they want tagged versions so thinks are hermetic
#circuitpython-dev
slender iron
tulip sleet
do they recommend that you tag the action in the .yml?
i think that's what you just said
slender iron
ya
ionic elk
@slender iron Would you like me to store which stm32 pins are 5V tolerant in the pin table?
slender iron
@ionic elk I wouldn't bother. I'd suggest starting with just the info you need
otherwise you are using up flash space for data you'll never use
ionic elk
@slender iron On that note - since STM32 has all the communication devices separately, rather than in sercom units, it'd save some space if I made a new object to store the peripherals and their possible pins separately, and then had macros to whether pins exist in the peripheral object, rather than a bunch of empty slots per pin. I think that'd be more intuitive for how the pins are assigned, and it'd save kbs of space (though I guess we're less worried about that for the STM32s).
Do you think that'd be viable?
Doesn't have to happen now, just looking at all these UART1_CTS, SPI3_MOSI etc scattered everywhere and thinking it'd be nice to just have them in one condensed table.
slender iron
looks at a datasheet
ya, I think it's fine to factor that info out as long as it looks the same from the API perspective
dull kestrel
I'm having an issue with using evals. Do they work in CircuitPython?
tulip sleet
@dull kestrel they should, but we don't use them in general
@ionic elk could you suggest a due date for the new PR. this is an estimate not a sweating deadline
ionic elk
Analogio?
dull kestrel
@tulip sleet I imagine its similar to how they're used in JS. The issue is, in my current setup, I need them I think, but I tried asking over in #help-with-circuitpython and no dice
tulip sleet
@dull kestrel I thought you meant the eval() function, which should just work, as it does in "regular" Python
@ionic elk the pin table task
ionic elk
Probably by tomorrow but I think it'll evolve a lot with analogio and busio
tulip sleet
@dull kestrel I answered in #help-with-circuitpython
ionic elk
Analogio can function with just a per-pin definition, but busio I think is going to need a separate peripheral sorting table.
tulip sleet
yah, that's fine, if you rough it out and do an initial PR we can kibitz on the design, maybe just add analogio and i2c or something like that
pastel panther
hey <@&356864093652516868> does anyone have an issue with adding kwargs for shift and lsb_value to UnaryStruct in Adafruit_CircuitPython_Register?
slender iron
not me ๐
ionic elk
STM32 pin assignments are annoying. They skip lots of peripheral numbers, (like having only UART1, 2 and 6), the same exact pin may or may not have multiple copies across pins (SPI1_MOSI/etc all have two pin options, most others don't), and the peripherals can and frequently do exist on the same pins as others of their category (lots of SPI2/SPI3 pins are shared).
tulip sleet
it sounds even more irregular than SAMD
ionic elk
Bigly
turbid radish
๐
main meteor
I saw one CPU that was fully flexible, any pin (aside from the power pins) could be assigned to any function.
ionic elk
nrfs are like that
tulip sleet
yes, they're either dedicated or any pin (but there's the dreaded "use this pin only for low speed I/O or else you'll add noise to the radio reception"
ionic elk
So anyway I'm thinking the peripherals go in their own table. Otherwise I'm going to either have a bajillion empty slots per pin, or skip options for pin selection that ought to be available.
main meteor
Oh, good point: it would be good to have a compact, but fully generalized method of pin description, but I can't think of one!
tulip sleet
@ionic elk why don't you do enough for analogio pins, and PR that, and then implement AnalogIn. Yes, it sounds like it has to be mapped in reverse from what we did
there is only ADC, not DAC?
ionic elk
Well, probably only for the communication peripherals. The ADC and timers should be ok to go in the regular way
So I can easily implement the ADC pins and timer pins and get Analogio out of the way, and then go back and make a separate comm peripheral table that'll be more space efficient.
tulip sleet
we had a long discussion before about pin numbers vs addresses, and whether to add a level of indirection for that. Is that a little moot due to using LL?
ionic elk
I'm probably going to go ahead and add the indirection. My indecision about that was just some C confusion on where to add the helpers so they apply everywhere I need them.
tulip sleet
i think that would be more straightforward in the long run
ionic elk
But I think a single macro that just does an address shift for these GPIO banks should do the trick in the pins.c
tulip sleet
"by friday", did you mean tomw or next friday
ionic elk
Really the issue is that the STM32 hal doesn't acknowledge "pin numbers" or "pin names" at all, so a lot of the existing stuff you and Scott use has to be rewritten.
tulip sleet
it acknowledges them in the #defines but not in the code?
ionic elk
It has no defines of the "#define PIN_PA4" variety at all.
It uses exclusively GPIO ports and masks
tulip sleet
can you assign a periph function to more than one pin?
ionic elk
So the hal would use "I2C2_GPIO_PORT" and "I2C2_GPIO_PIN" as a pair for all the I2C2 dealings, for instance.
Yes, sometimes, sorta
tulip sleet
wow
and that reflects the hw registers? I should read the datasheet but not before eating
ionic elk
If you're referring to the define pairs, yes, the first is a pointer to the GPIO port location, and the second is a 8 bit mask of that port.
Anyway, I'll probably have analogio in by monday
tulip sleet
i'll add the due dates but you can change them as necessary
idle owl
Never read datasheets hungry.
slender iron
thanks for the merge @gilded cradle ! Now I can release and then bundle
gilded cradle
Of course. Thanks for making the changes.
slender iron
np, it makes the examples clearer
gilded cradle
Yeah
idle owl
@slender iron Would you rather have issues filed on GitHub that turn out not to be issues and you need to close them, or would you rather people mention it here on Discord first and then we tell them to file an issue? Trying to decide how to word an alert in the guide regarding 5.x being currently in alpha.
slender iron
good question. I don't mind 1 but 2 is a good way to direct them to a better place to get help
idle owl
Yeah, kind of where I was at with it too. I'll direct them here I guess and if Limor wants it changed, change it up then. The alert won't be there forever anyway, it's temporary.
slender iron
kk
idle owl
I'll put Discord first, and then link the CP issues anyway
lapis charm
I've seen #1629, saying porting the esp32 is a big task, challenge accepted.
I've start and run into a few compile issues I couldn't figured out
machine_pin.c: In function 'mp_pin_make_new':
machine_pin.c:176:30: error: passing argument 2 of 'mp_arg_check_num' makes pointer from integer without a cast [-Werror=int-conversion]
mp_arg_check_num(n_args, n_kw, 1, MP_OBJ_FUN_ARGS_MAX, true);
^
In file included from machine_pin.c:34:0:
../../py/runtime.h:80:6: note: expected 'mp_map_t * {aka struct _mp_map_t *}' but argument is of type 'size_t {aka unsigned int}'
void mp_arg_check_num(size_t n_args, mp_map_t *kw_args, size_t n_args_min, size_t n_args_max, bool takes_kw);
^
machine_pin.c: In function 'machine_pin_call':
machine_pin.c:200:30: error: passing argument 2 of 'mp_arg_check_num' makes pointer from integer without a cast [-Werror=int-conversion]
mp_arg_check_num(n_args, n_kw, 0, 1, false);
^
In file included from machine_pin.c:34:0:
../../py/runtime.h:80:6: note: expected 'mp_map_t * {aka struct _mp_map_t *}' but argument is of type 'size_t {aka unsigned int}'
void mp_arg_check_num(size_t n_args, mp_map_t *kw_args, size_t n_args_min, size_t n_args_max, bool takes_kw);
^
machine_pin.c: At top level:
machine_pin.c:333:17: error: initialization from incompatible pointer type [-Werror=incompatible-pointer-types]
.make_new = mp_pin_make_new,
^
machine_pin.c:333:17: note: (near initialization for 'machine_pin_type.make_new')
machine_pin.c:414:13: error: initialization from incompatible pointer type [-Werror=incompatible-pointer-types]
.call = machine_pin_irq_call,
^
machine_pin.c:414:13: note: (near initialization for 'machine_pin_irq_type.call')
cc1: all warnings being treated as errors
../../py/mkrules.mk:55: recipe for target 'build/machine_pin.o' failed
I've looked over at the esp8266 port, and this code looks identical around those two functions
lapis charm
yeah now I've seen the message when trying to compile esp8266
"CircuitPython 4.0.0 and later do not support esp8266 boards."
