucky flint
๐
fluid moss
Nicholas!
old smelt
I'm trying to get a CP build going on my Mac, and I'm getting the following error out of Make:
GEN build-pyportal/genhdr/mpversion.h
python ../../py/makeversionhdr.py build-pyportal/genhdr/mpversion.h
make: *** No rule to make target `freetouch/adafruit_ptc.c', needed by `build-pyportal/genhdr/qstr.i.last'. Stop.```?
onyx hinge
@old smelt looks like you did not get the "submodules".
checking for a doc link for you...
git submodule update --init --recursive
``` these steps should have done it, but maybe they encountered errors -- documented near the top of https://learn.adafruit.com/building-circuitpython/build-circuitpython under "Fetch the code"
.. run those submodule commands from the top directory of circuitpython, which is different than what that gif included above shows
old smelt
Darn... I did forget to do that.
Thanks!@
BTW - the guide suggests creating a case-sensitive image for the builds. Any recommendations for size? How big will this thing grow when doing the builds?
main meteor
It's not huge. I set up a 10GB disk image for building CP, and only 1.2GB are used.
old smelt
Thanks @main meteor. Still bigger than I was expecting, but gives me some guidance.
solar whale
IIRC the only reason for the case-sensitive FS was for ESP8266 which is no longer needed. I don't think it is necessary.
old smelt
oh really?
Ok - that's good news. I'll try without the separate image.
I just didn't want to chase unnecessary issues.
main meteor
There's some other stuff in there, principally the library bundle, and some bitmap and display code I'm working on. OTOH, it's not up to date, so if CP 5 is larger, it might take up more room.
solar whale
I just built the curremt master on my Mac in a standard FS
old smelt
Speaking of CP5... How do I know what I am building, anyway?
solar whale
the master branch is latest code -- bleeding edge -- you have to select a branch or tag for something else -- then make sure you reload the appropriate submodules!
old smelt
Success! Thanks @solar whale, @main meteor, and @onyx hinge .
solar whale
I recommned checking with @tulip sleet regarding the need for the case-sensitive FS in case something has changed.
I don't use my Mac much for building CP -- normally on a Linux box, but I like to keep my options open!
It certainly seems to build OK with out it.
old smelt
Will do.
old smelt
Directory grew to ~820MB with this initial build.
ionic elk
@tulip sleet is there a reason why the microcontroller pin interface is inconsistent about whether it takes pin information directly, or a pin object? For instance, the claim_pin function takes a pin object, but most others take a packed pin number. Is there a good reason against making them consistent one way or the other? Just kind of a stylistic choice.
I guess I just don't understand why "claim pin" should use an object, but "reset pin" should not.
onyx hinge
@ionic elk this impression is halfway baked, so take it with a grain of salt. Some of these I/O objects (like I2C and SPI right now) need to be able to survive reboots of the Python interpreter. That means that their members shouldn't be Python objects, and leads to storing the integer pin identifiers instead. I became aware of this only last night while working on UART console but even if this is not the REASON it sure is important for facilitating this.
ionic elk
If it is, I think there should be a note of that in the file itself.
onyx hinge
sure, I can agree with that statement
and I may be way off base with my assessment too
ionic elk
What uses the I2C and SPI raw right after a reset like that?
onyx hinge
displayio
it maintains the connection to the display over python reloads
I think this may all stem from scott's displayio work in 5.0
ionic elk
does it also go in and claim those pins pre-emptively?
onyx hinge
yeah there is a function called something like never_reset_pin which it passes those pin numbers to
@slender iron can you weigh in on any of this? I'm going 80% on guesses here
ionic elk
That makes sense, generally. I guess my issue is that since this is a port-level interface, not a common_hal one, the interfaces between ports are starting to get very inconsistent.
Scott's currently recommending that we use bitpacked fields, which means all my functions are going to have two parameters, specific to stm32, instead of a single packed int.
claim_pin won't have that problem
I don't know if any of this matters really though.
orchid basinBOT
onyx hinge
I did see some of this discussion go by, but I can't say I followed it all. It's entirely possible that scott is wrong (for instance, if his design means you can't really implement reset_pin, it is probably not a design to continue with!) and until you are confident he understands your concern, you should continue to work to help him understand.
(but at the same time, if it becomes clear to you that either choice is going to work out technically, then I would personally accept his preference after making my best case about whatever the topic is..)
afk for some lunch!
ionic elk
Yes either will easily work technically, this is really just a style choice. Thanks for the power cycle info, @onyx hinge
slender iron
claim pin could take the number as well. reset takes it already so that python objects only need to store the number (1 byte) instead of the pin object pointer (4 bytes). It has nothing to do with the displayio lifetime stuff. pin objects are static and live forever too
@onyx hinge I think the audio playback smoothness might just be a matter of having larger buffers to allow for larger gaps between background calls
idle owl
Ok, I hit the audioio/audiocore issue.
@onyx hinge Are you around? I'm guessing not.
@slender iron Is there no audioio in CircuitPython 5.x for CPB?
slender iron
correct
idle owl
Related: is there a way from the REPL to list all the available CP modules?
slender iron
help("modules") iirc
idle owl
Ok. How do you do AudioOut then?
idle owl
Ok
Found it. Thanks.
