@ionic elk ok, jumping back in
#circuitpython-dev
slender iron
idle owl
@tulip sleet So Adafruit CircuitPython Crickit uses the C-level seesaw firmware? Do I understand that correctly? Considering filing an issue on seesaw for quadrature support as per https://github.com/adafruit/Adafruit_CircuitPython_Crickit/issues/13 but I want to make sure I'm in the right place.
slender iron
@ionic elk nice work! they both worked
ionic elk
π
slender iron
the h7 should be faster than the f7 right?
onyx hinge
@idle owl The C++ code in question lives at https://github.com/adafruit/seesaw
But it also depends on which pins can be chosen. They have to be ones that have a TC, but on seesaw it looks like those are already dedicated to an internal function (the DC motor drive)
slender iron
ok, h7 is faster. 76ms vs 179 for an empty for loop 20k long
idle owl
@onyx hinge Hmm. I'll not file an issue then since I don't know what I'm talking about.
When's the last time we pulled from upstream MicroPython? What version?
onyx hinge
@idle owl I appreciate you are trying to help a community member with their idea!
@idle owl our docs say "This release is based on MicroPython 1.9.4 @25ae98f" (5.2.0)
er the release note says
idle owl
Still ok. Thanks.
slender iron
hrm, with this test the f405 beats the f767
onyx hinge
@slender iron just to mention, you can remove these two caveats from the next release notes:
ulab can give erroneous results when sorting 2D arrays
ulab can give erroneous results with expressions like a[a<1]
slender iron
import time
print("hello world!")
start = time.monotonic_ns()
for i in range(20000):
pass
print(time.monotonic_ns() - start)
@onyx hinge kk
manic glacierBOT
The F76/7 devices have 128k DTCM and 16k ITCM. Please add a TODO to add it. Totally ok to follow up with it.
ionic elk
@slender iron by the way, I was hoping to ask for your opinion on a hobby project sometime, if you've got a minute
slender iron
@ionic elk I have time now
simple pulsar
Is pulseio healthy in 5.x on nrf? I am seeing some difficult to explain occasional variations when I use it to repeatedly charge a capacitor. I'd normally suspect a loose connection but I'm not nudging it at all...
crimson ferry
The STM32F405 isn't in the support matrix?
raven canopy
@crimson ferry its in work... https://github.com/adafruit/circuitpython/pull/2598
crimson ferry
Thanks!
slender iron
any idea what it's waiting for?
raven canopy
the proposed changes to the CIRCUITPY_----_BUILD mechanics.
to summarize the why not now? the builder doesn't recognize MINIMAL build.
ionic elk
@crimson ferry I'm working on that tomorrow actually
slender iron
π
ionic elk
I need to put in new build flags and stuff
slender iron
k, cool
ionic elk
but it's my next up alongside a reworked readme
slender iron
π
tulip sleet
@idle owl you can file an issue even if we can't do it; it would be aspirational
ionic elk
@slender iron the hobby project I wanted to ask you about was my fork of UncannyEyes - I'm trying to figure out what the best platform to port it to would be
would it be better if I asked that in help with arduino or some other room though?
raven canopy
@ionic elk are you going to adjust the doc builder part too (if needed)? not calling dibs, just wondering if i'll need to put it on my docket...
ionic elk
yes
ionic elk
I just mean in terms of Hardware. I don't know atmel too well, so my biggest question is how hard it is to get SPI DMA running on Arduino vs the raw atmel libraries
slender iron
we have pretty thin wrappers for it in circuitpython
ionic elk
I'd like to actually have the whole thing be Circuitpython but it'd be too slow unless I made it a module
tulip sleet
@slender iron I was out of the house and thought I got a message from you about not merging something (I didn't have my glasses), but now I can't find it. Did you send something?
slender iron
@tulip sleet check your dm
simple pulsar
Ignore my earlier comment, it's unlikely to be pulseio, I think it's probably just something like usb activity adding inter statement delays
simple pulsar
I'm finding myself in need of reading lots of samples at a set rate or highest rate of the ADC on the nRF52840. This came up in the discussion (on SAMD21) on https://forums.adafruit.com/viewtopic.php?f=60&t=142006 - has there been any work in this area?
thorny jay
Hi @onyx hinge I have the following adafruit board https://www.adafruit.com/product/3211 and a 64x64 matrix that worked fine (with the recommended hardware modification) on a Raspberry Pi 3b. Do you think I should be able to drive that matrix with your Protomatter/RGBMatrix code? Did you have a chance to work with a 64x64 matrix (they are "special", I believe they use one more pin/GPIO)? Will you write a learn guide to explain the steps (the electrical powering of those matrix is the most frightening for me, and that's why that Adafruit Bonnet and instruction really helped getting me started). Thanks.
onyx hinge
@thorny jay I haven't used the 64x64 matrix. The code should support the 5 address lines A through E, but the existing featherwing for feather m4 express would need to be modified to actually bring the signal to the header.
Would love to get your feedback if you try it
pastel panther
heya @onyx hinge is it possible to play independent waveforms using the DACs on a an M4's A0 and A1 pins? When I try I get an error about the DAC already being in use
raven canopy
@pastel panther looks like a no, unfortunately.
https://github.com/adafruit/circuitpython/blob/master/ports/atmel-samd/common-hal/audioio/AudioOut.c#L128
pastel panther
@raven canopy wub wub
I think it does support stereo so maybe I can hack that to set each channel with separate data
raven canopy
audiomixer, though? doesn't support balance (yet), but you could channel the source waves...
raven canopy
π
onyx hinge
yes, it does support stereo
pastel panther
@onyx hinge is the restriction to make the memory handling easier or something?
onyx hinge
@pastel panther I don't really know, the samd audio was largely by Scott, before I came on board.
pastel panther
ah, ok thankx
onyx hinge
#copoutanswers
raven canopy
doesn't make it the wrong answer. π
ionic elk
@slender iron any more testing you want to see from the H7/F7 port or should I go ahead and merge?
slender iron
@ionic elk merge away. I was just waiting for CI
ionic elk
@slender iron it seems that Dan's review didn't get dismissed by the update, did we change a setting? I'm no longer able to merge
slender iron
π
manic glacierBOT
ionic elk
@slender iron Thanks! Working on docs now. If you have any direction for what you'd like me to start looking at in the i.mx port after that though let me know - I'll start brushing up on the datasheets/libraries today and early next week.
slender iron
@ionic elk https://github.com/adafruit/circuitpython/issues?q=is%3Aopen+is%3Aissue+label%3Amimxrt10xx
slender iron
@tulip sleet any updates from github about the stalls?
slender iron
@gentle bronze I'm updating tinyusb in circuitpython now
tulip sleet
@slender iron nobanswer to ticket. Official forum responder said to try cancel and rerun
manic glacierBOT
thorny jay
@lone axle and @tulip sleet Your suggestion to look at ble_uart_echo_* and Bluefruit Connect is perfect. I took the Nordic UARTService code and just replaced the 3 UUID (main, RX and TX) by those in use by my device. And now I can send command and get response. π
slender iron
esp32s2 branch: https://github.com/tannewt/circuitpython/tree/esp32s2
onyx hinge
TIL that this error doesn't stop a github actions/checkout@v2 in its tracks: ##[error]fatal: unable to access 'https://github.com/v923z/micropython-ulab/': Failed to connect to github.com port 443: Operation timed out ##[error]fatal: clone of 'https://github.com/v923z/micropython-ulab' into submodule path '/Users/runner/runners/2.169.0/work/circuitpython/circuitpython/extmod/ulab' failed
marble hornet
@slender iron think long living classes could be turned off for a specific board?