timber mango
Question: What libraries should I be using to test 5.0.0 alpha2? Using the latest bundle on a pyportal I still see a call to refresh_soon(), which fails the initial call to setup the Pyportal (i.e. "pyportal = PyPortal(default_bg=BLACK)" ) SHould I open an issue on this?
exotic pumice
@idle owl re: pointing ppl to discord over issues -- maybe put something in the issue template?
idle owl
@exotic pumice We've avoided issue templates because they often result in issues containing only the template and no new information. We're quick to point people to Discord where it makes sense, so we've made it very visible.
crimson ferry
re: @timber mango 's question... There's probably a plan already, but maybe one issue (perhaps with a checklist like featherwings had) to handle the bunch of libraries, examples, guides, etc. that need their displayio updated to the refresh() model? (Dave, for now I believe you have to download the .py of the library and remove any calls to wait_for_frame() or refresh_soon() ).
timber mango
ok, will do. Thanks
crimson ferry
For 5.0.0-alpha, see: refresh() and auto_refresh here: https://circuitpython.readthedocs.io/en/latest/shared-bindings/displayio/Display.html
lapis charm
@exotic pumice a question, wouldn't be clearer to remove esp32 and esp8266 directories, if it's not supported anymore ?
exotic pumice
Probably, not my call
idle owl
Pro tip: when your code is being ignored, and you know it works, maybe check that you didn't somehow add a space to the beginning of the name code.py. ๐คฆ
@lapis charm Having the directories available makes it easier when merging from upstream. We merge from upstream MicroPython when it makes sense, and having the same directory structure keeps it simpler.
lapis charm
@idle owl ooh, those upstream merges must be painful anyhow, I can see how this can elevate some of the pain.
idle owl
Quite.
lapis charm
that's a shame, I wanted to take circutpython for a ride after hearing more about it in test&code pod cast
but I only have esp32 / esp8266 / pyboard v1
exotic pumice
You could get an ItsyBitsy m4 or something. Pretty cheap
Cheaper than an arduino uno and 1000x the power lol
esp architecture is pretty whack
lapis charm
the esp wer'e cheaper and had wifi built-in. if i'm putting a bit more money, i'll go with https://www.adafruit.com/product/4242
anyhow, that's for the help. have a good day/night
manic glacierBOT
slender iron
@lapis charm we're pretty close to supporting the very basics of the pyboard
@idle owl do we need the debug_bus_device and debug_i2c repos? they are empty
idle owl
@slender iron Yeah, they are still in the works, but aren't a priority so Limor had me waiting on them. They will eventually be a thing.
idle owl
Right on. Thanks!
slender iron
np, addin einks
idle owl
@slender iron Any reason why analogio wouldn't work the same on nRF? The potentiometer example isn't working on Bluefruit. Works on CPX with the same hardware.
idle owl
It's running, but it's acting like nothing is connected. It's returning the same thing as if it's simply floating.
Not sure, will check now
Light sensor appears to work.
Temp sensor uses thermistor?
Hold on.
Yes, light and temp sensors work.
slender iron
hrm, weird
idle owl
Tried on multiple pins.
onyx hinge
I'm not home now but I might be able to try it on a cpb in an hour or so
idle owl
@onyx hinge That would be great. I have two here, I'll try it on the other one.
I might not be around, but please let me know what you find.
onyx hinge
I'll report here
idle owl
https://learn.adafruit.com/adafruit-circuit-playground-express/circuitpython-analog-in is what I'm using.
Same results with both.
tidal kiln
@idle owl seems to be working for me? just testing analog in on a CPB?
Adafruit CircuitPython 5.0.0-alpha.2 on 2019-09-04; Adafruit Circuit Playground Bluefruit with nRF52840
>>> import board
>>> from analogio import AnalogIn
>>> pot = AnalogIn(board.A2)
>>> pot.value
96
>>> pot.value
20096
>>> pot.value
29920
>>> pot.value
44544
>>> pot.value
54112
>>>
@tidal kiln Yeah... I tried on two separate boards. The potentiometer works on a CPX fine...
tidal kiln
weird. do you have an early board rev?
idle owl
Got it in the first batch from the store.
tidal kiln
same here
idle owl
Then unlikely to be a different rev.
Ugh.
The potentiometer is soldered to alligator clips. There's so little to it, why would it work on a CPX if it wasn't going to work on a CPB, and why the same results on two different boards. Does analog work differently on the nRF and for some reason would render this potentiometer useless? I'm confused.
I need to head out for the night.
I'll pick it up again tomorrow. Thanks for testing, @tidal kiln
tidal kiln
ok. np. gotta be something simple.
yah, all seem working for me, tested A1..A6, all good
idle owl
@onyx hinge FYI ^^ worked for @tidal kiln. So it might be a me problem. Please test it anyway and report back what you find. Thanks!
onyx hinge
I'm seeing 0 or a value close to it when I connect A2 to GND, and 60000 or a value close to it when I connect A2 to 3.3V
and seeing intermediate values with a potentiometer
so in that configuration, I connected the outer legs of the pot to GND and 3.3V, and the middle leg to A2
I ran a manually typed in program like cater did fwiw @idle owl
plucky flint
morning
manic glacierBOT
manic glacierBOT
old smelt
Piggybacking on (@crimson ferry and @timber mango)'s thread regarding displayio and 5.0.0-alpha, I am seeing the "wait_for_frame()" error. @crimson ferry , you recommended grabbing the .py library and removing the calls, but isn't displayio a core module? How can I do that? Thanks!
crimson ferry
Those calls appear in some bundle libraries like PyPortal, PyBadger, etc (and if people have grabbed examples from the libraries or Learn guides).
old smelt
I see.
A-HA! Perhaps I should read the stack trace more closely.
Right there in adafruit_pyportal.py. I am back up and running. Ugh. I'll be more diligent next time. ๐ Thanks!
crimson ferry
grep of Library Bundle shows those two calls (wait_for_frame, refresh_soon) used in:
pyportal (drivers and helpers)
miniqr/examples/
turtle (drivers and helpers)
display-text/examples/ (drivers and helpers)
pybadger
pyoa (drivers and helpers)
slideshow
``` issue submittedYes, I commented the wait_for_frame() calls out. Interestingly, it's not choking on the refresh_soon() calls.
crimson ferry
Your code is definitely hitting the refresh_soon() calls? I wonder if it's aliased or something. It's not in the latest readthedocs: https://circuitpython.readthedocs.io/en/latest/shared-bindings/displayio/Display.html
trail pike
Hello. ๐
old smelt
@crimson ferry - Looks like it is not executing those calls. The only call to wait_for_refresh() I had to comment out was in the play_file() function. There's no refresh_soon() call in there.
solar whale
@tulip sleet the reason I posted the stm32 waring issue was that it only appeared for that one BSP -- did that make sense to you?
is that build the only one that uses that file?
tulip sleet
@solar whale makes sense, and I think it's good to fix - tnx!
plucky flint
@trail pike hey hey... fancy seeing you in here. ๐ How're things..?
trail pike
@plucky flint g'day! Things are good. Reading circuitpython.
plucky flint
Nice... are you doing PyWeek?