@slender iron Is the audiopwmio specific to the CPB or is it a 5.x change? As in, are all boards switching to that or is it an nRF situation?
slender iron
np
onyx hinge
@idle owl did Scott get you the info you needed?
idle owl
@onyx hinge Yeah, but I'll want to discuss some more details with you later if you're around.
onyx hinge
@idle owl how's 7 or 730 us Central work for you?
idle owl
@onyx hinge 7CT should work.
onyx hinge
Feel free to ping me but I think I'll be home
idle owl
Right on, will do.
idle owl
@onyx hinge 7:30CT might work better or possibly 7:45. I'll ping you later to confirm.
onyx hinge
Okay
meager fog
@tidal kiln p0ng
tidal kiln
p1ng
meager fog
@tidal kiln whatcha hackin on
tidal kiln
misc. doing a general house cleaning sweep.
basecamp / gh / forums / etc.
got something specific?
manic glacierBOT
meager fog
@tidal kiln lemme think - well, dya wanna try that recorder project with the cplay ble?
its faster, maybe works now?
tidal kiln
sure. can at least try it and see what happens.
solar whale
Are there any actual differences between the 4.x and 5.x bundles at this time?
manic glacierBOT
solar whale
there is no difference between the VERSIONS.TXT files so I guess not.
idle owl
@solar whale No, it's to continue the way we have been to alleviate confusion as to which bundle to use.
solar whale
@idle owl OK -- thanks -- I get confused easily ๐
crimson ferry
diff reported that the .mpy files may differ, but I assume that's just build differences?
solar whale
I saw that too, I suspect is is just some stuff in the binary that contains build details.
I tried doing some binary compares of .mpy files and just get more confused ....
I'll look into it more another time -- I'd like understand the differences, but it's not critical now .
onyx hinge
hm looks like I'll need to read that guide about running the linter locally ๐
yay local pylint
ruby atlas
yeah. as long as you remember before giving travis a workout ๐
idle owl
@onyx hinge In a meeting, so I can't test sound until we're done. I'll take a look at your PR in a bit here.
old smelt
Hi All - I did a local build of CP this morning, and I got a successful working UF2 which was pretty encouraging. It looks like it's CP 4, though:
Adafruit CircuitPython 4.0.1-5-g63b253c33 on 2019-09-11; Adafruit PyPortal with samd51j20
What do I need to do to build CP5?
Was I missing a switch?
onyx hinge
@old smelt any chance you simply transferred the wrong .uf2 file?
old smelt
Donโt think so. Iโll double check though.
onyx hinge
what does git describe --tags from anywhere inside your circuitpython source say?
onyx hinge
okay, well, that at least makes sense then!
are you at all experienced with git at this point?
old smelt
Some yes.
slender iron
@old smelt what does your git log show?
onyx hinge
It seems like you have run a command to switch to the "4.x branch". This is the "git checkout" command, for example. Perhaps the guide suggested to do it without explaining what the purpose was.
since getting the source, did make any changes? If not, you can easily move to the branch for 5.x, which is called "master": git checkout master
old smelt
git log kicks a bunch of output. Anything specific you want to see?
But a lot of commit dates in May
@onyx hinge , no changes yet. Just running a pipe cleaner right now.
Mac terminal does mysterious things with command history, so I can't reconstruct my build commands very easily. Kind of annoying...
onyx hinge
OK, well, recommend doing git checkout master and then git describe --tags. Then, it should say something more like 5.0.0-alpha.2-55-g89fed709a
onyx hinge
git remote -v | grep origin says ?
old smelt
Ah -ha! I think I may have figured it out. I orginally fetched my fork on GitHub which would have been CP4
onyx hinge
@old smelt okay, that could def. explain it
old smelt
Ugh... Sigh.
onyx hinge
it might be easiest to start with a fresh clone from https://github.com/adafruit/circuitpython if that is not a problem
(though eventually, to create a pull request, you do need git to know about both adafruit's circuitpython and yours as different "remotes")
old smelt
Sounds good. I'll start with cloning the adafruit repo. Learning...
idle owl
@old smelt You can add a remote that is the adafruit repo, and then fetch/merge that, and then push it to your repo to update.
onyx hinge
yes, git gives you a lot of choices of how to manage things. you could sure git remote add adafruit https://github.com/adafruit/circuitpython, git fetch adafruit, git checkout -b my-new-branch adafruit/master^{} for instance
idle owl
@old smelt Let me know if it doesn't work for some reason and I can try to walk you through it in more detail. Or Jeff can beat me to it! ๐
onyx hinge
which would let you start on a branch called my-new-branch, beginning where adafruit's master branch is
(please please please don't ask what ^{} means, just roll with it ๐ฌ )
old smelt
Ha! Okay...
Thanks for all the help.
Gotta go drive some cameras for a while. I'll try this out later tonight.
onyx hinge
see you later!
since you're interested in contributing a bug fix we are interested in helping you with the boring steps of getting git to work for you
idle owl
@onyx hinge What did you test the simpleio update with?
With the simple demo code I'm getting ValueError: A2 in use
Resetting the board didn't fix it.
onyx hinge
@idle owl I tested on Metro M4 and CPB. I used board.SPEAKER on CPB and board.A0 on M4
idle owl
Hmm.
onyx hinge
give me two minutes and I'll be back at my desk to check again. Are you on 4.1? I was on 5/master.
idle owl
I'm on a CPB running 5.x
Tried it with board.SPEAKER and it says AUDIO in use.
Was attempting to use a piezo on A2 first
code.py output:
Traceback (most recent call last):
File "code.py", line 8, in <module>
File "/lib/simpleio.py", line 97, in tone
File "/lib/simpleio.py", line 49, in _AudioOut
ValueError: A2 in use```No other code imported?
idle owl
import time
import board
import simpleio
while True:
for f in (262, 294, 330, 349, 392, 440, 494, 523):
# For the M0 boards:
simpleio.tone(board.A2, f, 0.25) # on for 1/4 second
# For the M4 boards:
# simpleio.tone(board.A1, f, 0.25) # on for 1/4 second
time.sleep(0.05) # pause between notes
time.sleep(0.5)
PIezo worked on A2 using PWM only.
onyx hinge
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "kattni.py", line 8, in <module>
File "simpleio.py", line 96, in tone
File "simpleio.py", line 49, in _AudioOut
ValueError: A2 in use
``` OK me toooh it plays one sound and then breaks. ๐ฆ ๐ฆ I am a bad programmer, it must be a problem in the core PWMAudioOut...?
idle owl
You're hardly a bad programmer. I'm not sure. Limor requested changes anyway.