tulip sleet
@marble hornet The mechanism could be overriden in the functions that take a boolean saying whether an object is long-lived or not. But what would you gain? (π€ soon so we can discuss later)
marble hornet
i'll ping you tomorrow then. sleep well
gentle bronze
@gentle bronze I'm updating tinyusb in circuitpython now
@slender iron kk, tag me when you create PR for update. I will double check it π
orchid basinBOT
lone sandalBOT
ionic elk
What are practical options for databases on Circuitpython? Is btree it?
lucid solar
Someone here knows how the serial console works? It outputs fine over the USB-CDC...but I'm looking at getting it to output on the VCP 'debug' serial on a STM32 nucleo board.
lucid solar
Update: I've fixed it π
marble hornet
@meager fog are you familiar with what might cause raytac nrf52840 's bluetooth to stop working? did this ever happen to you when developing the boards?
manic glacierBOT
lucid solar
@ionic elk We should talk sometime...I've ported C.Py to nucleo_f746g which works fine...and have now got it working on disco_f746zg. Thing is...there is a lot of changes that would break existing ports, so we need to move things either more generic, or less specific depending on what it is.
crimson ferry
Would it make any sense to have a #help-with-micropython channel, or do we want to steer people toward MicroPython.org resources? Some folks come here for help, typically in a general channel or the CircuitPython channel. There are many Adafruit Learn guides for MicroPython on Adafruit hardware. Adafruit Forum is combined CircuitPython and MicroPython, though almost all posts these days are CircuitPython. I don't have a strong preference, just trying to get a general sense of direction.
timber mango
We should just point people to the Micropython forums.
tulip sleet
@crimson ferry our MicroPython guides are mostly historical. We don't try to do extensive micropython support. We can answer simple MicroPython q's, but their forums are better for regular users. Often the q is some kind of comparison of CPy vs MPy.
crimson ferry
Thanks, will do, kind of what I figured.
tulip sleet
@slender iron could I ask you a PacketBuffer q, or is this not a good time?
slender iron
@tulip sleet I have 10 mins
tulip sleet
i'm trying to use a PacketBuffer on the peripheral side as a server, to queue writes to a fixed size characteristic (i.e., not notify, just straight writes). This may be the first time this is being doen, though I thoguht it was done for HID or MIDI.
common_hal_bleio_packet_buffer_get_packet_size() returns 0 if there's no connection, which is not what I'd expect (I'd expect max_length), and the code subtracts the att_overhead from the returned packet size, which is also not what I'd expect.
ok, I see, I guess it tries to return the min(mtu, max_length), and then adjusts for the overhead. I thought it would just return the max_length if < mtu
slender iron
how would it write when it's the server?
tulip sleet
no, the client writes to the characteristic (it's the tone service for adafruit services, so it's sending a tone to play)
i.e. the bluefruit playground app writes to the char
tulip sleet
i was htinking the packet size was the max size of the written data, but it doesn't seem to be that
in this case, the tone is a fixed lenght, 6 bytes
slender iron
notifications are limited by the mtu
tulip sleet
yah, but I needed the packet size before any connection was set up. I mean, I can get it in another way (by the struct format string), but i'm misunderstanding what packet_size means, I think
I thought it meant the max size of the data of a single packet, which is usually way smaller than mtu
slender iron
I think it's fine to return max length
it should be whichever is longer since packet length is used to validate the buffer passed in
tulip sleet
i was using it to allocate the buf initially, but if it's mtu-size, that's unnecessarily huge
not the PacketBuffer buf, the writeinto() buffer
slender iron
when sending it should check against the smallest
slender iron
ya
tulip sleet
so maybe I'll PR a fix for common_hal_bleio_packet_buffer_get_packet_size(), since it usually is going to return mtu size rather than actual char size (if char size is smaller).
this is all trying to debug some buffer overflow
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "ble_adafruit_circuitplayground_bluefruit.py", line 68, in <module>
File "adafruit_ble_adafruit/tone_service.py", line 81, in tone
ValueError: Buffer too short by 5376 bytes
obviously some length is wrong here. It's taking some data values as the length, I think
ok, tnx, I'll proceed with the debugging. I wasn't completely misunderstanding, which is what I was worried about
slender iron
Β―_(γ)_/Β―
tulip sleet
good night!
tulip sleet
@onyx hinge I'm getting a compilation error from the tip of master in ulab, which is odd, because it should work:
In file included from ../../py/obj.h:32,
from ../../extmod/ulab/code/poly.c:12:
../../extmod/ulab/code/poly.c: In function 'poly_polyfit':
../../py/misc.h:89:43: error: 'y' may be used uninitialized in this function [-Werror=maybe-uninitialized]
89 | #define m_del(type, ptr, num) ((void)(num), m_free(ptr))
| ^
../../extmod/ulab/code/poly.c:88:26: note: 'y' was declared here
88 | mp_float_t *x, *XT, *y, *prod;
| ^
../../extmod/ulab/code/poly.c:127:46: error: 'x' may be used uninitialized in this function [-Werror=maybe-uninitialized]
127 | XT[i+j*leny] = XT[i+(j-1)*leny]*x[i];
| ^
cc1: all warnings being treated as errors
this is in nrf
Submodule path '../../extmod/ulab': checked out 'a746bd8e0953853056ee405e2fa02c8ebca4fb79'
/home/halbert/bin/gcc-arm-none-eabi-9-2019-q4-major/bin/arm-none-eabi-gcc
hmm, this only happens if I compiled with DEBUG
I'll make an issue
tulip sleet
@slender iron I found some issues with using ringbuf.h for PacketBuffer, can discuss tomw
i wonder if this is causing other issues, if you are using it for MIDI, etc.
slender iron
@slender iron think long living classes could be turned off for a specific board?
@marble hornet I don't think it should be turned off. If there is a bug, then we should fix it.
timber mango
@slender iron I agree. Fix bugs when they are found. Do not turn features off because of a bug.
marble hornet
I have not seen any bugs. Besides the in-list Id issue. I would want to disable it for just one board, the GUI imports and destroys classes often and over. Do you think it would be advantageous to disable long living for that board?
lucid solar
STM32f746g_Disco: Working β
tulip sleet
@marble hornet why are you destroying imported classes? Is it because they are just used for constructing something and then aren't needed after that?
marble hornet
@tulip sleet I am destroying them when I get rid of the whole module
tulip sleet
@marble hornet why would the user program want to unimport a module? I know you've been trying this for a while, but what is the use case?
marble hornet
The user would t. My βsystemβ would. The watch can switch between apps. Like open one then go home then open a different. The apps loaded in memory can change. Each app is a class in a module. To prevent memory errors, when memory is low the oldest imported module is deleted. (Using the above sys.modules and careful handling of anything that has a reference to it )
Does that answer your question?
tulip sleet
yes, I remember that. This is difficult enough to do in regular Python. Does a strategy of restarting the whole thing after remembering what do to after restart not work out? Would it help if there was a non-flash way to remember state between VM instantiations? I imagine the only strong reason for this is lack of RAM.
marble hornet
yes that is for this scope. inter-VMinstance would probably help and "boot-up" time of the system is > 1 second. although freezing in some of the source code will speed up "boot-up" time. via importing it is much faster
it is working right now except for the couple hundred bytes. i just realized i did the x y problem switch.
tulip sleet
@marble hornet did you make .mpy files of the source to speed up loading?
marble hornet
not yet i've still been editing a lot of the files, i'll give it a shot to estimate.
when reporting the issue I should have said , i think theres an error when garbage collecting/ unimporting a module. after removing/garbage collecting all the instances and references to a module and its content a few hundred bytes still stick around.
tulip sleet
using .mpy will make the loading a lot faster, and will reduce some garbage
marble hornet
a touch faster
tulip sleet
@slender iron there were two issues with using ringbuf.h:
- Due to the way it does buffer-full checking, it only stores
n-1values when allocated as sizen. The calling code was not aware of this and was trying to fill up the buffer with exactlynbytes. - When the buffer was written to when full, it started overwriting the oldest bytes. This is not so terrible for UART and CharacteristicBuffer, but it played havoc with PacketBuffer, because the length bytes would get overwritten.