@trail pike BTW it was the ldnpydojo 10th birthday last night.
trail pike
Are you? (hoping to! Let's see if the freelance gods line up to let me have time off)
plucky flint
We hacked together a bunch of Instagram like filters using pillow and numpy in Jupyter notebooks. ๐
plucky flint
I was trying to get OpenCV to identify eyes in faces so I could automatically stick on googly eyes. ๐
We used our current PM as the test subject in the photo. Which was pretty hilarious.
trail pike
ldnpydojo is something special. Haven't found a group quite like it.
plucky flint
Yeah... we had about an hour of reminiscences before we actually got to coding. Lots of really cool stories, memories and some emotional moments too.
trail pike
Drifting on topic a little bit... is there a doc which answers the question (or part of)... "How is circuitpython tested?"
plucky flint
I'll be doing PyWeek - I want to use it as an excuse to make a game with my goofy hack on PyperCard. I'm thinking of making a Myst clone (originally created in HyperCard FWIW) called "Missed". ๐
@trail pike CircuitPython is a friendly fork of MicroPython. There are docs on micropython.org for devs which contain details of how it's tested.
trail pike
Oh... a myst clone would be grand.
plucky flint
As for CircuitPython specific stuff I'm less clear (at the moment), but the folks here will be able to point you in the right direction
trail pike
Does PyperCard have a zen mode? I really liked the black&white look.
plucky flint
Hah... you should check out the Adafruit original.... https://learn.adafruit.com/circuit-python-your-own-adventure/writing-an-adventure <- very cool hack. If you'd watched me while reading this blog post you'd have seen a lightbulb flashing above my head, just like in cartoons. ๐ ๐ก ๐ค
...and then I got a bit carried away making something similar for desktop and mobile (it's a simple shim around Kivy)
Ideal for beginner programmers.
Anyway...
trail pike
Coolest hack. I remember you mentioning you were going to make it at EP.
plucky flint
Yeah... well I need to get the docs finished for next Friday, I'm trying it out on a class of young coders in an evening class. Let chaos reign. (Actually it'll be really useful to see the bumps in the road, get feedback and check the hunch about such a simple model for UI work is easy to both teach and learn).
trail pike
Did you consider removing some capital letters?
Can't really find any documentation on the release process or how micropython is tested either. But the travis file has a bunch of interesting bits in it. https://github.com/micropython/micropython/blob/master/.travis.yml
plucky flint
Capital letters..?
@trail pike start here..? http://docs.micropython.org/en/latest/develop/index.html <- links to guides and other fun docs.
trail pike
Ah, there's not that many Capital letters in there, because mostly it's done through json and python classes are avoided?
plucky flint
OK... I've lost context. Capital letters where?
trail pike
(because, I think python made the mistake of using Capital letters as is done in Dutch for nouns, so many sentences have lots of Capital letters in there if classes are used)
plucky flint
Ahh
trail pike
But also typing capital letters is harder.
plucky flint
But capitalisation makes reading easier. There's some sort of karma going on there... ๐
slender iron
@trail pike we keep the micropython unit tests going but the hardware apis are only tested by hand
I was just on the Test and Code podcast and we talked about it there a bit: https://testandcode.com/84
trail pike
@slender iron interesting. Thanks ๐
idle owl
@tidal kiln @onyx hinge I'm at a loss here. It worked for both of you. I have two boards exhibiting the same behavior. I eliminated the potentiometer and jumpered A1 to 3.3v. It acts like it's grounded, returning 0.01ish or 500 depending on whether I'm using the helper function or the raw value. If I ground it, no change. If I connect it to VOUT, it returns 2.2V or 44k. I tried it on a different laptop with a different USB cable. I've tried both with code.py and the REPL.
You're both also using board from the same batch as far as I know.
Ok, I just tested it on A2-A6, and it works.
Are you sure it worked on A1 for you?
onyx hinge
I think I tried A2, but let me scroll back....
onyx hinge
you can see from my photo last night that I used A2
I think it's what @tidal kiln used too
idle owl
@tidal kiln Said he tested A1-A6.
(โฏยฐโกยฐ๏ผโฏ๏ธต โปโโป
It's got to be something with A1.
I'll ask Limor.
onyx hinge
is there a pdf or png schematic? which pin is for SDA/SCL?
onyx hinge
okay so it's nothing to do with i2c pull ups
idle owl
The schematic is in the guide but the guide isn't done yet.
onyx hinge
on the schematic I only see the connection to the microcontroller and the pull down resistor
idle owl
I was going to check but figured it was going to be too complicated for me to grok
I'll trust your take on it
Conclusion: it doesn't imply anything is weird
onyx hinge
Yeah
idle owl
So either there's an issue with these boards I have which is possible, or A1 is weird.
onyx hinge
what I did was, open it up in eagle, go to the scmehatic, and in the commandline type "show a1"
all the wires and pads on A1 get highlighted
idle owl
I've never used the command line in Eagle. I should probably look into that.
I usually click "info" and then click the thing I want to see
GUI TOWN.
onyx hinge
like all commandlines it can make SOME things a ton faster
I won't be home until later this evening, but if I haven't updated you with findings about A1 on my device, please ping me about it
plucky flint
OK folks... I'm stepping AFK for the weekend. Thankyou for your help and advice about circup this week. ๐
idle owl
@plucky flint Have a lovely weekend! I'll try to take a look at it soon and give some feedback.
onyx hinge
also confirming from the schematic it looks like the "TX" pin is maybe not digital capable -- it's called "D1/TX" as a net name
idle owl
@onyx hinge Yeah, I noticed the silk doesn't say A7. Need to verify that with Limor and change the guide because I copied most of it from CPX.
plucky flint
@idle owl ๐
ionic elk
Bit of a noobie question here but is using binary constants via gcc frowned upon? Doing a lot of mask work so it sure seems like they'd be nice for visualization.
I've used them occasionally in personal projects where I wasn't worried about portability, but it sounds like they could cause damage in larger projects.
idle owl
@tidal kiln A1 doesn't work right. The pin assignments are wrong. Dan and Bryan both had the same results I did.
@onyx hinge ^^
ionic elk
@tulip sleet that werror hotfix has passed CI by the way
onyx hinge
@ionic elk You mean writing numbers like 0b10110 ? As far as I understand it, these are not a part of the any C standard. However, in practice, CircuitPython is only buildable with gcc. It looks like there are already a few uses of this extension, see PixelBuf.h
@slender iron might have a preference to avoid it though .. comments?
ionic elk
@onyx hinge good to know. I'll check out the PixelBuf. Don't have any current urgent use for it but it'd be nice to know the precedent.
tidal kiln
@idle owl sry. i'm afk for the most part today. got a sec right now though. let me double check it's working for me...
idle owl
@tidal kiln No worries. There's nothing to try. We found the issue. Testing the update right now.
A1 and A5 were assigned to the same thing, A1 wasn't useful for anything as it was.
tidal kiln
ok. yah. i'm seeing that too. A1 is wonky.
tested real quick the other day...must have thought the changes were real.
and this makes sense then...
>>> pot = AnalogIn(board.A5)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: A1 in use
sry for the mis-info. glad it's been found at least.
idle owl
Thanks for checking again.
pastel panther
@slender iron where can I find docs for using a ssd1305/6 oled with 5.0?
tulip sleet
@ionic elk I'll merge #2123 after the latest PR finishes. We are running into the problem that check runs can have their commits disappear out from under them if there's another merge while they're running. Did you see that further back here.
ionic elk
@tulip sleet that's right we ran into that yesterday.
slender iron
@pastel panther https://github.com/adafruit/Adafruit_CircuitPython_DisplayIO_SSD1306/blob/master/examples/displayio_ssd1306_featherwing.py
pastel panther
@slender iron thanks; looks like this particular display needs a different setup as it's lighting up a few random pixels but not displaying text that is printd
Had a similar thing on arduino so I'm pretty sure it's not CP
slender iron
which display is it?
pastel panther
I take it back
it's the large oled bonnet
I think my "hello world" scrolled off the screen when the script ended
slender iron
which display module is it?
slender iron
kk
ssd1305 is the controller ic. usually the display has its own datasheet to describe how it's all hooked up
pastel panther
there are a few rebellious pixels on the right side that I'll have to figure out how to address
oh right, sorry
๐
slender iron
the pixels are probably do to the width being wrong
pastel panther
I have it set to 128x32; do you think it's not working?
slender iron
try setting colstart to 4
pastel panther
that did it ๐
slender iron
the ram of the ssd1305 is 132x64
I wouldn't mind having a separate driver for the 1305 too
it'd make it easier to find
pastel panther
ya, that sounds good
ionic elk
Has external ram for circuitpython ever been considered? There's a couple of STM32 Discovery boards that include it.
tulip sleet
@ionic elk we would certainly consider it for chips that support it. Up to now there haven't been some. It would be for F7 and similar chips.
in a lot of cases those chips have minimal on-chip RAM so it's basically required
ionic elk
An F103 Bluepill with 2MB of ram ๐ค ๐ค
tulip sleet
can you execute from the external RAM?
ionic elk
That I'm not sure.