Yeah one sound, then fails.
so something isn't deiniting I assume.
onyx hinge
yeah I bet I did not test that in core
I will file an issue in core for this. It would be useful if you verify it works on M0/M4 and 4.x right now, though
I will also look for Limor's requested fixes
idle owl
I'll test it on CPX running 4.x
onyx hinge
and yeah I shouldn't say things like "I am a bad programmer", it isn't helping anybody. It's just expressing how I feel in a moment, but it's a habit better off broken.
idle owl
@onyx hinge Works on CPX running 4.x.
onyx hinge
.. I'm going to evaluate a little more before declaring it a bug in core, I didn't reproduce it without using my modified simpleio yet.
but I'll follow up this evening after some more investigation
idle owl
Ok
onyx hinge
thanks for the quick testing! Please don't let me keep you from other stuff you need to do tonight.
idle owl
@onyx hinge Ok, so Limor wants me to do this try/except thing with importing audiocore as audioio to make the code the same, but in a piece of code that uses AudioOut and WaveFile how do you do that? Because it's two different things, audiopwmio and audiocore.
onyx hinge
quick and actually thorough
idle owl
You're welcome. I'm dealing with similar code at the moment it turns out.
onyx hinge
so you need to deal with the move of WaveFile to audiocore? Are you also trying to deal with AudioOut and PWMAudioOut at the same time?
idle owl
Yes.
So I can't simply import audiocore as audioio and audiopwmio as audioio it doesn't work that way
onyx hinge
first before I start giving a lot of advice, let me check on what the review notes on that simpleio PR say
onyx hinge
Yeah, let me sketch that. I think it's a valid request/criticism.
this is nice, it gives me a break from locking up my whole computer trying to do clever things with UARTs
from audioio import AudioOut
except ImportError:
try:
from audiopwmio import PWMAudioOut as AudioOut
except ImportError:
pass # not always supported by every board!
try:
import audiocore
except ImportError:
import audioio as audiocore
``` is what I have now. Would that code potentially help with what you are doing?it is probably better than what I had, I was trying to be too fancy
idle owl
That was close enough to get me what I needed. Thank you!
onyx hinge
oh hmmmm!
idle owl
This look rightish? ```python
try:
from audioio import AudioOut
except ImportError:
try:
from audiopwmio import PWMAudioOut as AudioOut
except ImportError:
pass # not always supported by every board!
try:
from audiocore import WaveFile
except ImportError:
from audioio import WaveFile```
onyx hinge
Yes I think that looks good
you want to just write WaveFile, not audiocore.WaveFile, when you use it
idle owl
I'm finding all the bugs it seems. ๐
onyx hinge
>>>
>>> import pulseio
>>> import board
>>> with pulseio.PWMOut(board.A0): pass
...
>>> with pulseio.PWMOut(board.A0): pass
...
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: A0 in use
finding bugs is so valuable, thank you! I will file an issue on core
idle owl
Why did you import audioio first on the first try/except and audiocore first on the second one?
onyx hinge
that's a good question!
I wanted to put the "most 5.xish, most mainstream" choice first. So that means audiocore for where to find RawSample/WaveFile pretty clearly
for where to get AudioOut, I think that boards with real DACs (all the M0 and M4 pretty much, right?) are more prevalent, so that's audioio first
onyx hinge
I think you could make different arguments, but that is my thought process as I can reconstruct it
idle owl
Makes sense
onyx hinge
now what is embarassing is that .. I am almost sure I ran into this and noted it as an issue to file the very first evening I set out to work on nrf52 "officially"
and did I file it? No. The dog ate my google doc
idle owl
I've had that happen!
onyx hinge
yay, fix in core is pretty simple. It's so easy to overlook things
old smelt
Progress:
Adafruit CircuitPython 5.0.0-alpha.2-55-g89fed709a on 2019-09-11; Adafruit PyPortal with samd51j20
onyx hinge
hum, I wonder if some of my problems with locking up are due to using the "-rt" kernel set. I booted into a "vanilla" debian kernel this last time, and haven't had a crash yet. ugh, I am used to the "-rt" kernel being pretty stable these days.
plucky flint
๐
cerulean sparrow
@onyx hinge What are you using that needs a hard realtime kernel? Seems like a large pain point when -lowlatency / -preemptive should handle a lot of cases.
onyx hinge
@cerulean sparrow I am a sometimes developer on LinuxCNC (http://linuxcnc.org) and expect to be doing more of that this fall/winter. It requires either the "-rt" kernel or even an RTAI-patched kernel, though the second option is becoming a more and more distant second choice. Usually being on debian's "-rt" kernel is just fine actually, but I'm hunting for any reason to explain how flaky my development experience has been lately.
main meteor
Might be time to do some forensics on your lockups/crashes.
onyx hinge
They're always preceded with kernel GPFs like this one ```Sep 8 12:40:58 soba kernel: [80072.030736] general protection fault: 0000 [#1] PREEMPT SMP NOPTI
Sep 8 12:40:58 soba kernel: [80072.030745] CPU: 1 PID: 16427 Comm: kworker/1:3 Not tainted 4.19.0-5-rt-amd64 #1 Debian 4.19.37-5
Sep 8 12:40:58 soba kernel: [80072.030748] Hardware name: BIOSTAR Group TA970XE/TA970XE, BIOS 4.6.4 02/02/2012
Sep 8 12:40:58 soba kernel: [80072.030762] Workqueue: events acm_softint [cdc_acm]
Sep 8 12:40:58 soba kernel: [80072.030802] RIP: 0010:usb_clear_halt+0x16/0xc0 [usbcore]
topaz schooner
Is there a CircuitPython install guide for the Feather M0 Bluefruit? From everything I've read, it should install, but I don't want to brick my Feather.
main meteor
I wonder if the kernel doesn't like the board re-enumerating its USB port.