I am doing a PR that will clean this up, and will just drop overflow bytes instead of overwriting the oldest bytes. I don't think this is really an issue for UART, it's still terrible anyway, but we can discuss.
ionic elk
@lucid solar regarding your mention of RCC_OscInitStruct.HSEState, do you have a board in mind that does not use a high speed external oscillator?
lucid solar
Well, no...but I'm about to comment something else on the PR π
lucid solar
@ionic elk no external SRAM on F412_disco....but I've seen some on other boards...I don't have any right now yet.
ionic elk
You'd have to check with @slender iron about where that'd be in terms of priority. It's probably not something I'll be able to revisit for a while but I'm sure he'd be open to PRs.
lucid solar
@ionic elk peripherals/stm32f7/f767xx/periph.h
Any reason the mcu_tim_banks is not extern, and the mcu_tim_pin_list is not defined there?
manic glacierBOT
idle owl
CircuitPython Weekly Meeting in ~40 minutes. Everyone is encouraged to attend! Please add your status updates and hug reports to the notes doc even if you'll be attending - it's super helpful! Thanks and hope to talk to you soon! <@&356864093652516868> https://docs.google.com/document/d/1YIp-LKPdIzsN2LN4yRo4FzxzTTnLuFeDOJPboQKDVhc/edit
turbid radish
Folks, if you have any CircuitPython/MicroPython/Python projects, news, etc. for the CircuitPython Newsletter, please send to me at anneb@adafruit.com as soon as possible as I'm finalizing things this afternoon. Thanks!
modern wing
Good afternoon all you wonderful coders -- happily lurking today 
tidal kiln
@idle owl i won't be able to make meeting today
idle owl
@tidal kiln Thanks for letting me know. Hope all is well.
errant grail
Happy, healthy, and lurking today. Hope everyone is doing well!
serene warren
Lurking, but not leering π
onyx hinge
If you'll be lurking, add it to the notes or let me know in here. Got you, Mr. Certainly, cater, cgrover, duewester
timber mango
Text only
onyx hinge
@lucid solar will you walk to talk during hug reports or status updates, or are you just listening (lurking) today?
thorny jay
@idle owl Not going to the meeting either, and did not provide notes . I have a few hug report, but I can provide them next week. And I cannot really talk about my current CP related project.
lucid solar
I can just lurk....My mic doesn't work at the moment anyway π
idle owl
@thorny jay Thanks for the update. No worries. Hope to see you next week.
@lucid solar You can type your hug reports and status updates in this channel as well if you would like to participate without a mic.
serene warren
Yes I can'
timber mango
You can also listen in by joining the voice channel even if you are not going to talk.
serene warren
Hello hello back at you
lucid solar
Scott + Water = wet keyboard
serene warren
Hello @idle owl
sterile bronze
lurking
gilded cradle
I have one of these in case water spills: https://www.amazon.com/gp/product/B074DTL449
errant grail
In anticipation of global warming issues, I'll be wearing my stillsuit.
serene warren
When I was on subs in the Navy, if you came even near a control panel with liquid they knock you away.
tiny oriole
Man.... double the volume slider seems like its 4x as loud lol
timber mango
lurking today
onyx hinge
@timber mango noted, thanks
turbid radish
lurking
serene warren
Thanks Charles
slender iron
I was just admiring the new leaves on the trees across the street. π³
stuck elbow
lurking, sorry
raven canopy
i blame the Discord client crashing! π
slender iron
π
old smelt
Lurking (and late)
raven canopy
i was merely the messenger. π
slender iron
π€
tiny oriole
heh, im over here switchless
slender iron
w00t! nice job @lucid solar
onyx hinge
@lucid solar an impressive debut
lucid solar
Thanks 
onyx hinge
slender iron
@lucid solar I added you to the circuitpythonistas role so you get notified about meeting stuff
@raven canopy it might have been the pin I picked
raven canopy
if it is pin related, we can handle that better than just pushing out zeroes. π
onyx hinge
slender iron
modern wing
Animal Crossing? Enjoying being in debt? π
modern wing
I see you're digging up these puns π
raven canopy
how are we not further discussing Jeff's mention of the RNG LEEK?
slender iron
π
onyx hinge
ionic elk
onyx hinge
[I turned on a software audio monitor of my mic and the lag makes me have a ton of trouble talking. I'm going to turn it off right after this]
ionic elk
@slender iron whoops posted it just to you yeah I put it on show and tell last week
onyx hinge
cute as promised
slender iron
@onyx hinge that's why I like the blue yeti. it pipes audio back with no lag I can tell
modern wing
Which mic are you using @onyx hinge ?
onyx hinge
@modern wing today I'm using the built in mic on some gamer headphones, HyperX brand
serene warren
Great day. Be Safe to all.
onyx hinge
@slender iron I'm sure considering some upgrades to my conferencing equipment maybe we can talk some more about what you recommend.
gilded cradle
Thanks everyone
modern wing
Thank you everyone, and stay awesome π
onyx hinge
Heading AFK, thanks all!
solar whale
Have a good week, all -- the dog is pacing -- need to go for a walk π
modern wing
@onyx hinge There's a sale at B&H right now for some AT gear: https://www.bhphotovideo.com/c/buy/audio-technica-large-diaphragm-usb-mics/ci/46964
I use the AT2020usb (not plus), and I'm very happy with it. No need for a pre-amp or cloudlifter.
I remember opening a serial session via Putty, with a brief panic when I got no output. Turns out that I didn't even plug in the cable to my laptop nor the device....it was sitting in my bag.
It happens to all of us at one point π
Animal Crossing turnips are literally doing better than real-world oil right now. π’οΈ
slender iron
it's M4 and the same clock speed
ionic elk
slender iron
I grabbed both the discovery and nucleo from digikey π
ionic elk
want me to?
slender iron
sure π
crimson ferry
This is expected?
Adafruit CircuitPython 5.2.0 on 2020-04-09; Adafruit PyPortal with samd51j20
>>> lines = 'line 1\nline2'
>>> lines.splitlines()
['line 1', 'line2']
but
Adafruit CircuitPython 5.2.0 on 2020-04-09; Feather STM32F405 Express with STM32F405RG
>>> lines = 'line 1\nline2'
>>> lines.splitlines()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'str' object has no attribute 'splitlines'
i would guess some of the MICROPY_ flags haven't been tweaked in the STM port/boards.
crimson ferry
Some string methods are in one but not the other. And vice versa.
Should all ports support identical language features? If so, I can file an issue, but I'm not sure how to find all of the differences beyond string exhaustively.
raven canopy
@crimson ferry the STMs are still a little young I think. and we have various things turned off at the MICROPY_ level even on M0 (Gemma, Trinket). not sure it requires an issue, as it will likely be worked out as the STM port matures. then again, an issue never hurts! #NonAnswerAnswersAreHelpful
lapis hemlock
@slender iron Scott, do you have two minutes?
slender iron
chatting with someone else now
lapis hemlock
OK.
slender iron
I can do a quick text answer
crimson ferry
@raven canopy Maybe I'll just give it some time. rtc is another thing., not currently supported.
lapis hemlock
https://github.com/micropython/micropython/commit/30840ebc9925bb8ef025dbc2d5982b1bfeb75f1b brings in warnings, which mess up ulab. What would I have to do to shut it up?