It just has a FSMC controller
I'm not sure it can extend directly off system RAM. Might be why micropython hasn't done it
tulip sleet
sounds like it would be OK, but I don't know how fast it is
ah, ok, it's serial, so it might be pretty slow to use as regular RAM
Some (not all) STM32's have FSMC mapped to an address space (at 0x60000000).
good for displays without their own ram
ionic elk
^yes that seems to be what it's primarily used for
I wonder, though, if there'd be some way to have it link into the core, given that CPy has to store everything in RAM
tulip sleet
_Note two serious problems with running from external SRAM:
It's going to be slow. Like 6...8 times slower than running from flash or even internal SRAM according to this. This microcontroller architecture simply isn't optimized for this kind of use.
Some STM32 chips have a silicon bug where FSMC access by two bus masters crashes the chip. This will basically preclude you from using DMA. This concerns STM32F103 revision Y IIRC, probably other chips too._
it is an interesting q about how much slower it would be to use as a destination for importing code, since a lot goes on after reading a bytecode
pastel panther
@tulip sleet I'm guessing I'm conflating two things, but doesn't samd51 support execution in place from qspi?
tulip sleet
yes, it's "XIP". We don't execute code from RAM very much if at all, though MicroPython has ways of including assembly code, so it's possible. But here I'm talking about import an .mpy into the FSMC memory, where it would be read as bytecodes to execute.
pastel panther
<-- attempting to take that as a learning opportunity rather than "I has the dumb".
I think I understand, mostly
thanks
@slender iron will you have time later to give me the high level of what would be needed for a displayio ssd1305 driver?
slender iron
@pastel panther I'm headed out early for a party so the next hour or so is all I have
basically it's copy and rename ssd1306
pastel panther
@slender iron that's easy enough. I was thinking some more init tweaking needed to be done but I can figure it out
slender iron
ya, I usually get the init sequence for a datasheet flow chart or sample code
remember that things like colstart are not a property of the IC but rather how the IC is hooked to the display
pastel panther
ah, interesting. I did not know that.
I shall consult the 'duino driver and datasheet later tonight. thanks!
stuck elbow
@pastel panther the only thing you have to do is to reset, switch off, enable charge pump, and switch on
All the rest are just fine tuning and tweaks
pastel panther
@stuck elbow fair enough, but in my case it's the "fine tuning tweaks" that are the issue ๐
stuck elbow
Why?
pastel panther
stuck elbow
Is that a reflection?
pastel panther
no
stuck elbow
I didn't know those displays had grayscale
pastel panther
I'm not sure that's what's happening
when I saw something similar with the arduino driver, it was smearing "down"
stuck elbow
Too high contrast?
pastel panther
possibly, I'll have to dig more later. thanks
meager fog
@tidal kiln hihi
onyx hinge
@idle owl do you need any more info from me about the CPB analog in, or did you solve it?
idle owl
@onyx hinge We solved it! The board def was wrong. A1 and A5 were on the same thing.
tidal kiln
@meager fog ping
tidal kiln
@idle owl yeah! good test / good catch
meager fog
@tidal kiln ah i assigned it on basecamp
this person fixed the neotrelis but didnt do a PR
tidal kiln
ok. seeing it in basecamp. is there a related github issue?
meager fog
is not!
tidal kiln
where's the fix? not seeing any link?
crimson ferry
In one of the SSD1675 examples, epd_reset (D5) and epd_busy (D6) are used, but not in the other. But aren't those EPD pins really for SRAM and SD on the FeatherWings?
meager fog
@ionic elk ihii
i just got a discovery '412 ๐
tell me what to do and ill test! do i upload with st-link?
exciting
ionic elk
You actually won't need st-link at all
it's mbed compatible so you just drag it on in there
meager fog
the bin file from s3?
ionic elk
Is s3 a release? behind on lingo
ionic elk
I think that should work. There's a chance that it might have been broken by Dan's HID update, I didn't have the chance to fully test that out. But it should work with the defaults.
meager fog
ok so then what ๐
ionic elk
Sorry, double checking things, it's been a little while since I got to work on it.
Ok. For now, compile the board using USB=FALSE. Then, you can use the built-in UART available on the CDC line to the st-link.
You should be able to view that with a standard terminal. I'd just ignore the "user USB" until Thach and I get a chance to hack at that again.
ionic elk
Sure.
meager fog
๐
ionic elk
OK, here you go. Just tested it, should be good. Baud is 115200
Note: there's a buffer bug that deletes every 10th character when copy-pasting into the repl, and can even crash in extreme cases. It won't bother you if you just type. Wasn't a very high priority fix since the whole system gets replaced by USB later anyway.
@meager fog just let me know if you have any issues and I'll check it out.
meager fog
@ionic elk ok burning!
when i drag over the bin it doesnt reboot
is that normal?
looks like red, orange, green leds are lit
how do i get to the repl
thru the stlink USB?
ionic elk
Yes, there should be a CDC line with the UART interface
Not sure what terminal you use - it's tty.usbmodem1411203 for me
ionic elk
HOORAY
meager fog
are the 'duinoy headers like D12 in board?
meager fog
ok great, whats keeping USB from working?
ionic elk
We're not sure - I was running into kind of obscure register differences between the F412+ and the ones under it.
meager fog
ok so to be specific, F412 does not work for USB at all even with tinyUSB
ionic elk
yes
meager fog
F405 and F411 do work with tinyUSB
ionic elk
I haven't gotten the F405 BSP in yet but it shouldn't be a problem, and the F411 is done, yes
meager fog
for f411, you have the circuitpy build and USB enumerates and you get USB repl?
ionic elk
Thach needed to make a new BSP for the F411 before that one worked as well.
F411 has full CDC and MSC flash support - it'd have HID too, but the F411 doesn't have enough endpoints for it
meager fog
ahh ok
fascinaing you cant program the F411 firmware in and have it 'just work'
odd but ok
is there an issue for the F412 tinyusb failure?
ionic elk
Yeah it's a pain! As are... many things with ST
I'm not sure we've made one I could throw one up there. But I think Thach was planning on wrapping that up pretty soon.
ionic elk
Will do.
meager fog
https://github.com/hathach/tinyusb/issues/103 looks like the existing issue
?
please add a note it doesnt work at this time ๐
ionic elk
Where would you like the note? Github, comments, documentation?
meager fog
@ionic elk now that i htink of it, it should be an issue in circuitpython and link to that tinyusb
so you get a notice when thach closes it
and then you can re-try
ook good news i have a low cost stm32f411 dev board
manic glacierBOT
meager fog
theres no bootloader for the stm32f411 i have, can i prog the bin with stlink? what address does the bin go to, 0x0?
@ionic elk ^ if ya know lemme know
ionic elk
@meager fog yeah the F411 doesn't have mbed so it needs a tool. I use https://github.com/texane/stlink, but ST also has a flash utility tool https://www.st.com/en/development-tools/stsw-link004.html
meager fog
@ionic elk all good we stock 'em
i use them a bunch ๐
btw there's adalink if youhaven't seen it yet
makes it ez
ionic elk
Gotcha - to be clear the board still has an stlink on it, you just gotta use the command line since it doesn't have the mbed boot filesystem. Do you recommend most users to the official ST tool?
meager fog
st tool works great - i think its win only?
ionic elk
yeah
meager fog
yah adalink is python so great for automation
ill test the stm32f4 PR
good excuse ๐
ionic elk
I actually JUST found a new bug on it so it might be better to wait until tomorrow.
Somethings claiming pins incorrectly. Only happens for some of them though so I missed it at first.