The M0s are pleasantly resistant to bricking. This UF2 guide covers (re)installing CircuitPython along with a lot of possibly useful related information https://learn.adafruit.com/adafruit-feather-m0-express-designed-for-circuit-python-circuitpython/uf2-bootloader-details
topaz schooner
Thanks.. I got my Feather M0 Bluefruit a while ago when BT wasn't supported in Circuit Python yet. Now that it is, I really want to try it out. The Express doesn't seem to have any real differences from the Bluefruit, aside from some pin assignments and, of course, the BTLE module. Still a little scary. If anyone else has tried this, I'd really like to hear how it went.
One thing does stand out.. The Express is described as having a ATSAMD21G18 ARM Cortex M0+, where the Bluefruit just says "M0", no plus.
But I can't find any info that says what differences there are, or aren't.
main meteor
I think the main other difference is the Express has an additional flash memory chip (that additional storage is quite useful when running CircuitPython).
topaz schooner
Wondering if I should just bite the bullet and upgrade to the Huzzah32 (ESP32 Feather)
main meteor
Depends on what you want to do with it. The Huzzah offers WiFi but I don't think it supports Bluetooth. I think the M0 offers more in the way of I/O.
topaz schooner
The Huzzah32 might not, the Huzzah32 Feather does. Didn't know they were different until I just looked. But it still doesn't have official CircuitPython support.
solar whale
@topaz schooner Circuitpython will not support the BLE on the Feather M0 Bluefruit -- It is essentailly a Feather M0 Basic.
CP BLE support is only for the nrf52840 chipset at this time
as to express -- as noted, that implies there an external flash for the CP filesystem. The Feather M0 Bluefruit does not have one. It has a very small File system. on the internal Flash.
topaz schooner
That clears it up.. Thanks. Guess I'm sticking with C
solar whale
Sorry to bear bad news -- the feather nfrf2840 express is a nice way to go!
ionic elk
I'm having a lot of trouble getting the CI to work on my PRs - it's encountering mostly git issues with tarballs and authentication. I see one of @onyx hinge's builds has failed for the same reasons. Is this just a Github thing we can wait out?
manic glacierBOT
cerulean sparrow
Does the Monster M4sk run circuit python? Is there any documentation on how to edit the eyes?
Ok, it doesn't, but I found this: https://learn.adafruit.com/adafruit-monster-m4sk-eyes/customization-basics
ionic elk
@tulip sleet how did you protect the USB pins, and other vital stuff, from the chipwide reset_all_pins() function? I don't see anything that adds them to the never reset array.
When I searched for never_reset_pin_number, all that comes up is some busio and PWM dynamic additions. I was expecting some table in peripherals or port init to protect critical system pins, but I don't see anything. Are there just no GPIO pins shared with oscillator or system purposes in nrf?
manic glacierBOT
umbral dagger
@cerulean sparrow Yes it does: https://circuitpython.org/board/monster_m4sk/
slender iron
note that only one eye is supported from circuitpython
I haven't spent time to verify multiple displays work together ok
ionic elk
@slender iron is there a way for a user to check what pins certain functionality is on? Say if they want I2C, they ask what pins have it, or something like that?
idle owl
@ionic elk I wrote scripts that attempt to create the object on each pin or pin combination and returns the valid options.
Like this one for I2C which is easily adapted to any other protocol or functionality: https://learn.adafruit.com/circuitpython-essentials/circuitpython-i2c#wheres-my-i2c-15-17
If that's what you're asking.
ionic elk
@idle owl interesting. On the STM32, we won't have peripheral information per pin, since STM32 peripherals are extremely specific. But I wonder if I could tie into this somehow
idle owl
Also interesting. Did not know that about the STM, though I know nothing about them, so it's all new info for me.
ionic elk
sure it's good to know how it's done now
Ideally I'd like to have some option available to the user to either declare where every peripheral is mapped, or search for the pins of a specific one
idle owl
Right on, so sort of the opposite of the script.
ionic elk
Yes, and it'll have to be built that way all the way down to the port level
idle owl
This is stretching beyond my knowledge, but something like dir() for peripheral pins from the REPL?
ionic elk
I'll check that out, the python stuff is kind of out of my depth
so I'm not sure how this is exposed higher up
idle owl
That's what makes having a team so great ๐ Someone else can know the things you don't.
Right
I'm definitely referring to higher up.
dir() gives you everything something can do.... so it works on modules to show what you can call, but also works on, for example, board to tell you all available pins. Which is what made me think about the potential to have dir(i2c) or something and have it show you the available options... not exactly what it's meant to do, but something like it might be what you're getting at.
ionic elk
Yeah kinda. So, for instance, on the stm32 F411, I2C3 has exactly 2 pins: SDA on PC9, and SCL on PC8. So I'd like the user to either be able to punch in some "Peripherals" command and see a list that includes "I2C3: SCL-PC8, SDA-PC9" as a row, or punch in "I2C3" and see "Pins: SCL-PC8, SDA-PC9"
ionic elk
would that fit with the dir() command or would it be better suited elsewhere?
idle owl
That one I'm not sure on. I feel like you could create a peripherals thing and call dir() on it, but I'm not sure that's what it's really meant to do. Scott would know better.
slender iron
@ionic elk the goal is for user's not to know anything about I2C3 and just know i2c can be done on PC9 and PC8
which is what @idle owl 's scripts can tell you
ionic elk
The I2C3 label can certainly be left out, but the issue is that if they try to use I2C1's SDA and I2C3's SCL, it will fail
slender iron
how would they try to use it?
ionic elk
I'm not sure I understand
slender iron
want to voice or video chat?
slender iron
๐
ionic elk
@slender iron is there a way to receive notifications on github whenever an issue has a certain tag? It'd be convenient for tracking stm32 issues
slender iron
not sure, I get emails for all of them and filter them into a folder
silver tapir
BTW @meager fog, on this week ask and eng somebody asked about RISCV boards on adafruit. This chip has the requirements you mentioned. We mentioned it breifly in #risc-v.