Warnings like these: https://github.com/v923z/micropython-ulab/issues/102
These are actually errors, that's the snag.
lapis hemlock
@slender iron Scott, I think, I have found the cause of the error.
@onyx hinge Jeff, https://github.com/micropython/micropython/commit/f31f9a8b70db03cbcbcf39b493f959d0e284962a changed the treatment of floats in micropython, and that messed up ulab. I have implemented a fix in https://github.com/v923z/micropython-ulab/pull/104, and the question is, whether it is still OK in circuitpython, or we have to patch it up.
onyx hinge
@lapis hemlock I didn't actually check if it builds but we have both mp_obj_new_int_from_float and MICROPY_FLOAT_C_FUN so it shouldn't be too far off
I'm not likely to have time to look into it in detail this week
lapis hemlock
I'm not likely to have time to look into it in detail this week
@onyx hinge It compiles on github now. Would it be OK, if I merged it as it is, and we can modify it later, if you run into difficulties incircuitpython.
onyx hinge
that is fine by me
lapis hemlock
OK, I'll do that. Thanks for the feedback!
You can ignore the review request, that was accidental.
onyx hinge
your changes won't stop circuitpython from building, because we have to make a pull request to update the ulab submodule first, and we'll see any problem then.
okay, I do appreciate you keeping me in the loop
lapis hemlock
Thanks for spotting the error in polyfit! I could now pull in your modifications.
manic glacierBOT
lucid solar
This is expected?
Adafruit CircuitPython 5.2.0 on 2020-04-09; Adafruit PyPortal with samd51j20 >>> lines = 'line 1\nline2' >>> lines.splitlines() ['line 1', 'line2']but
Adafruit CircuitPython 5.2.0 on 2020-04-09; Feather STM32F405 Express with STM32F405RG >>> lines = 'line 1\nline2' >>> lines.splitlines() Traceback (most recent call last): File "<stdin>", line 1, in <module> AttributeError: 'str' object has no attribute 'splitlines'
@crimson ferry https://circuitpython.readthedocs.io/en/5.0.x/docs/library/index.html#omitted-functions-in-the-string-library guess this is why....but we should be able to enable it.
slender iron
@lucid solar ya, looks like a build config issue
timber mango
The one thing I do not like about my ItsyBitsy M4 is the reset button. It is almost level with the board and is difficult to press, especially if I need to "double click" it.
crimson ferry
@lucid solar Thanks, appreciate the link π
ionic elk
@lucid solar @crimson ferry @slender iron It sounds like this is part of what I'm already working on, which is the adjustment of build flags for the docs. STM32 has some redundant build flags to avoid FULL_BUILD that should all basically just get removed
with minor exceptions which is what I've been working on
crimson ferry
@ionic elk Thanks, is rtc one of the things that will pop into the rebuild or does that need development?
slender iron
rtc will need some love after my lower_power branch is merged in
since it starts using the RTC for time keeping
timber mango
Is there a new bootloader for some boards? I seem to remember something about this.
crimson ferry
For M4s, yes, they're on the circuitpython.org download pages
timber mango
OK, got it, thank you.
obsidian compass
Is there any blog/tutorial on how to use raspberry pi as SWD programmer? I am especially looking to re-program particle xenon board to be used with CircuitPython but don't have a SWD programmer, I read rpi can do SWD, so is it possible to flash bootloader on xenon using rpi?
tidal garden
Hey guys, newb here. Anyone know of the best method for reading/editing a .mpy file?
crimson ferry
@tidal garden .mpy files are bytecode, you'll want to find the .py version. If it's a CircuitPython library from the bundle, they can be found on Adafruit's Github https://github.com/adafruit
manic glacierBOT
crimson ferry
My commit seems to have broken Github (and these very tall print statements from black are going to take some getting used to)
idle owl
@tidal garden @crimson ferry The .py versions are available on https://circuitpython.org/libraries - the "Bundle Version py" is the .py versions of every lib. At least when GitHub is working.
tidal garden
@idle owl thank you!
simple pulsar
@tidal garden The .mpy normally come from a bundle and the VERSIONS file in the zip bundle will tell you which version you have if you need to match them up perfectly.
slender iron
@obsidian compass it's definitely possible. I'm not sure of the process though. I think openocd is involved
anyone have an ili9341 handy? https://forums.adafruit.com/viewtopic.php?f=60&t=164561
lucid solar
@slender iron I've uploaded a test firmware for the ili9341 issue...if it doesn't work for him still, then I can try with a different board and the ili9341 display (I don't have the feather_m4_express)
slender iron
did you manage to replicate the issue?
lucid solar
I don't have the board to test it...but at least I know I've made the firmware with the right steps.
onyx hinge
@slender iron we're making 5.3.0-rc.0 from master right?
slender iron
that was my plan. the ili9341 bug above could be an issue
lucid solar
@obsidian compass https://learn.adafruit.com/programming-microcontrollers-using-openocd-on-raspberry-pi
onyx hinge
I'll read back
crimson ferry
Do you need a test? I have a Feather M4 and ili9341 / 2.4" FeatherWing... Mark's UF2 + code.txt (both from forum)?
onyx hinge
@crimson ferry I think testing with the latest github is what we're after
slender iron
@crimson ferry that'd be helpful!
crimson ferry
OK, give me a few
onyx hinge
he said, despite being only half-way up to speed
slender iron
thanks!
onyx hinge
I'll get started drafting release notes right now anyhow
slender iron
π
onyx hinge
@slender iron who can add 5.2.x and soon 5.3.x docs on RTD?
manic glacierBOT
slender iron
Do we have a 5.x branch?
onyx hinge
I don't think so
tulip sleet
I'm not sure we should have a 5.x branch (yet). in 4.x it just confused things about whether to use 4.1.x or 4.x
crimson ferry
Feather M4 + 2.4" FeatherWing (ILI9341) + latest S3 ...a4d86b9.uf2 + forum code.py = safe mode [5.2.0 UF2 is fine]
onyx hinge
I've touched displayio indirectly for rgbmatrix, seems like I should take a look at it then
lucid solar
@crimson ferry can you try this, see if it crashes
lucid solar
does mp_raise_RuntimeError cause CPY to go to safe mode?
crimson ferry
@lucid solar same behavior
onyx hinge
@lucid solar it shouldn't
I don't seem to have any ili9341. I wonder if I can run that code without HW connected.
lucid solar
slender iron
@onyx hinge I bet you can
@lucid solar safe mode is triggered by reset_into_safe_mode
crimson ferry
@lucid solar still safe mode with ...1c8c.uf2
lucid solar
@crimson ferry https://adafruit-circuit-python.s3.amazonaws.com/bin/feather_m4_express/en_US/adafruit-circuitpython-feather_m4_express-en_US-20200414-ada102d.uf2 << love the git hash for this one
timber mango
@slender iron I am really looking forward to your low power stuff. This will be great for robots too! π
slender iron
π if only I could get it checked in
onyx hinge
I got ValueError: Unsupported display bus type then a moment later dropped to safe mode
this is with some random "protomatter in progress" commit, which reinforces the idea it may be due to my changes.
timber mango
That is weird. Display only code should not care whether a display is actually connected or not.
slender iron
@tulip sleet want to review https://github.com/adafruit/circuitpython/pull/2789 ? it looks good to me
timber mango
Of course, it is nice to know whether a display controller is present or not.
onyx hinge
self=self@entry=0x20001d80 <displays+192>, brightness=brightness@entry=1)
at ../../shared-module/displayio/Display.c:162
162 self->updating_backlight = true;
(gdb) p self->core->bus
$4 = (mp_obj_t) 0x0
``` well that's clearly not going to work out π π¦@lucid solar Boom. Of course, the "ada" firmware works π
lucid solar
@onyx hinge Yeah, it does seem like the 'bus' thing is the issue. commit 09dc46a98480 seems to be where it broke. (Now I'm going to help my wife...back later)
onyx hinge
Thanks @lucid solar I agree with your finding!
onyx hinge
hooray for our community, I'm glad we didn't ship a release candidate with this bug.
slender iron
@onyx hinge looking now
onyx hinge
i'll be back in ~ 1h to check on the status of things and presumably continue on to blog & forum posts.
orchid basinBOT
onyx hinge
I'll probably schedule the blog post for tomorrow, as it'll be getting late
slender iron
@stuck elbow any ideas for slimming down the pewpew m4 build? somehow my sleep work is making it larger
stuck elbow
@slender iron it's mostly the graphics for the menu program and the pew emulator, I include them in the firmware because I don't want them to use RAM
@slender iron it already has the analogio disabled, that should have freed a lot of space?