Gotta head out for the night but I'll be back at it tomorrow.
meager fog
@ionic elk ill just try it for the sake of tryin
pastel panther
@gilded cradle do you know if having two Displays works? like one on SPI and one on I2C?
gilded cradle
I'm assuming you mean in CP and I'm not sure. I haven't really touched the I2C displays, but theoretically yes since you can do it in arduino.
pastel panther
ok, thanks
onyx hinge
I gotta learn how to upload a fresh .elf file from within gdb, resetting my session because I have to exit gdb, stop the j-link gdb server, perform the flash, and then put everything back together .. is no good. Anyone have a recipe for commandline / terminal gdb? I'll read back after making breakfast, this is just a drive-by question.
onyx hinge
you mean I have to research it for myself ๐
maybe it's just "load"? I'll try it next session.
exotic pumice
yeah it's just load
onyx hinge
nRF I2S.CONFIG.MCLK looks kinda-sorta like it's an integer clock divider .. but .. it isn't.
16MHz * (0x80000000 / 0x50000000) isn't 10.666667MHz (it's 10MHz), but programming 0x50000000 gives you 10.66667MHz anyway!
smh and taking a break
crimson ferry
Would it be a reasonable issue to add to imageload for support of True color BMPs? Or is there a better way than doing everything within a with open("/truecolor.bmp", "rb") as bitmap_file: construct? (I need it in memory and I don't think we have deepcopy)
onyx hinge
hm, I have my world clock and bit clock clocking like I would expect for a 44.1kHz, 8-bit stereo signal
but the supposed data out is not synchronized to them. It's just a square wave at close to 1MHz
bit clock, left/right clock, but not i2c data
also, when I cleared out the PSEL for that pin, the scope didn't change, so .. where's this 1MHz signal coming from? Not I2S I guess.
okay, hm, still not sure, but a reset fixed it. now it looks like we have data and it's stable in relation to the clocks.
looks like maybe there's a bug where a soft reset can leave pins associated with PWM instances
and I guess maybe PWM takes precedence over I2S in whatever the output pin switch matrix does
alright, I'm just "DMA'ing" the same single sample over and over again, but when I change it via the debugger it changes on the scope so the next step of playing an actual sample should be pretty quick.
waxen loom
2nd of a larger series of very simple apps working with the BLE to help spur development on the Bluefruit. https://github.com/mytechnotalent/CircuitPython-BLEThermometer/tree/master
simple pulsar
FYI I made the requested change https://github.com/adafruit/Adafruit_CircuitPython_PyPortal/pull/39
craggy galleon
I have a CircuitPython / REPL question:
I am trying to read back the console through the USB COM port (COM2). This is not an issue. I just use pyserial on Windows and it reads back.
The bit I am struggling with is sending commands to the console. I think my issue is that I am sending encoded data to the console and it doesn't know what to do with them. Does anyone know how I can send text to the REPL / Console for the CircuitPython board over a serial-like connection?
Windows:
import serial
test = serial.Serial("COM2", baudrate=9600)
while True:
test.write("\n".encode('utf-8'))
status = test.read()
print(status)
CircuitPython:
while True:
data = input("Data: ")
print(data)
I cracked it! The REPL expects /n/r as the return character!
raven canopy
@craggy galleon you can also use the "raw REPL" that's designed for machine-to-machine. example of usage: https://github.com/adafruit/circuitpython/blob/master/tools/cpboard.py
orchid basinBOT
onyx hinge
ugh that's not a sine wave! ๐
onyx hinge
okay, THAT is a nice pure tone
๐
there is a little bit of scratchiness on top of it that I wish wasn't there, but I don't know what to expect of this i2s amp. I should test it on a samd part.
onyx hinge
well that's the first time I lost progress with git in quite some time. My computer hard-locked very shortly after doing a WIP commit ๐ฆ
error: object file .git/objects/7e/b3549cd8b21c9429dcc1e4a33bfc1b1707023a is empty
error: object file .git/objects/7e/b3549cd8b21c9429dcc1e4a33bfc1b1707023a is empty
fatal: loose object 7eb3549cd8b21c9429dcc1e4a33bfc1b1707023a (stored in .git/objects/7e/b3549cd8b21c9429dcc1e4a33bfc1b1707023a) is corrupt
git log, git reflog, etc all fail
gentle bronze
@ionic elk stm32f412 discovery board is ported https://github.com/hathach/tinyusb/pull/120 It has more endpionts than 407,411 on Fullspeed core. Therefore it could work with cpython hid ๐
sorry for the wait, being a bit busy lately ๐
forgot to mention, you need to put jumper to JP3 (USB) on the board, it is left open by default.
onyx hinge
sigh, why am I stuttering when playing 22050Hz samples. Surely my code's not that inefficient.
main meteor
SD card delays?
onyx hinge
it's the QSPI (I assume) internal to the particle xenon
I would have also assumed it was plenty fast enough at this task
idle owl
<@&356864093652516868> Here is the notes doc for tomorrow's CircuitPython Weekly meeting at 2pmET/11amPT. Everyone is welcome! Please add your hug reports and status updates before the meeting, even if you'll be attending - it's super helpful! If you're missing the meeting and you'd like to include your updates, add them to the notes and we'll read them off. Hope to see you there! https://docs.google.com/document/d/12p4adKbmLf-2QIu2vdvnRw98MfD1jh-mLwq-OXcXxRY/edit?usp=sharing
simple pulsar
@onyx hinge re: "computer hard-locked" - are your hard disks/ssds healthy?
onyx hinge
@simple pulsar well, it seems to be mostly correlated with circuitpython hardfaulting and dmesg implicates USB
simple pulsar
Oh, unfortunate, is this on linux?
onyx hinge
"Sep 8 12:40:58 soba kernel: [80072.030802] RIP: 0010:usb_clear_halt+0x16/0xc0 [usbcore]"
yeah
simple pulsar
I'm thinking out loud here but it feels like the driver should be robust for any usb badness going on. Have you got the CIRCUITPY drive mounted?
onyx hinge
sometimes, and/or I'm using the serial console
simple pulsar
I've nothing useful to add here. I only asked about SSD/etc because I was surprised last year when my first one failed after reaching EOL. Not surprised that it failed just surprised by Windows (on a desktop) doing a wide range of bizarre things with no user notification of impending doom or first signs of doom. I thought in 2018 that might have been basic functionality.
onyx hinge
instead of actually characterizing the problem I just complain on chat ๐
smartctl data all looks OK
I do intend to replace the whole system later this year so if it is the HW so be it
and I appreciate you trying to make me be systematic about a problem I reported ๐
hm would an attached j-link affect performance if no breakpoints are being hit? I can't reproduce my choppy audio anymore ๐
I might have had a hardware watchpoint and certainly some breakpoints set "before"
onyx hinge
how likely is it that doing "a lot" of read activity from CP (I2SOut) would "starve" the device so badly it couldn't USB-enumerate?
I'm seeing that when my code.py starts playing such a file, linux just can't enumerate it
just using a different (e.g., 22kHz mono instead of 44kHz stereo) file "fixes" this
tulip sleet
@onyx hinge we've encountered this kind of problem before: https://github.com/adafruit/circuitpython/pull/901 but this was on atmel-samd before tinyusb. nevertheless the usb irq prio may be an issue. I remember adjusting this myself as well, I think in a different situation.
onyx hinge
Run till exit from #0 audioio_wavefile_get_buffer (self=0x20008280,
single_channel=false, channel=<optimized out>, buffer=0x20008210,
buffer_length=buffer_length@entry=0x2003ff74)
at ../../shared-module/audiocore/WaveFile.c:210
0x0005e4a8 in audiosample_get_buffer (sample_obj=<optimized out>,
single_channel=single_channel@entry=false, channel=channel@entry=0 '\000',
buffer=buffer@entry=0x20008210, buffer_length=buffer_length@entry=0x2003ff74)
at ../../shared-module/audiocore/__init__.c:110
110 return audioio_wavefile_get_buffer(file, single_channel, channel, buffer, buffer_length);
Value returned is $8 = GET_BUFFER_MORE_DATA
(gdb) p *buffer_length
$9 = 0
``` hm now I am having a problem with audiosample_get_buffer returning 0 bytes at a time.f_read from the wavefile returns 0 for some reason
oh it's another corrupted wav file
cp: error writing '/media/jepler/CIRCUITPY/stereo44.wav': No space left on device
I think my problems were bad wave files mostly. What a pile of wasted time
onyx hinge
I bet for other audio outputs this was not a "fatal" error. Mine calls over and over again until a desired number of bytes are made available.
shy kiln
I'm considering buying an Arduino Nano 33 BLE (based on nRF52840 with a 9-axis IMU) - I want to experiment with sending IMU data via BLE. I would like to develop using circuitpython, but on circuitpython.org I see that the Nano 33 BLE is not a supported board. Is anyone working on providing support for it?
onyx hinge
@shy kiln good question -- I'm not familiar with the module. I don't see any signs that support for this module has been started. I suspect that someone familiar with circuitpython and uf2 bootloaders would be able to create support for the board, but there could always be unanticipated problems.