Same ram an an M0, and has USB full speed which @slender iron wants.
http://dl.sipeed.com/LONGAN/Nano/DOC/GD32VF103_Datasheet_Rev1.0.pdf
Sipeed is actually making STM32 compatible board with this chip on the cheap.
https://www.seeedstudio.com/Sipeed-Longan-Nano-RISC-V-GD32VF103CBT6-Development-Board-p-4205.html
manic glacierBOT
manic glacierBOT
old smelt
@slender iron - have a minute to talk about the make_opaque()/make_transparent() issue?
slender iron
@old smelt may have some time tonight
old smelt
Ok. Just a quick question...
Your comment in the issue thread said that make_transparent()/opaque() needs to include a self-refresh.
Is that as simple as adding self->needs_refresh = true;
slender iron
ya, I believe so
old smelt
Ok. Tried that, and I'm seeing the same results
That answers my question. Thanks. I'll poke around a bit. Trying to understand how all of this stitches together.
slender iron
k, thanks!
plucky flint
๐
timber mango
sorry if this is not the right place to ask this but how can i get into using circuitpython on the game boy ?
stuck elbow
@timber mango you would need a special cartridge that @slender iron has developed
timber mango
is it possible to purchase these preassembled ? i do not have much hardware or electronics experience
stuck elbow
I don't know, you might need to wait for him to wake up
stuck elbow
this is very much work in progress still
by the way, I'm on a quantum computing sprint now, and now we are using CricuitPython to write a quantum game
there is a quantum computing library for it: https://github.com/quantumjim/aether
timber mango
thats amazing !
timber mango
Got hands on a uBlox NINA B3 evalkit, and tried to install CircuitPython on it. So I followed the tutorial on how to install it on the nRF devboard. Worked like a charm to get the UF bootloader up and running. (nrfjprog link is dead BTW).
Only thing, I just had only a 128MB quad SPI flash at hand. Is this a problem? Since I only once for a short time a REPL, and it never showed up as a CircuitPython USB drive.
ionic elk
@meager fog sorry I missed your question yesterday.
Let me know if you still need anything answered
let me get you a new binary too.
manic glacierBOT
slender iron
@timber mango The SAMD51 only supports 16MB QSPI. the nina is an nRF52840 right?
@timber mango The GameBoy cart isn't currently available. I'm not sure when it will be. I've been tweaking the software still because it has some major bugs still (like hangs and improper writes)
timber mango
@slender iron jep it is... Note to self, next time order a bunch of different parts ๐
I tested it again, and now i got a stable REPL. Just no flash drive, seems i need to get me less flash.
ionic elk
I'm having an issue with mu where if I write a second time after a program has finished executing, it freezes, and the disk image for circuitpython stops being available until the board is reset.
Is this more likely to be an issue with my code, or with mu?
orchid basinBOT
slender iron
@ionic elk is that with an stm board? try with a known good build of circuitpython to verify mu is working. that sounds like what happens when the board crashes
ionic elk
I think it's a write crash, yeah
It's more destructive with Mu than other editors, so I've switched over to sublime
hmmmmmmmm actually what appears to be the case is the board soft reboots after 30 seconds of runtime, and then can't be written to again without crashing...
slender iron
@ionic elk sounds like it's time for a debugger
ionic elk
I've been in there but I can't get a trace on what's causing the reset
slender iron
is it a full reset or just a reload?
what is the backtrace once it's broken?
good places to break are reset_into_safe_mode and HardFault_Handler
ionic elk
@slender iron "soft reboot". Might be host computer related, though. I'm having some trouble replicating it consistently.
ionic elk
Analogio adds USB pin protection
But I should look into whether it might be messing up some other system essential feature through a pin.
I remembered an important reason to stick to all LL, by the way. The HAL relies on a file called hal_msp which contains a bunch of pin definitions which we'd much rather have flexible control over. LL avoids it completely.
I'm looking into how micropython handles it at the moment.
slender iron
๐
onyx hinge
@slender iron bringing up this "samd51 thing plus", what do I do about usb pid/vid ?
slender iron
use the same as arduino or uf2 and we can bug sparkfun about it
onyx hinge
okay, will do that for now.
I want to talk about the UART console stuff but that's a longer conversation than I want to get into now.
but that's "long term if ever stuff" to me, I'm not worried
manic glacierBOT
onyx hinge
suddenly everyone is interested in this board I guess
I'm not sure which part limor is suggesting santaimpersonator work on -- and whether they are on discord? -- but I think I have a good start locally, but not tested/built yet. I started on the pin definitions last night but didn't get through before it was time to hit the sack.
old smelt
@onyx hinge - out of curiosity, does it require a J-Link to port to a new board initially? I have this vague memory of something like that from when I was looking at putting CP on a Particle Argon.
meager fog
@slender iron @ionic elk QUESTION!