I want to rewrite the pew emulator to not need external graphics, but I don't really have much usable time for this recently
slender iron
I'm only failing the DE and zh builds
stuck elbow
can you tweak that code inlining stuff in the makefile?
otherwise, I suppose you could disable AUDIOMIXER for those languages only?
I don't think any game uses it at the moment
onyx hinge
that reminds me .. assuming I didn't see it here, this is possibly relevant: https://github.com/google/bloaty
forum announce made, blog post schedule for tomorrow AM ET. thanks y'all!
slender iron
w00t 1872 bytes free now
stuck elbow
\o/
tulip sleet
@onyx hinge I just merged master into my branch, and I'm still getting that ulabuninitialized variable error I reported a couple of days ago when compiling DEBUG. Is there a backport from 5.2.x that needs to be done to master?
onyx hinge
@tulip sleet oops I didn't pull request to update ulab submodule
tulip sleet
halbert@salmonx:~/repos/circuitpython/extmod/ulab$ git status
HEAD detached at a746bd8
nothing to commit, working tree clean
is the ulab commit
onyx hinge
I'm on mobile, can you file a bug in circuitpython?
tulip sleet
sure, but that's not in the rc, right?
onyx hinge
It's not fixed in the rc
half geyser
Is it possible to include a binary module in circuitpython?
I have an AES module I've written, and right now I add it to shared-module and shared-bindings. I'm not sure if anyone else would find this AES module useful, so I'm considering just carrying it locally.
But it requires modifying py/circuitpy_mpconfig.mk and py/circuitpy_defns.mk.
Would anyone else be interested in this AES module? Or could I carry it in the board port without needing to modify files in py/?
lone axle
@half geyser your AES module is written with C is that right?
half geyser
@lone axle Yes, it's https://github.com/kokke/tiny-AES-c/blob/master/aes.c
The actual patch I have to add AES support is at https://github.com/simmel-project/circuitpython/commit/15aa364be7a63ee80cad564f4711d1c8e8e2119f
It passes the NIST tests at least!
lone axle
@half geyser I don't believe there is currently a way to use C modules (or any binaries made from them) by just including the files on the device. They must be "baked in" to the system I think. If you had a python implementation of the AES module then you could include it as a file on the device in the libs folder and use it within your code.
half geyser
Is there a way to "bake in" modules by appending them to e.g. the mpconfigboard.mk and/or mpconfigboard.h files? I can't see any modules that do that -- most seem to simply override various common-hal files.
lone axle
Also if you had a python implementation of it, there is a "frozen" file that is made from the .py file it's called .mpy and the tool that makes them is mpy-cross.
tulip sleet
they are all conditionalized with CIRCUITPY_whatever
I'm not sure what you mean by "appending", you mean just listing the sources? We don't do that; we use the make ifdefs and turn things on and off
half geyser
@tulip sleet: I made a conditional CIRCUITPY_AES that I check against, which is how I'm able to get it to build.
half geyser
But there's no concept of PORT_MODULES or BOARD_MODULES in the same way that there is a SHARED_MODULES, is there?
tulip sleet
no, they'd just be in common-hal, if they were port-specific
half geyser
That does seem like a good policy. It prevents board- and port-level fragmentation.
tulip sleet
_bleio is like that, only exists in nrf, so it's port-specific (mostly), but controlled as if it's for all prots
yes, our intention is to make the ports as uniform as possible
this is not micropython's philosophy, which is to expose capabilities of a port in the interest of increased functionality for that port, even if it differs from others
half geyser
Would there be any interest in an AES module? I'm still not sure if it's necessary for my usecase. I suspect it is, which is why I did the port.
tulip sleet
it doesn't hurt! and it may be handy for some network stuff? We can always just leave it off
half geyser
It might end up being called _aes.
tulip sleet
we'd add an underscore if we didn't want to guarantee the API across minor versions. If there's a wrapper library, that's fine. That's why it's _bleio
half geyser
Ah, that's the significance of it! I thought that referred to the fact that it was a low-level library. Gotcha.
One more question: Is there any SPI flash memory API?
Right now I'm using internal memory for the filesystem, and I want to use SPI flash as a database. Reads and writes are infrequent, so I don't want to do XIP.
From what I can see, you can either have internal flash or external XIP SPI. Is there any way to have block-level access to the SPI, or should I just write SPI wrappers in Python to manually manage the flash?
tulip sleet
we have microcontroller.nvm for simple bytearray access, but it's not block-level. i wrote a simple consecutive-logging db for bonding information in _bleio, but that's not visible to python
sorry, I misread your question, you meant the external flash
there is block-level access for SD flash that's used via python.
we actually want to rewrite that in C, because it's a big problem it's python calling C calling python
half geyser
I ran into the logging db from _bleio when my firmware pointers were incorrect. Circuitpython really gets unhappy when parts of its text section get erased π
That does look like a good approach, though! A logging db is exactly what I had in mind.
tulip sleet
https://github.com/adafruit/Adafruit_CircuitPython_SD this style was copied from micropython
i'm sorry i messed up the calculations of the regions; i was going to look at your fixes. I had to move things around a couple of times and it sounds like I forgot to fix the arithmetic.
half geyser
It's okay! I just wasn't sure if I got it right. It's not a well-exercised code path.
tulip sleet
it is -really- simple. There are fancier logging db's in the nrf SDK
half geyser
My big fear was that I mess it up for someone else.
tulip sleet
i will double-check it; i didn't get a chance to review that PR
hathach uses LittleFS for storing the bonding info in the Arduino nRF code. That looked interesting too, but it was more work to port it than I wanted
half geyser
Okay, adafruit_sdcard looks like it does state management in python.
Ah, I see that the C code inside spiflash.c just calls out to common_hal_busio_spi. So maybe it's not so bad to do everything in python.
tulip sleet
the problem we were having is displayio reading data for the display from the sd card, and also other stuff going on: https://github.com/adafruit/circuitpython/issues/2417
half geyser
If I'm using the internal flash filesystem, there's absolutely no utility in defining EXTERNAL_FLASH_DEVICES is there?
tulip sleet
right, you don't need it; take a look at the pca10059 board files, for isntance
half geyser
This is good information.