In the "good" case it would be less than 300 lines of code to write
based on the amount of code added in CircuitPython for the Particle Xenon
hmmm I think all the CircuitPython boards may have QSPI flash modules, reserving the full 1MB for CircuitPython code. checking..
tulip sleet
@shy kiln @onyx hinge all our boards do have flash. We have a build for PCA10059 which does not have external flash, so that could be adapted to the new board. Could also buy our board and a 9-DOF breakout
onyx hinge
the accelerator is supported https://learn.adafruit.com/adafruit-lsm9ds1-accelerometer-plus-gyro-plus-magnetometer-9-dof-breakout/python-circuitpython
@tulip sleet I was just checking that. What defines the amount of storage reserved for CIRCUITPYTHON in the PCA10059 board?
ah is it LD_FILE = boards/nrf52840_1M_256k.ld ?
tulip sleet
yes, I was just going to paste that path
that .ld file is specified in the mpconfigboard.mk file
onyx hinge
ugh somebody needs to make a board revision to move that silkscreen label CSAG. ๐
tulip sleet
I think the CSAG is lower deliberabtely to differentiate from the adjacent labels
or am i missing a joke?
onyx hinge
I'm just being a tiny bit OCD, I think it would be above the pin (like the other even pin labels) but that interfered with the pads of some resistor so it ended up moved where it was and offset
so yeah it's a non-issue
shy kiln
@onyx hinge @tulip sleet Thanks for your comments. I can see that using Feather nRF52840 Express plus 9-DOF breakout would be much simpler. On the other hand a smaller size/mass of hardware is advantageous to me, and the nano is already smaller than a feather even without adding the 9-DOF. I'll have a think about this.
tulip sleet
@shy kiln it's not too hard to make your own build, and you could submit a PR to us for the board (or that class of boards)
shy kiln
@tulip sleet How would I submit a PR?
exotic pumice
yw
shy kiln
@exotic pumice Thanks. I was thinking PR meant "product request" ๐
exotic pumice
lol
onyx hinge
'night all! It was a productive weekend.
manic glacierBOT
manic glacierBOT
cerulean sparrow
@slender iron , thanks for working with me with getting the PR.
slender iron
@cerulean sparrow happy to! I appreciate you sticking with me on it
cerulean sparrow
@slender iron No problem
plucky flint
@slender iron @idle owl @tulip sleet I'm going to have to miss this evening's meeting (I was hoping I'd be able to make it). My son has just informed me he needs taking to Tae Kwon Do. ๐
My Hug report is for @tulip sleet for his really helpful ideas last week while I was working on circup.
(in addition to a general TellyTubby like group hug)
๐
idle owl
@plucky flint Thanks for letting us know. We'll add your hugs to the notes!
plucky flint
Thank you! ๐
pearl notch
Hi. Whatโs the algorithm for finding the link to the current dayโs meeting? (Scroll up hasnโt always worked for me.) Iโd settle for a link to todayโs meeting.
plucky flint
-> dad taxi
idle owl
<@&356864093652516868> Here is the notes doc for today's CircuitPython Weekly. Everyone is welcome! Please add your hug reports and status updates before the meeting, even if you'll be attending - it's super helpful! If you're missing the meeting and you'd like to include your updates, add them to the notes and we'll read them off. https://docs.google.com/document/d/12p4adKbmLf-2QIu2vdvnRw98MfD1jh-mLwq-OXcXxRY/edit?usp=sharing
errant grail
In the studio today tracking and overdubbing, so I'll be lurking. Group hug!
onyx hinge
"(in addition to a general TellyTubby like group hug)" I may or may not return to this image whenever anybody says "group hug" from now on
status: trying to get some sort of lunch together before the meeting starts
idle owl
Meeting starting soon; we're finishing up another meeting.
sterile bronze
just lurking today
onyx hinge
[fx: pots and pans tumble uncontrolled from the counter]
timber mango
lurking
meager fog
hi folks, text only just say'in hi ๐
gilded cradle
hi
pastel panther
@pearl notch The link used to be a pinned post but it looks like we've stopped doing it. For now you can search for posts that tag <@&356864093652516868>
wups
slender iron
๐
tidal kiln
lurking
pastel panther
@pearl notch https://docs.google.com/document/d/12p4adKbmLf-2QIu2vdvnRw98MfD1jh-mLwq-OXcXxRY/edit?usp=sharing
river quest
The Top Programming Languages 2019 - Python tops the charts with a CircuitPython nod!
https://spectrum.ieee.org/computing/software/the-top-programming-languages-2019
CircuitPython 5.0.0 Alpha 2 Released!
https://blog.adafruit.com/2019/09/04/circuitpython-5-0-0-alpha-2-released/
Machine Learning with Microcontrollers Hack Chat, lots of Python on Pi! 3pm ET Weds!
https://hackaday.com/2019/09/09/machine-learning-with-microcontrollers-hack-chat/
Speaking of' Hackaday turned 15, tune in to the story and more
https://youtu.be/kMkgwSLj0Uc
meager fog
@ionic elk heya im just...not able to get this workin'
river quest
Feather STM32F405 โ How did we do?
https://blog.adafruit.com/2019/09/04/feather-stm32f405-how-did-we-do-st_world-stm32f405-stm32-stm-adafruit/
wiringPi, wiringPi GPIO library โ deprecatedโฆ Lessons for the open-source community
https://blog.adafruit.com/2019/09/02/wiringpi-wiringpi-gpio-library-deprecated-lessons-for-the-open-source-community-drogon/
meager fog
@ionic elk whats weird is i totally get all the USB appearing when i use thach's tinyusb bin
ionic elk
Hmm maybe it's a submodule update issue?
meager fog
@ionic elk did you try the binary you posted on your board?
this definitely works:
(the tinyusb test bin)
ionic elk
Yep, it's the same one that's working on my board right now
meager fog
and you're positive its the '412 discoboard not the '411?
ionic elk
I just downloaded the link I posted off of github and loaded it into the mbed, and I'm getting both the Cpy drive and CDC successfully
meager fog
:/
ionic elk
Could you send a photo of your board setup?