@onyx hinge id like to prod sparkfun into submitting PRs for their boards
its going to be hard for us to test/verify their BSP's without said hardware - and they're shipping with circuitpython so they def made the BSP at some point ๐
onyx hinge
@old smelt if it doesn't ship with UF2 bootloader then yes it seems so
@meager fog OK, better if I deassign that issue from myself then and let santaimpersonator take the lead?
meager fog
@onyx hinge yah sure, i dont think we can assign them cause they're not in the libraries group but thats ok
onyx hinge
๐ I'll go do that
onyx hinge
ooh looks like I can assign it
old smelt
@onyx hinge - got it. Thanks.
timber mango
@slender iron okay thanks for the update ! keep up the great work ๐
cant wait to get my hands on one, i have so many cool ideas that weren't possible with assembly
old smelt
Well, it's Beer:30 here in WI, and I have a bit before Friday Football. Anyone around who has some deep displayio core knowledge that could help me out?
meager fog
i know a little
old smelt
Cool!
I'm trying to chase down this make_transparent/opaque issue, but I don't really understand when/where a display refresh is executed.
I've tried adding the code that I thought would trigger a refresh for the display, but it doesn't seem to do it.
It's this issue: https://github.com/adafruit/circuitpython/issues/2076
So, essentially, the calls to _pallet.make_transparent() in roundrect.py aren't really doing anything.
The original configuration of a shape (transparent/opaque) seems to be "sticky". And can't be changed after initialization.
meager fog
i dont know that kind of detail
if you have an issue, please hold until someone can answer it
i dont know the core displayio stuff ๐
manic glacierBOT
onyx hinge
@slender iron I see how to use version checks for audioio vs audiocore. Do you have a preference as to how audioio vs audiopwmio is handled? Doing it by os.uname().sysname == 'nrf52' doesn't really appeal to me. What would you like me to do?
board.AudioOut did occur to me...
but that's a bigger change and would require touching all boards
slender iron
@onyx hinge maybe try/except is ok for audio vs audiopwmio
it'll be cleaner with the version check for core/audioio
onyx hinge
OK, I tried to do that in my latest update on that PR.
slender iron
@onyx hinge k, great!
onyx hinge
I also realized, I don't know any case where nRF would go through the audiopwmio path successfully
onyx hinge
OH I see how they fit the qwik connector on the thing plus -- it's actually bigger than the standard feather form factor
๐ถ standards are great ๐ต
raven canopy
returns to the living. the flu sucks...especially when you have to keep adulting... ๐
onyx hinge
argh I thought I was going to debug some stuff on pyportal this weekend, but to my horror the swd header is not fitted
onyx hinge
do I need to connect 3.3v to the j-link for it to function?
solar whale
yes -- at least in my experience...
I connect SWDIO, SWDCLK, GND, 3.3V and RST
3.3V goes to Vref on the JLink
onyx hinge
right
that confirms what I thought/read, thanks
too bad there aren't labeled test points for 3.3v and gnd
solar whale
on which board?
Ah --- I see -- on pyportal -- they are on the pads for the SWD connector -- you could access them there
onyx hinge
Connecting to target...Connected to target
Waiting for GDB connection...
``` looks like we may be on the roadhuh not the problem I expected to start out debugging -- with my master-ish-branch firmware loaded, pyportal hardfaults at startup in common_hal_pulseio_pwmout_construct setting up the display early early on
solar whale
I just put alpha-3 on a Pyportal -- so far its OK
onyx hinge
hmm that's exactly where I am as well ```jepler@soba:~/src/circuitpython$ git describe --tags
5.0.0-alpha.3
hm, when I load the firmware from inside gdb with danh's jload it messes up. when I do it via PORTALBOOT, the same binary is fine
smh
fine in that it gets started to circuitpython; I have an incompatibility in my python files or something, ``` File "adafruit_pyportal.py", line 438, in set_background
AttributeError: 'Display' object has no attribute 'refresh_soon'
do you have latest adafruit_pyportal.py -- I am not seeing that
Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 5.0.0-alpha.3 on 2019-09-14; Adafruit PyPortal with samd51j20
>>>
origin https://github.com/adafruit/Adafruit_CircuitPython_PyPortal (fetch)
origin https://github.com/adafruit/Adafruit_CircuitPython_PyPortal (push)
origin/master:adafruit_pyportal.py: board.DISPLAY.refresh_soon()
origin/master:adafruit_pyportal.py: board.DISPLAY.refresh_soon()
``` huh despite pulling I still see the use of those functions in adafruit_pyportal.pyrefresh_soon and wait_for_frame
solar whale
I think there are may places that need update -- I fixed Slideshow a few weeks ago, but more to be done
onyx hinge
what a surprise, I think this todo is why audio stutters while updating the display on pyportal ... ``` // TODO(tannewt): Make refresh displays faster so we don't starve other
// background tasks.
usb_background();
pyportal_simpletest.py fails due to wait_for_frame as well....
onyx hinge
this does fix the audio stuttering right up in my reproducer, but clearly it'll never fly: ``` // TODO(tannewt): Make refresh displays faster so we don't starve other
// background tasks.
+
- #if CIRCUITPY_AUDIOIO || CIRCUITPY_AUDIOBUSIO
- audio_dma_background();
- #endif
usb_background();
(reproducer loops a wave file while continuously printing to the console)
slender iron
Hi all, the weekly community meeting is Monday at 11am Pacific / 2pm Eastern here on Discord in the CircuitPython voice channel. All are encouraged to attend. The notes doc for the meeting is here: https://docs.google.com/document/d/1rWT7iyx39KK_qpDKbT_SnyGplmTR-gaNi_zGQCvT43E/edit?usp=sharing <@&356864093652516868>
idle owl
@ionic elk Could be either, but I would personally suspect code first. Or you found a lovely bug ๐
I've managed things like that with code before is the reason for considering that first.
timber mango
@slender iron you said you're tweaking the software a lot for the game boy cart. does that mean the hardware is mostly final? could i manage to put the board together myself and then flash any updates to the software?
slender iron
@timber mango I believe the hardware is final but since there are bugs I'm not 100% sure I won't need to rev hardware.