And what about the code generator? I don't see CIRCUITPY_ENABLE_MPY_NATIVE enabled for many ports. Does it offer much benefit?
tulip sleet
it's maybe twice as fast in some cases, but it makes the code bigger. we don't have room on the M0 ports for that. Our inclination is to code stuff that needs to be really fast as a native module
half geyser
Nice to know it offers some benefit at least.
tulip sleet
see https://github.com/adafruit/circuitpython/issues/1248 , which has some discussion, and PR #2271
half geyser
Well, this'll be fun. Now that we have the basic port done, I will need to work on some of the audio stuff. Unfortunately the PCB is designed such that it needs an I2S input and PWM output. Which is the opposite of what the nrf port currently has. At least there's a starting point. And PWM output should be simple.
tulip sleet
? for audio ? we are doing PWM audio output on the Circuit Playground Bluefruit already
mic is PDMIn
half geyser
Oh! I was looking at the common-hal/audiobusio/ directory and I only saw I2SOut and PDMIn. I didn't see common-hal/audiopwmio.
tulip sleet
we split it up because it wasn't a "bus". It is a bit confusing
half geyser
Is there support yet for complimentary PWM? This board has the speaker hooked up to two PWM channels with the plan to drive one signal as a differential pair for 2x the output power.
tulip sleet
nothing like that, I think; we had only our own boards in mind π
half geyser
Thanks for the information! It's good to get a sense of what exists and what doesn't. And this has answered a lot of questions I've had over the past week.
tulip sleet
feel free to ask scott or jeff or me any time
half geyser
It really was a quick and relatively painless board bringup. You've put together a really great ecosystem overall.
tulip sleet
thank you; we generate so many boards ourselves at such a pace that we really have to avoid too much special-casing
we really appreciate all the work you have put into it too; we really enjoy getting a larger and larger community
manic glacierBOT
@dhalbert we may need to upgrade the SD to v7 for nrf52833. S140 v6 is written before 833 is released
onyx hinge
good morning! I'll see about merging ulab and then I have some RGBMatrix oddities to chase down.
ionic elk
@tulip sleet what was the original purpose of small build? I know it was to enable smaller builds of circuitpython but is there a specific cutoff in terms of the chips it applies to?
Is it just boards using the SAMD21G18A
tulip sleet
@ionic elk it is SAMD21 builds that don't have external flash, so 64kB of the internal flash is reserved for the filesystem.
e.g. Trinket and Gemma M0
ionic elk
@tulip sleet sorry to bug again, do you think you could help me fill out this list? I'd like to put it in the documentation somewhere... I'd still like to make a template port at some point.
CIRCUITPY_DISPLAYIO = 0 # Technically has common-hal, but it's just a stub. Depends on SPI
CIRCUITPY_BITBANGIO = 0
CIRCUITPY_GAMEPAD = 0
CIRCUITPY_I2CSLAVE = 0
CIRCUITPY_PIXELBUF = 0
CIRCUITPY_RANDOM = 0 # Depends on OS
CIRCUITPY_STORAGE = 0
CIRCUITPY_TOUCHIO = 0
CIRCUITPY_USB_HID = 0
CIRCUITPY_USB_MIDI = 0
CIRCUITPY_REQUIRE_I2C_PULLUPS = 0
CIRCUITPY_ULAB = 0
Storage just requires internal flash, right? Pixelbuf I'd assume is Neopixel write?
tulip sleet
@ionic elk pixelbuf is a C optimization to speed up RGB LED writes. it is separate from neopixel_write. storage for is any filesystem, internal or external
when you say "fill this out", I'm not sure what you mean
there are many of these options, and a few a MICROPY instead of CIRCUITPY.
ionic elk
I'm trying to comment what each item depends on - it's the list you need to include in a port mpconfigport.mk so it compiles in the early stages.
I don't see pixelbuf in the common-hal list for atmel or the other ports
tulip sleet
i think pixelbuf is not quite ready
there is a python equivalent of it to save space on small ports
ionic elk
so normally it'd be in common-hal?
tulip sleet
no, it's shared-module, because it's port-independent. it's C helpers for some stuff that is kind of too slow in python
ok, it is turned on for CIRCUITPY_FULL_BUILD, so we're not waiting on anything
there is prerequisite info for these modules implied in https://github.com/adafruit/circuitpython/blob/master/py/circuitpy_mpconfig.mk
but I think you are looking there already
ionic elk
Yeah I'm looking at prerequisites that are a good bit earlier than were considered in that file.
CIRCUITPY_DISPLAYIO = 0 # Technically has common-hal, but it's just a stub. Depends on SPI CIRCUITPY_BITBANGIO = 0 # Requires DigitalIO CIRCUITPY_GAMEPAD = 0 # Requires DigitalIO CIRCUITPY_PIXELBUF = 0 # Does nothing without Neopixel Write CIRCUITPY_RANDOM = 0 # Depends on OS CIRCUITPY_STORAGE = 0 # Requires a filesystem implementation CIRCUITPY_TOUCHIO = 0 # Requires Microcontroller (for generic implementation) CIRCUITPY_USB_HID = 0 # Requires USB CIRCUITPY_USB_MIDI = 0 # Requires USB CIRCUITPY_REQUIRE_I2C_PULLUPS = 0 # Does nothing without I2C CIRCUITPY_ULAB = 0 # No requirements, just takes space
tulip sleet
pixelbuf can be used for DotStars too, so neopixel_write is not necessarily a prereq
if you want to add more conditionalization or testing in that file, that would be great
ionic elk
Well, actually right now I'm basically reverting all my conditionals since I'm realizing that they're kind of useless.
We could chat about it if you're open to it real quick
tulip sleet
@ionic elk do you want to audio chat?
spice crypt
Is there a functional flow-chart/diagram of the nitty gritty of how the CP ecosystem works? From bootloader to execution.
I see references to an API, but curious what that looks like visually.
Found this: https://learn.adafruit.com/adafruit-feather-m0-express-designed-for-circuit-python-circuitpython/uf2-bootloader-details but curious if there's more.
slender iron
@spice crypt I think that's the best reference. I'm always happy to answer questions
spice crypt
Alright. I hate to bother you with my frivolous questions. I'm on a quest to speed up the boot times, as you may have noticed. π
slender iron
Any time I spend teaching you comes back in spades. π
spice crypt
Digging into the nrf bootloader; see some delays I can shrink down.
slender iron
you are trying to speed it up when waking from sleep right?
spice crypt
Right now, just curious where the handover from bootloader to "firmware" is, and how they talk for things such as the SoftDevice(BT)
No, from fresh boot
I've decided to just do a global power off, rather than use sleep for now
spice crypt
I'm at 7 seconds to UI loaded
slender iron
how long does it take to get into circuitpython?
slender iron
seems like most of that 7 seconds would be graphics stuff
slender iron
you could set a pin high from C
spice crypt
Yeah, I figured I'll have to use some GPIOs to profile it.
I'm enjoying digging through the code
spice crypt
wakeup source? Assume that's a NRF register
spice crypt
I see that's used in the bootloader to determine if to start dfu. Anyway, I didn't understand what you meant by : "I'd love to see each delay check the wakeup source"
A ctrl+c, ctrl+d reset takes 2 seconds for first sign of life from CP
then 3-4s for the rest of my UI to load. So from raw poweron, looks like bootloader eats about 2s
slender iron
basically, only delay if reason is pin-reset or vbus
spice crypt
aha! vbus is what I'll be shooting after
slender iron
ctrl-d should be pretty quick. I guess the screen update may slow it down if showing the terminal
tulip sleet
there is usb enumeration time too
slender iron
but vbus should delay because it's the same as plugging in
spice crypt
there is usb enumeration time too
@tulip sleet Is this regardless if USB host is present or not?
tulip sleet
I think the check for USB present is quick, but I'm not really sure. i think it's an electrical check by the USB peripheral
spice crypt
I shall experiment. Thanks for your input guys.
proceeds to continue premature optimizations
slender iron
π learning is helpful
anyone planning on automating the download stats again? @tulip sleet or @raven canopy? Every release causes the downloads page to be sorted unhelpfully. Maybe we need to move assets back to github so we can grab the stats easily?
tulip sleet
@slender iron I would like to process the download stats. Did you put your scripts anywhere? I think moving back to github is not a good idea at all
tulip sleet
at the very least we could sort it alphabetically for now
slender iron
I can manually do it but that doesn't happen immediately
I don't think we should have moved off github until we had this setup
the fix for download stats with github would have been simpler
tulip sleet
i kind of disagree. The issues with github was not download stats,they were failed uploads and huge release pages. I will work on this.