I mean you're getting it working with tinyusb so that shouldn't be it
ionic elk
Oh, could you try it on the mac? Maybe it's somehow related to case sensitive stuff, somehow
ionic elk
HMMM
I... honestly don't know what's up with that. Why would it affect a binary? @slender iron do you have any ideas?
slender iron
it enumerates on mac but not win10?
meager fog
yeah
the tinyusb demo build works fine on win10 - get cdc + msd
i could try rebooting my compy
i dont think that will affect it but ๐คท
pastel panther
@slender iron links plz
tulip sleet
i'll try this on my '412 board on Linux and another windows comp a little later
onyx hinge
https://github.com/microsoft/pyright @pastel panther
pastel panther
thanks @stuck elbow ๐
slender iron
@meager fog @ionic elk I'd start by looking at a beagle trace for it
ionic elk
I'll check that out after the meeting. otherwise it'll mess up my headphones -___-
slender iron
๐ @inland tusk
meager fog
@ionic elk do you have a windows computer to test against?
ionic elk
I do I'll need to set it up
slender iron
@pearl notch I can't edit the spreadsheet to add myself
meager fog
ok thats a good idea - i can give you a trace later if ya need
its got promise! at least it works on mac
you'll figure it out ๐
pearl notch
Sorry, I just expanded the access to the https://docs.google.com/spreadsheets/d/1NWwiUuQnam_abdv02ugLxz9uFrvhtMfy8U306HJ4_qU/edit?usp=sharing
ionic elk
@meager fog can confirm does not work on my win10 machine
I'll run a trace once the meeting is over
pearl notch
Amazing! The directory appears to be filling out itself. ๐
onyx hinge
@pearl notch it is impressive to watch
pastel panther
pastel panther
@pearl notch Looks like that was just an image if it was supposed to be a link to a video
pearl notch
onyx hinge
I would love to talk FPGAs too
slender iron
@old smelt have you looked at migen at all? I'm hoping to get those folks building datasheets for shared SoC files that are easy to load
old smelt
Yes - a little bit.
slender iron
once we have that, we can port circuitpython to it like any other mcu
old smelt
Great interview on The Amp Hour podcast with Tim Ansell, Sean Cross and Michael Ossmann.
old smelt
We've been historically Intel-focused, but I am intrigued by the open source activities
meager fog
@ionic elk good to hear you had the same non-enum! that means its not just me ๐
old smelt
...mostly on Lattice.
pearl notch
Canoe: awesome!
pastel panther
@solar whale whitewater? Touring? All of the above?
ionic elk
@meager fog AHA. I was restricting the usb endpoints to only CDC and MSC, since that's required on the F411. When I turned off that setting (enabling HID) I got full enumeration.
It seems like windows has some kind of required relationship between the CDC and HID, which isn't required on mac
This may be obvious to those more experienced with USB?
prime flower
@solar whale thx for looking into the RFM issue, im curious about your future findings
ionic elk
I will update the PR to represent the change but it does mean that the F411 will continue to not work on windows... I wonder if there's a way around it?
tulip sleet
@ionic elk @meager fog there is a bug in the "turning off USB devices" PR, which I found yesterday and will submit a fix for soon.
the interfaces get numbered wrong
tulip sleet
it's related to the interface numbers in the descriptors, not the hid report ids
ionic elk
As in, are the excluded endpoints still pushing up the numbering, or something? I think I remember something like that
Windows doesn't like the endpoints being numbered out of order, or something.
tulip sleet
yes, that was the mistake. There's a function that renumbers the interfcaes after they are concatenated, but I was renumbering over the whole possible list, not the selected sublist
ionic elk
Welp that's that then.
tulip sleet
i'll submit the PR and you can try it on the stm build (or I can)
stuck elbow
gotta run
ionic elk
@tulip sleet it's not that big a deal since the F412 should have the full descriptor list anyway, since it can fit it
I've pushed a commit that removes the restriction.
But it will be important for the F411
manic glacierBOT
ionic elk
@meager fog in the short term here's a build that should work on windows.
onyx hinge
meager fog
@ionic elk tryn !
solar whale
Have a great week all -- have to go offline for a bit ๐
meager fog
@ionic elk @tulip sleet odd that import os doesnt work, maybe check you aren't missing something?
can import board ๐
ionic elk
I don't think that's supported yet
tulip sleet
because no filesystem support
meager fog
o ok
@ionic elk @tulip sleet lets get you using adafruit VID/PIDs for these boards - dan, plz assign him some
old smelt
Thanks all. Have a great week!
onyx hinge
Thank you for running the meeting, @idle owl
meager fog
that way mu will detect it - and also we can have STM drivers on win7
slender iron
๐
ionic elk
@meager fog probs the best starting point is just the digitalio example
errant grail
Thanks!
meager fog
yah i will just like to do os.uname() ๐
ionic elk
we can add it to the list!
meager fog
yes
this is great so far
excellent work
next up will be the pin restructuring and analogin?
ionic elk
Could use some help ordering that, BTW. I've got QSPI Flash, F405 support, UART, SPI, and OS
Analogin is almost done, that should be in today
meager fog
@slender iron @tulip sleet wanna chat about whats next?
QSPI flash feels like the next best thing, that way we get consistant storage
ionic elk
But the communication devices will need new pin stuff. Possibly a new object, since they're handled very differently than nrf and SAMD
tulip sleet
do you want minimal F405 soon?
ionic elk
F405 stuff is one of those things that could be really easy or really hard depending on whether something is unexpectedly wrong
meager fog
i will need that...but dont know if it makes more sense to get structure in place
so adding any boards is simpler
ionic elk
like some weird register change, like we had with the f412
I think some level of schenanigans risk is inevitable with STM32. Might just be a good idea if we had a place to document stuff like that
tulip sleet
you can do external flash as SPI, not QSPI, first, if that makes it easier in some way, but since there are examples already QSPI is probably fine
meager fog
well we'd need SPI then ๐
maybe thats a good next one, SPI usually is not a bonkers peripheral
ionic elk
which requires the com peripheral restructuring...
meager fog
also the 'f405 doesnt have QSPI
ionic elk
Yeah that's why I was kinda pushing for the F412
Or one of the other ones
F413 has 10 UARTS
just sayin
meager fog
i def have to make a F405 feather
ionic elk
Sounds good. Do we want to chat briefly about the comm peripherals object?
meager fog
can make a 412 feather after
tulip sleet
did you want a unified object? I don't think you really need one. nrf has separate also (unlike samd)
slender iron
ya, I'm not sure what you mean by comm peripherals object
tulip sleet
well, SPI and I2C peripherals overlap, but we're allocating them statically right now
ionic elk
I just wanted an object for storing what pins are associated with the UART, SPI, I2C, etc peripherals that's separate from the pin objects,
since each peripheral has a specific pins for each function, and they're pretty inflexible
Including it with the pin array would mean a ton of extra space for basically no benefit
meager fog
i'll note that for better/worse the peripheral setup for stm is not as-flexible as sercom, so the table of what you can use is not as bonkers
like there's only 2 or 3 spots for each peripheral
slender iron
@ionic elk I would just have separate tables for each type
meager fog
on smaller chips, only 1 or 2
tulip sleet
is there any sharing of periphs? If not, then you can just have a table per periph type
slender iron
@onyx hinge @old smelt https://twitter.com/tannewt/status/1171142678962008064
Twitter
tulip sleet
revamp I2C to use periph table
meager fog
(i gotta go to my 3:30)
ok #1) table per chip, #2) make sure i2c still works #3) add SPI
ionic elk
@tulip sleet yes, periph table, test with i2c, add spi, add spi flash, add qspi flash, add 405
hows that
slender iron
I'd do 405 before qspi
ionic elk
ok
meager fog
yah good idea!
tulip sleet
you can add all these to-dos in basecampe and we can reorder as necc
meager fog
you can always use QSPI flash as SPI ๐
ionic elk
How about documenting all this stuff? I've been hoping to do that at some point
Setting up a learn guide for the structure of putting a port together
If I wait too long I won't remember :\
meager fog
yeah but i dont want ya to have to document things after you do something and then it gets refactored heavily
how about after SPI, then document the peripheral table
because you had gotten i2c/spi to work
ionic elk
This would be more of a guide on how to get the barebones of a port up, then the steps you need to take to flesh it out
But I think after SPI would be a good point to do it
meager fog
ok
slender iron
I don't think it's worth doing a guide unless we revamp it first
it's not something we do thaaaaaaat often anyway
tulip sleet
taking notes on what was hard or should be revamped is good. You could make a guide outline and just take notes. Don't need to make a draft guide yet
ionic elk
Sure, but it might be nice for explaining how ports are structured
I'm just hoping to save others some confusion - there's lots in the early port process that's pretty easily explained and could save a lot of time. But whatever structure you feel is appropriate for that I'd be happy with.
Could just be on the readthedocs for now, maybe?
tulip sleet
i think taking notes about it even in a Google doc is fine, so you don't forget. I wouldn't spend time formatting it all up now
we can share the google doc to other porters for now
onyx hinge
"gateware" just means the configuration image for the FPGA, "gate" because FPGAs are made of gates?