Ya, if you assemble one yourself then you could help with the software. ๐
timber mango
time to teach myself electronics ๐
slender iron
are you near seattle? I could loan you one
onyx hinge
I have been doodling a C64 "cartridge port" <-> fpga <-> feather setup. tinyfpga bx has enough signals, so it could read/write any memory via DMA, or drive instructions onto the bus like a ROM. feather could send it instruction lists via SPI or something. I haven't quite figured out how much stuff @slender iron 's board does without needing an FPGA in the mix.
also I'm not sure whether DMA can drive I/O devices. Sure might be some interesting demo effects you could accomplish if you could update VIC registers every cycle.
slender iron
@timber mango I'm hesitant to send them permanently because I only have three to show people
@onyx hinge my board acts like memory and simply returns instructions when the host fetches them
it never controls the memory bus directly
it tells the cpu what to do
timber mango
understandable @slender iron
onyx hinge
@slender iron that's what I thought I understood.
timber mango
what are the two ports on the left of the board ?
slender iron
midi in and out
onyx hinge
@slender iron so I'm also making some doodles about background/idle tasks. Basically, an "earliest deadline first" scheduler, still cooperative, will now let you re-enter other tasks, but never re-enter the same task. It also requires coming up with (largely imaginary or at best hand-wavy) deadlines for all background tasks, and turns them from a block of code into a structure..
@slender iron thoughts before I get too far into implementing it?
slender iron
@split zodiac totally ๐
timber mango
could you technically encode artbitrary data and send it over the midi port?
slender iron
yup, it's just uart with an optoisolator on the input
@onyx hinge sounds like it might be something switching to an rtos could give us. I've never used one myself though
timber mango
i did some PS/2 keyboard to serial stuff with assembly but that sounds more fun
onyx hinge
@slender iron I have some experience in that area with LinuxCNC, though we NIH'd everything.
in the name of having our own portability layer of course
slender iron
of course ๐
timber mango
i really want to make a little operating system for the game boy using this
do you think it would be possible @slender iron ?
onyx hinge
@slender iron are there rtos-like parts of micropython we can integrate instead?
exotic pumice
what speed do y'all run the qspi flash at?
slender iron
@timber mango it depends on what you mean by operating system. the cart basically makes the gameboy buttons, sound and a screen for circuitpython
@onyx hinge not that I know of. I know some folks use micropython from within freertos though (the imx rt PR upstream does)
exotic pumice
I can't figure out where the storage module is actually implemented for samd
arduino defaults the qspi to 4MHz but that seems slooooooow
timber mango
@slender iron just a software operating system with little applications and stuff
slender iron
@timber mango ya, that should be doable!
timber mango
put wifi on the board so it can have an internet browser ๐
slender iron
@exotic pumice for external flash it's done outside the port in supervisor
(not actually real)
timber mango
yea haha but still cool
there are companies that do the full board print and assembly, right?
slender iron
yup! they won't be able to test it though so you may need to fix it anyway
timber mango
yea i can probably give that a go though !
timber mango
that's okay i think it's worth it
do you think it would be possible to have a 32-bit framebuffer that is quantized and blitted onto the game boys vram each frame?
slender iron
you'd have to do the math
timber mango
probably getting ahead of myself but its just so exciting ๐
slender iron
I doubt you can push that much data each frame
timber mango
mm you're probably right and only CGB has VRAM DMA ๐ญ
onyx hinge
if it's "just" running instructions, then that gives you a quick check on how many pixels you can push...
a long enough sequence of STA instructions, f.e.
but I also don't know what the limits are on the samd side
manic glacierBOT
timber mango
ye i havent touched gbz80 in a while so ill have to work out the cycles
it still has to copy everything from the ROM into VRAM right @slender iron ?
@slender iron since you have control over all the cartridge i/o, where does the cartridge rom point to ?
onyx hinge
oh right it's not a 6502 in gameboy
LD A, IMM and LD [HL+],A are each 2 cycles, so you can do about 1MHz / 4 of them per second or 8kB per 30fps frame ?
timber mango
sorry was waking up
25920 cycles which is more than a whole v-blank
@onyx hinge yea that gave me a good idea
just to double buffer and run at 30fps
slender iron
@timber mango there isn't a ROM to speak of because the samd dynamically returns data (ignoring the address almost all of the time)
onyx hinge
well, returning to the M4 dac stuff for an hour, I can reproduce the "stairstep DAC when increasing" behavior under Arduino by setting CCTRL above 0. This is the "current control" and is supposed to control the rise and fall rate of the DAC output, I think.
What arduino configures and what CP configures, neither quite match the datasheet .. argh!
CCTRL = 0 is to be used with clock <= 1.2MHz, but Arduino uses 12MHz.
slender iron
@timber mango technically the cpu thinks it's reading and address but we ignore it. We've issued the code itself so we already know what to return and can ignore the address except to detect the vsync interrupt
onyx hinge
and yes, if I copy those settings -- including the GCLK settings -- the rise/fall match Arduino
I wonder what uses GCLK 4? I can't find it via grep.
timber mango
@slender iron thats so neat ๐
onyx hinge
I'm quitting for the night, but somebody's got to explain me why ports/atmel-samd/peripherals is a submodule and I'll read back tomorrow. (one element of the current best DAC fix I've got is in that submodule, which tricked me)
slender iron
@onyx hinge it's a submodule so that it can be revved/used separately from CircuitPython
manic glacierBOT
onyx hinge
I've had about enough of samd51's DAC.
this is playing the looping sample [0, 57600]
both peaks should be the same
if I go down a bit, say to 56000, they are stable
if I go higher, say to 64000, they are "stable but not what was requested"
solution: when playing audio out, scale everything to (56000/65536) of full scale ๐ข
takes a break
raven canopy
@onyx hinge that seems maddening. thinking back a couple weeks, i wanted to blame the open-mode DFLL as the source. but iirc, Arduino uses it too. at this point, i'd only be throwing darts at the WSJ stock listing...
orchid basinBOT
onyx hinge
@raven canopy Does Arduino actually use the DACs for Audio? I looked around for a library and didn't find it.
raven canopy
proceeds to throw a dart
@onyx hinge have you played around with the DACx.REFRESH register at all?
hehe. concurrent thoughts.