i spent a lot of painful time fixing up releases
slender iron
π I agree github wasn't perfect. I'm just frustrated by the regression of the stats.
tulip sleet
i will put it on my short list
onyx hinge
tfw you go looking for a bug, and the bug goes into hiding
I would have estimated that 1/20 of the time I would see a crash while updating code.py when an rgbmatrix display was in the mix. but now it hasn't happened in 100+ tests
ionic elk
What's the best way to generate documentation that needs line breaks at a specific column width? Do we have a python script, or is there an editor I should be using, or what?
tulip sleet
@ionic elk what's the formatting? Is this an .rst or an .md or ?
tulip sleet
.rst should reflow paragraphs, which are separated by blank lines. so the input width shouldn't matter
ionic elk
so I can just write with word wrap on, and then save, and it should be ok?
tulip sleet
or is this in code?
ionic elk
I'm editing porting.rst in Sublime Text 3
tulip sleet
yes, just set your local word wrap to match the file approximately
it should reflow the paragraphs (e.g. compare the input with the readthedocs output)
obsidian compass
Is there any blog/tutorial on how to use raspberry pi as SWD programmer? I am especially looking to re-program particle xenon board to be used with CircuitPython but don't have a SWD programmer, I read rpi can do SWD, so is it possible to flash bootloader on xenon using rpi?
@obsidian compass
so i managed to do it and wrote a blog on the same : https://codensolder.com/blog/rip-swd-programmer
It would be great to do SWD bit-bang using an existing CP board, @slender iron I saw your issue for the same, would be great to have that feature π
tulip sleet
i actually tried this to do some fuse fixing before I got a J-Link; sounds like you did the same thing. I had more trouble with not-great chip support in in openocd
but that was a few years ago
obsidian compass
yeah it's way easier if companies provides configuration files in openOCD, for NRF52 it was easy, I just spent two hours figuring how to put particle bootloaded to DFU which goes like push RESET and MODE button, then release RESET until you see GREEN blinky LED, after that it was easy
Also i didn't had the mini jtag connector so I soldered directly on the small pin headers π
ionic elk
@raven canopy I'd like to work on https://circuitpython.readthedocs.io/en/5.0.x/docs/porting.html#porting. Can you recommend a good editor for RST text? I'm new to it and Sublime really doesn't like the word wrapping.
slender iron
@ionic elk how much time do you expect to spend on it?
ionic elk
Like 10m? I can just ignore it if you want
slender iron
what is the issue?
ionic elk
I just ran into it as I was skimming and it doesn't make a lot of sense. Figured I could fix it up pretty quick
It's just very confusingly worded
slender iron
I use sublime just fine
ionic elk
For me sublime acts like there's a newline after every 80 columns, and if I type in the middle of a paragraph, I have to go down and fix alignment on every line after it until I hit a paragraph gap
slender iron
ya, I think that's what I do
ionic elk
Oh π¦
had the same issue with Zephyr documentation
Figured people were using some kind of autobreak tool I didn't know about
slender iron
I just have the column line turned on so I know
lucid solar
@ionic elk there's probably some autoformatter for rst in VSCode
slender iron
I imagine there is one
ionic elk
Yeah it just makes everything more work when I gotta get in there and fix like 6 lines just to edit one word choice
slender iron
yup. sometimes I just do small things the hard way
tulip sleet
sometimes I start each sentence on a new line and don't worry about refilling. That makes it easy to edit, and the formatter takes care of the reflow
lucid solar
Does someone have some trick they use for debugging whilst developing. I can output to the USB-CDC, but that closes down on re-flash...If there isn't something there already, maybe we enable the VCP/ST-Link to output the 'console' also?
tulip sleet
i just reopen the serial connection; i have some aliases to maek that easy to type
ionic elk
Ok, I found a good thing - Sublime->edit->wrap->"Wrap paragraph at 80 Characters" for you @slender iron
tulip sleet
I do mp_printf(&mp_plat_print, ...) a lot
lucid solar
@tulip sleet I just think, that if a board has the debugger onboard (nucleo/disco) then we should make use of it.
@ionic elk did you have some thought on this whilst making the STM32 port?
ionic elk
@lucid solar not sure what you mean. I use onboard STLink sometimes with arm-none-eabi-gdb and OpenOCD or texane/stlink
tulip sleet
i am doing that in conjunction with gdb
ionic elk
But that doesn't require anything extra from Circuitpython other than turning on symbol flags in the build process
lucid solar
I mean....being able to printf something to the VCP console through the st-link
ionic elk
I rarely do a printf when debugging, I just use print
slender iron
@lucid solar you are proposing using the SWO right?
the serial output through the debug connection
lucid solar
@slender iron we could...but there is the VCP...much easier to use I think
slender iron
that's a uart out to the debug chip right?
lucid solar
yup
slender iron
I'd accept a PR for it. I either reconnect over screen when printing or switch to iterating with gdb and no prints
tulip sleet
mostly we are not using dev boards with VCP built-in, so it hasn't come up that much
lucid solar
Ok...I'll make the code π DAPLink also has that feature.
tulip sleet
that is, we are debugging on the adafruit boards
lucid solar
If we make it so a CPY board can work as a debugger/reprogrammer (CMSIS-Dap), then we could also enable that feature there.
@tulip sleet what do you use as the SWD programmer? j-link?
tulip sleet
yes. I also have a BlackMagic, but until recently it didn't have SAMD51 support. The J-LInk has been mosty fine
and it supports so many chips
Adafruit bought me a commercial one since I'm using it for pay
lucid solar
aha...nice π
ionic elk
@slender iron is sommersoft around? I realize we need a regex thing to properly identify certain modules in mpconfigport.mk for stm32 now but thats not something I'm practiced in
slender iron
how would I know?
tulip sleet
he is not online (click the "persons" icon at the top), but if you tag him he may answer later
lucid solar
@ionic elk I was looking in the wrong file, now I understand
ionic elk
@slender iron sorry, I just meant if he was out of the country or something.
tulip sleet
he has a day job
ionic elk
ah
slender iron
(but is US based)
lucid solar
@ionic elk what do you want to do? enable the modules depending on the MCU?
slender iron
goes to exercise
ionic elk
@lucid solar just pushed. parse_port_config in shared_bindings_matrix.py needs to be adjusted to fit stm/mpconfigport.mk
lucid solar
@ionic elk I see...
ionic elk
Issue is the H7 defines are turning off modules for the F4
lucid solar
@ionic elk nearly done...I can see why it's not matching
ionic elk
I added
elif port in ["stm"]:
re_board_chip = re.compile("MCU_SERIES\s=\s(\w+)")
chip_keyword = "MCU_SERIES"
but still no dice so far
tulip sleet
use r"..." because you have backslashes that are interpreted by re
the backslashes will get eaten by ordinary strings. using r (raw strings) prevents them from being interpreted.
lucid solar
ifeq\s($(MCU_SERIES),(\s)*(\w+)
it didn't seem to find it because (\s)* wasn't there
tulip sleet
it needs to be r"MCU_SERIES\s=\s(\w+)"
also as you point out should be \s* to make whitespace optional
r"MCU_SERIES\s*=\s*(\w+)
and any length of whitespace
lucid solar
@ionic elk FYI, I use https://regex101.com to test things...
tulip sleet
now you have two problems
ionic elk
still not getting anything
tulip sleet
what is the new code
ionic elk
the tester should be r"MCU_SERIES\s*=\s*(\w+)" to recognize ifeq ($(MCU_SERIES), F7)? do I have that right?