(I was not familiar with the term)
manic glacierBOT
idle owl
@slender iron Ok, circuitpython-build-tools released.
slender iron
thanks @idle owl !
meager fog
@ionic elk @tulip sleet @slender iron ok back - i am having photo take pix of the '412 discover
manic glacierBOT
ionic elk
@meager fog were you still having issues even with the new binary?
meager fog
@ionic elk no! it works now
huzzah
i was in meetin's
is this dep on danh's PR
or is the PR compelte now
ionic elk
super, I was worried there we had a bug beyond what dan described!
meager fog
all good - i have lotz o meetings that slice my day up
but i did try it now and it enumerate
i was going to try to blink some LEDs
ionic elk
The PR is done. Having all the USB endpoints sidesteps the bug and should be done anyway
meager fog
if you have an example py that will save me a few mins
ionic elk
We'll just need dan's fix for the F411, which can't use all the endpoints
meager fog
ok i have this adorable F411 board we should support after the discovery
ionic elk
import board
import digitalio
import time
led = digitalio.DigitalInOut(board.LED1)
led.direction = digitalio.Direction.OUTPUT
while True:
led.value = True
time.sleep(0.5)
led.value = False
time.sleep(0.5)
Oh great what's it called?
meager fog
ok please order 2 of these https://www.elecrow.com/micropython-development-board-pyb-nano-compatible-with-python.html
@slender iron lotta flight controllers use the stm32f411
@ionic elk ok get also 2 of https://www.aliexpress.com/item/4000068998794.html be sure to pick the '411 option
when you get shipping, use DHL/FEDEX/UPS not postal, cause it will take 3 weeks
ionic elk
k
meager fog
@ionic elk did dan give you a VID/PID?
meager fog
PID 0x8055 # arduino
PID 0x8056 # circuitpython```here ya go
you can use the circuitpython for this PR
manic glacierBOT
meager fog
@ionic elk i cant seem to save files to flash
thats normal?
also, cant re-use pins on reset, they aren't deinited?
import digitalio
import time
leds = []
for pin in (board.LED1, board.LED2, board.LED3, board.LED4):
led = digitalio.DigitalInOut(pin)
led.direction = digitalio.Direction.OUTPUT
leds.append(led)
while True:
for led in leds:
led.value = False
time.sleep(0.1)
led.value = True
i was out but just got back
ionic elk
I think there might be an issue where it isn't triggering self reset properly
can you try hitting the reset button and see if it starts working again?
meager fog
yeah
i am doing that
ok just verifying
thats a different PR
wanted to verify it wasnt me
that example blinks 3 of the LEDs, LED1 seems to not blink but who knows why
oh wait i forgot LED1
duh one moment
manic glacierBOT
tulip sleet
@ionic elk do you want an F412 .bin with restricted USB or an F411 binary as is, for testing?
ionic elk
restricted F12 bin
meager fog
@ionic elk ok once you update the VID/PID im happy to call the USB for f412 working well enough!
pretty lights ๐
ionic elk
I'll hang on to it, thanks!
ionic elk
I do need to put in a little bitty PR where I standardize LED names
meager fog
i think its ok as is, ez to blink ๐
good test of time/digitalio
good work! we're getting there
tulip sleet
@ionic elk ok, I think this is right. It has my interface numbering fix, and it's your branch rolled back to before the last commit, which removes the F412 USB device list narrowing
tulip sleet
@ionic elk hierophect I took my '412 out of the packaging. Can I save the factory firmware load easily before overwriting it, or is it readily available elsewhere?
ionic elk
you mean the factory example? I think it's pretty readily available on the board page but I'm honestly not sure. I've never used it
That's the one with the display, right? Let me take a look
Yep, compiled demo binary: https://www.st.com/en/evaluation-tools/32f412gdiscovery.html#resource
tulip sleet
@ionic elk Yes, well, I went ahead and overwrote it. ๐ I am powering the board with both USB cables, and it enumerats fine on Windows 10 and Linux with USB_DEVICES = "CDC,MSC". I can see the insides of CIRCUITPY and connect via REPL. Yay!
tulip sleet
I'll merge my PR after this finishes building so we don't get GitHub actions skew conflicts
ionic elk
Great. Could you take a look at my old pin remap one too?
slender iron
@ionic elk I think @meager fog meant that pins couldn't be used after a CircuitPython reload. It's super common bug to have early in development. basically all peripherals should be reset after python code is done
meager fog
@slender iron @ionic elk indeed, i can open an issue
meager fog
issue created
slender iron
thanks @meager fog
manic glacierBOT
Overall nice work! I tested it using the UDA1443 and the nrF52840 FeatherWing with the sine sample: https://learn.adafruit.com/adafruit-i2s-stereo-decoder-uda1334a/circuitpython-wiring-test
I did the same test to compare with the Feather M4 and it did way worse. I could hear the tone but it wasn't as loud or as clear as the nRF. So hopefully you don'...
onyx hinge
@slender iron I'm surprised at volume/clarity differences, i2s being a digital protocol
slender iron
@onyx hinge yup, seems like a bug to me on samd51 side
exotic pumice
have you checked samd51 errata?
onyx hinge
I'm having rather rotten luck today with using "load" in gdb
exotic pumice
are you using openocd?
onyx hinge
it keeps putting my device (particle xenon) in a state where I have to "make sd; make flash" otherwise it just reboots over and over again
no, all segger j-link tooling
exotic pumice
ok good
try like monitor halt, then load, then reset or something
it might be trying to load while it's running
exotic pumice
yeah try a halt first
onyx hinge
I don't have a theory why it could end up so confused. I swear this same workflow worked most of the day saturday.
particularly how I could have to "make sd"
tulip sleet
@onyx hinge it would be worth getting a real j-link, I think
onyx hinge
@tulip sleet I do have the "j-link base" now, not the cheapest one
non-edu and everything
onyx hinge
np
tulip sleet
I do what you are doing all the time. I have a set of aliases
but I usually use a feather 52840
tulip sleet
the board or the computer?
onyx hinge
The whole computer. Recently installed Linux box, but the HW is all 10 years old
tulip sleet
is this your startup for gdb server? JLinkGDBServer -device nRF52840_xxAA -if SWD
and you're up-to-date on the segger software?
I am using a 7010 Dell Optiplex. Both USB2 and USB3 ports work fine with this setup
onyx hinge
drwxr-xr-x 8 root root 4096 Aug 31 08:21 JLink_V650a
-if swd -speed 4000 -device nrf52832_XXAA -port 2331
``` well this is wrongit will be interesting to see if the device change does anything. Are you using an offboard USB hub?
onyx hinge
Yes, a powered hub
tulip sleet
you might try a direct connection, just to make sure
but it is not generally a problem, it's usually a particular hub that's flaky
onyx hinge
regardless, it's a good evening of progress!
I am sure that I will find a routine that works for me with the j-link and/or maybe it's my PC which I plan to replace in October/November anyway
tulip sleet
and it may be the board too, so order one or more Feather 52840's when they're available
onyx hinge
doesn't seem likely, but not impossible either I guess
tulip sleet
do you have a UF2 bootloader on the board for convenience?
onyx hinge
Yeah
though j-link "load" in gdb would be even more convenient if it worked ๐
consistently
tulip sleet
did it lock up again after changing the device?
I'm not sure the code is any different in the jlink programs
onyx hinge
My whole PC hasn't locked up again yet, no.
haven't done "load" enough times to know if the problem is there or not, sometimes it works 2-3 times and then stops
I've had a whole raft of slightly different problems and I am not sure I've even got a handle on what the pattern is, just that it doesn't work 100% and I'd like it to ๐
tulip sleet
yah, I really don't see that I do use my aliases for loading, so I always do a mon reset before and after
onyx hinge
having the aliases would help me be more consistent that's for sure
but still, what can "load" possibly be doing that erases the sd?
(softdevice, right?)
tulip sleet
right.
well, maybe there are some subtle differences between the '832 and '840 about writing flash
manic glacierBOT