๐ค hmm...well, I assumed it did. i can take a gander.
onyx hinge
I got to the point where directly setting the DATA register in CP gives results that look like Arduino doing AnalogWrite, full scale 0 to 65535
but AnalogOut and AudioOut on CP are pretty different, and AudioOut is still giving weirdness
I actually DID get AudioOut to give the same waveform as Arduino doing AnalogWrite, except that it did it without respecting the actual sample rate
main meteor
Ouch.
raven canopy
haha. "Excuse me silicon-collective, but that doesn't seem correct."
onyx hinge
checks whether samd51 might be violating the code of conduct
raven canopy
@onyx hinge are you using Tone or analogWrite on arduino?
onyx hinge
@radiant wind AnalogWrite
pinMode(A0, OUTPUT);
analogWriteResolution(16);
}
void loop() {
analogWrite(A0, 0);
analogWrite(A0, 65535);
}
k. looks like it uses the DAC to me... https://github.com/adafruit/ArduinoCore-samd/blob/master/cores/arduino/wiring_analog.c#L418
onyx hinge
I agree, but it is not using DMA to send a sample, it's just sending one value after another
raven canopy
true. it has that "luxury".
why must the SAMD sheets contradict itself.. ๐
DACx.REFRESH Feature Description:
tREFRESH = REFRESH ร tOSCULP32K
DACx.REFRESH Register Description:
tREFRESH = REFRESH ร 30ฮผs
and, it looks like we're already using the lowest REFRESH anyway.
0 = no refresh
1 = reserved
asf4_config is set to 2.
onyx hinge
32768Hz is about 30.5us so I'll forgive that much
raven canopy
here is the characteristics at 85C chart. does the drift, plus the RuntimeCal using DFLL (which also drifts), cause any concern? serious question; i have little experience outside the ~2 years i've been here.
onyx hinge
I don't think I've heated anything to 85C while running my tests
onyx hinge
so on top of all my other fixes, rescaling the samples so they are in the range 1000 .. 65000 instead of 0 .. 65535 would fix the last waveform problem
taking some away at the high end and some at the low end gives more total range for some reason
.. I need somebody to look at this and say "yeah that's crazy but let's go ahead and do the workaround" (or "not do the workaround", I don't care...)
orchid basinBOT
waxen loom
I found an interesting observation on the Bluefruit specifically 'touch_A5' and 'touch_A1' are mapped the same on the A5 pin. The A1 pin has no effect. I am using the 'adafruit_circuitplayground.express' library which obviously does not map the same to the Bluefruit. Is there a similar 'express.py' library available for the Bluefruit to date? Thank you all in advance.
onyx hinge
@waxen loom hm, I believe this was fixed by https://github.com/adafruit/circuitpython/pull/2124 but that may not have been in 5.0.alpha.2.
waxen loom
@onyx hinge thank you very much.
manic glacierBOT
waxen loom
@onyx hinge updated firmware however now getting:
File "/lib/adafruit_ble/uart_server.py", line 31, in <module>
ImportError: no module named 'bleio'
from bleio import Attribute, Characteristic, CharacteristicBuffer, Peripheral, Service
onyx hinge
looks like adafruit_ble will also need to be updated
however it looks like the latest release does not have that fix, https://github.com/adafruit/Adafruit_CircuitPython_BLE/releases 24 days ago while the related pull was just merged 16 days ago https://github.com/adafruit/Adafruit_CircuitPython_BLE/pull/18
waxen loom
@onyx hinge thanks will revert back to older firmware in the meantime.
onyx hinge
@tulip sleet consider a release of Adafruit_CircuitPython_BLE in the near future?
onyx hinge
note to self: don't brush the jumper wires that constitute the soldered-on debug interface. Only seems to have rebooted the board, but it was a surprise!
tulip sleet
@onyx hinge did you find the BLE pre-release? Look at all releases.
onyx hinge
@tulip sleet 2.0.1-alpha1 is the top pre-release and I think it's too old based on dates to have the _bleio change. I didn't actually dl and check though.
tulip sleet
I think youโre right. Iโll make another pre-release. I thought I did that but I see not.
tulip sleet
manic glacierBOT
slender iron
@onyx hinge you should be able to merge now
slender iron
anyone want to replicate this? https://forums.adafruit.com/viewtopic.php?f=60&t=156235
timber mango
@slender iron how much memory is there on the circuitpython cart for program, gfx, etc?
marble hornet
@meager fog the stm32 port is so cool! f4 and f7 in the works or just f4? (very excited)
ionic elk
@marble hornet Just the F4 for now, there's a lot to do even within the STM32 families! Scott's mentioned interest in the F7 and NXP's new chips after that.
slender iron
@timber mango I think it's 256k ram total because it's the SAMD51x20 iirc
modern wing
Good afternoon all you wonderful folks โ Iโll be lurking on the call today :)
orchid basinBOT
onyx hinge
oh shewt I forgot it's monday. will be listening only
crimson ferry
Lurking today
errant grail
Text only today. Wrapping up the final two tracks on the album today.
modern wing
Yes
river quest
nice name
we have an album, i should prolly post up that we have one more often - https://www.adafruit.com/product/1688
errant grail
Great album, @river quest . I play it often.
river quest
๐