tulip sleet
could you point to this file in a repo, also?
tulip sleet
are you using match() or search() ? match has to match the whole input string
i'm confused, are you trying to match ifeq ($(MCU_SERIES), F7) ? That has no = in it. The above is going to match MCU_SERIES = <something>
ionic elk
Yeah I'm a knucklehead this is for the compile
got it! thanks @tulip sleet @lucid solar!
lucid solar
great!
slender iron
@ionic elk did you see how I reworked the hal_conf stuff in my lower_power branch?
ionic elk
Taking a look right after I get this docfix PR in
lucid solar
Anyone good at bitmasking?
lucid solar
I have a uint16_t value...I need to split it in 2...lower byte and upper byte
8 should take the upper?
ionic elk
yes
lucid solar
& 0xFF lower?
ionic elk
yep
lucid solar
thanks...turns out I did know
There seems to be an issue with never_reset_pin_number not working in stm32
lucid solar
Turns out I misunderstood how it saves the pin...it still doesn't work, but at least I can debug it.
manic glacierBOT
ionic elk
@slender iron ok reviewing low power now!
slender iron
π
lucid solar
@slender iron if an adafruit board has a built in display, you initialize it inside board_init right?
ionic elk
@lucid solar yes - check out the stm32 Meowbit or the Hallowing for examples
lucid solar
I see...just wanted to check that's the right way to do it.
slender iron
π
raven canopy
@ionic elk sorry for being a π today! i'm looking over it now. and yeah, i missed every r"" literal when i wrote it initially and haven't fixed it. deshipu fixed one a while back..
re: rST editing... i'm like Scott; the hardway. π
lucid solar
anyone know how to fix
../../py/circuitpy_mpconfig.h:335:50: error: 'MP_QSTR_displayio' undeclared here (not in a function); did you mean 'MP_QSTR_digitalio'?
335 | #define DISPLAYIO_MODULE { MP_OBJ_NEW_QSTR(MP_QSTR_displayio), (mp_obj_t)&displayio_module },
make clean and rebuild?
QSTRs don't regenerate properly most of the time, so they force you to clean when you add new modules.
lucid solar
aha thanks
orchid basinBOT
slender iron
@ionic elk please don't ask folks to remove attribution. if they want ownership of it that is a good thing. I'd only object to them removing other folks
tulip sleet
@slender iron nope, it doesn't work, so maybe it's Linux-specific
slender iron
it's flaky for me
usb problems usually means we're sleeping when we should be calling tinyusb
ionic elk
@slender iron Dan and I were talking about it for big list files like the STM32 has where we fold in macros or blocks of settings to add new board/SoC support, since they could lead to long and kind of confusing lists of attribution for stuff that doesn't really alter things structurally. If we do add them maybe we could attribute on a board-by-board level, where the name goes by the board that was added?
slender iron
@tulip sleet I can't replicate it when on gdb. can you?
tulip sleet
@slender iron did you compile DEBUG?
slender iron
I tried both though it looks like I may have triggered it with debug
er, without debug
tulip sleet
without debug, REPL arrived a little delayed the first time, but then the second time it didn't work
(with j-link and gdb connected)
could try lowering the background interval to say 500 or 800us instead of 1ms; is that possible or is it 1/1024 based?
but i'm wondering if the USB peripheral is shutting down or something
manic glacierBOT
slender iron
@tulip sleet I thought I removed the background interval
tulip sleet
oh, hmm
tulip sleet
could debug by setting a pin when call the tud task function, and saleae-ing the timing on working vs not working
slender iron
ideally anything usb will trigger an interrupt that then wakes up the loop and then runs background tasks
manic glacierBOT
tulip sleet
@slender iron maybe turn off the actual sleep now, to confirm that the new USB interrupt strategy works,
ionic elk
Ok I'm out for the day to make dinner be back tomorrow
slender iron
@tulip sleet that's easy to do. it works for me. want a uf2 or to poke it yourself?
tulip sleet
if i'm going to debug it, tell me the change, or a uf2 is easy
slender iron
comment this out to prevent the sleep: https://github.com/tannewt/circuitpython/blob/lower_power/ports/atmel-samd/supervisor/port.c#L491
wfi = wait for interrupt
this is the sleep loop: https://github.com/tannewt/circuitpython/blob/lower_power/supervisor/shared/tick.c#L89
tulip sleet
@slender iron commenting out the WFI didn't help. It worked, and then didn't work. (J-Link is disconnected)
tulip sleet
@slender iron~~ if I power cycle it by pulling the plug, I can get the REPL to appear. If i press the reset button, it doesn't~~
tulip sleet
ok, no, that's not true, I pulled the plug and now it didn't work. It's just flaky
slender iron
must be timing since it's ok on samd51
tulip sleet
the samd51 might be quicker on the draw on the USB interrupt
slender iron
ya
tulip sleet
i'll be off for a while but will check back later tonight
raven canopy
@tulip sleet just a random thank you! i hit some USB bugs that caused total crash, with a metro m4 when flashing new firmware with rosiepi. (when do these hurdles stop? haha)
updated the bootloader, and it seems to be fixed. π
slender iron
anyone planning on attending the virtual pycon edu summit? https://us.pycon.org/2020/events/edusummit/
simple pulsar
I've found some line-noise-like junk printed over USB serial console from 5.2.0. When GitHub starts working again I'll add a comment on existing issue...
lone axle
Does anyone know of an existing tool that is like Circup, but is for upgrading the version of Circuit Python itself on the device rather than the versions of libraries?
simple pulsar
@lone axle Do you mean one that doesn't require the user to initiate (could be viewed as authorise) the action with the double-click on reset button?
lone axle
@simple pulsar correct. My thought is to have the program connect via serial and get in to REPL and issue reset to bootloader command from there.
simple pulsar
lone axle
π©
simple pulsar
Here's my junk on serial console ```
0000000 Q u a n t i z e 2 3 2 . 5 6 1
0000010 2 3 2 . 5 6 1 e 3 4 8 0 . 0 5 1
0000020 _ X u 357 277 275 y 342 226 222 342 226 222 342 226 222
0000030 342 226 222 0 342 226 222 } 342 226 222 2 2 0 . 0
0000040 \n
0000041
Need to have a reasonably wide window for that to display nicely. The separator in print which should have been a space character has turned into a load of junk.
the args to print there would have been "Quantize", 232.561, 220.0 - curiously, 232.561 is repeated. Just added all of this to GitHub, I think it's in the stages of recovering at the moment.
raven canopy
@lone axle not a polished end item, but this is what i use with rosiepi: https://github.com/adafruit/circuitpython/blob/master/tools/cpboard.py
lone axle
nice, ty
raven canopy
requires changes for CPython3.7+; has instances of async as variable names. https://github.com/sommersoft/circuitpython/commit/c5415cb82edf6a42ba11d04069e975809be311fa#diff-51ce48f797ab56188d3824c76f125eb2
old smelt
Getting an error on GitHub trying to download zip file of the AzureIoT repo. Can't clone it from a terminal either.
https://github.com/adafruit/Adafruit_CircuitPython_AzureIoT
remote: Internal Server Error.
remote:
fatal: unable to access 'https://github.com/adafruit/Adafruit_CircuitPython_AzureIoT.git/': The requested URL returned error: 500```
Anyone else seeing GitHub oddities?@old smelt I was getting a lot of 504 earlier. Probably still buggered.
orchid basinBOT
tulip sleet
also see https://www.githubstatus.com/
old smelt
Thanks @tulip sleet
manic glacierBOT
tulip sleet
@slender iron when you are available, could we talk about .packet_size before I submit another commit?
prime flower
@slender iron thanks for the link to the edu summit, I'm going to be offline for class-rel. tomorrow but I'm now reg'd to attend this.