#circuitpython-dev
small comet
qq around the type annotation topic (with apologies if I overlooked some documentation) - is there a CI config one can enable to pre-commit/check for missing type annotations once they've been added? (I've been running mypy by hand, but looking for a more standardized config if such a thing already exists)
lone axle
qq around the type annotation topic (with apologies if I overlooked some documen...
running mypy by hand is the only way right now really. Once we finish up the remaining library types I imagine at some point we'll add the check to pre-commit or actions somehow. There was this one: https://github.com/adafruit/Adafruit_CircuitPython_PIOASM/pull/51 which is one way that it could be integrated. You could carryout similar changes in your local instance while your working so you could test slighly less manually. But in the actual repos I think we won't want to enable the automated check until we are ready to do it on all of the libraries accross the board, so that we don't end up with some libraries using different checks than other ones.
manic glacierBOT
small comet
running mypy by hand is the only way right now really. Once we finish up the rem...
Makes sense - a couple of quick follow-ups: can I find a set of options for mypy that would be the "canonical" one enabled across the libraries? (Was also interested in your thoughts around regression - i.e., might it become an issue to fix one repo, and new changes cause issues before the global pre-commit check is enabled?)
lone axle
Makes sense - a couple of quick follow-ups: can I find a set of options for mypy...
The command that was used to generate all of the initial issues can be found here: https://github.com/FoamyGuy/missing_type_issue_maker/blob/06cd8949061a812ff7648d5c020b7577eb9d5667/find_missing_type_annotations.py#L8 it was essentially mypy [path to lib] --disallow-untyped-defs And for the initial pass we only went after lines that mypy reported "Function is missing a type annotation" on. Those would have been the ones that appeared within the list in the issue. That is probably as close as we have atm to a canonical configuration. We'll probably take some time during a meeting one week when we're ready to enable it more widely to discuss what configuration options we want to use and how strictly we will enforce mypy's preference.
I'm not sure if I understand you're second question. If you mean would it have been possible for the libraries to get changed in a way that created new code that now needs to be typed in order to satisfy mypy, even when that code didn't exist at the time we ran the check initially. I do think that's possible and likely something we'll have to work through when we "flip the switch" The goal before that is to get the vast majority of the "low hanging fruit" taken care of before that switch flips so that we lessen how much additional work will be needed once the checks do actually begin.
small comet
The command that was used to generate all of the initial issues can be found her...
perfect - thx for the pointer to the script!
Re: second question - gotcha. I'm realizing I'm probably confused about the mechanics 🙂 at a high-level, when the flip is made would all repos get a new stanza in .pre-commit-config.yaml pointing to a common workflow with the shared options? Or would the shared options be copied into each stanza directly? The comon standard makes perfect sense; and I was sort of wondering if it might be also be achievable incrementally (aka, one repo at a time, but adhering to the same standard.)
lone axle
perfect - thx for the pointer to the script!
Re: second question - gotcha. I'm ...
I don't think we have the full plan nailed down yet, and I'll admit that actions / pre-commit is not an area of strength in my knowledge. That being said I do think adding a new section to the pre-commit-config.yaml that points to a common workflow would be what we do. Or maybe we have that mechanism in place already, so we'd be adding it into that already existing actions task repo (these are the specifics I don't quite know about).
It would be technologically possible to do at different times, but I think the preference will be to do them all at once so that we keep a level of consistency across all libraries as best we can. Enabling it on individual repos as they're ready would mean that some libraries have different "rules" for PR submissions which would add complexity to reviews and could cause confusion among less technical contributors if it's unclear why one library PR is failing for a reason that they happen to know a different library PR passed on.
All of that being said, these are all great questions. And all I can say is my best guess. If you happen to ever be available on a Monday this would be a great topic for 'In the weeds' during the weekly meeting to get a wider discussion going involving the rest of the team as well.
small comet
Thx for the extra context, makes sense!
(Not easy to manage a large collection of open-source repos for sure, and kudos to all of you for making it work well 🙂 )
onyx hinge
synthio has a different maximum polyphony depending on the board (because processing more notes takes more CPU time and does more RAM waveform accesses). The original implementation used 2, and I've increased it to 12 on various boards where that tested OK. CPU usage can be greater or lesser depending on the sample rate as well as whether advanced features are used (ring modulation, filters, and to a lesser extent math & lfo blocks).
With recent changes to docs generation we COULD include the information about polyphony in the docs automatically, rather than having an outdated sentence about it.
one problem is that the list can't replace the existing 'supported on these boards' block.
anyway, is this worth it?
orchid basinBOT
tulip sleet
synthio has a different maximum polyphony depending on the board (because proces...
Could the hard limit be removed? That is, for simple synthesizing, could the polyphony be greater than 12?
@slender iron @onyx hinge I have made draft release notes for 9.0.0-alpha.1. There is a PR I will merge soon to update the frozen libraries, and there are a few pending new board defs I might wait for.
onyx hinge
Could the hard limit be removed? That is, for simple synthesizing, could the pol...
It would be difficult to make it fully dynamic, I think
tulip sleet
ok, so there is an array def or something somewhere that depends on the value. I wasn't sure about that
onyx hinge
Yes that's right, I think there are 2 or 3 parallel arrays of that size in the synthesizer
shared-module/synthio/__init__.h: mp_obj_t note_obj[CIRCUITPY_SYNTHIO_MAX_CHANNELS];
shared-module/synthio/__init__.h: uint32_t accum[CIRCUITPY_SYNTHIO_MAX_CHANNELS];
shared-module/synthio/__init__.h: uint32_t ring_accum[CIRCUITPY_SYNTHIO_MAX_CHANNELS];
shared-module/synthio/__init__.h: synthio_envelope_state_t envelope_state[CIRCUITPY_SYNTHIO_MAX_CHANNELS];
and there are some things that do linear searches of all the note_obj[]s which might become a problem
onyx hinge
@tulip sleet I thought there were more PRs needing review by me but I just saw the one. If there are more, give me a ping
tulip sleet
<@329766224093249548> I thought there were more PRs needing review by me but I j...
I originally had you on some of the backports, but added Scott since you were away most of yesterday.
onyx hinge
I originally had you on some of the backports, but added Scott since you were aw...
thanks scott 🙂
onyx hinge
@lone axle hey, I filed https://github.com/adafruit/Adafruit_CircuitPython_Requests/issues/128 because I noticed the docs say that requests.set_socket is deprecated .. however .. I'm not sure what the equivalent would be with Session() and esp32spi, because of the whole fakessl thing. do you know?
it sorta feels like fakessl belongs in esp32spi in some way but that'd be another incompatible change 😕
lone axle
<@382939733107408897> hey, I filed https://github.com/adafruit/Adafruit_CircuitP...
I don't know for certain. I think the new way would look more like this:
socket = socketpool.SocketPool(wifi.radio)
https = requests.Session(socket, ssl.create_default_context())
but as far as I am aware ESP32SPI would still need the FakeSSLContext stuff in order to work.
Upon first consideration I would probably agree that FakeSSLContext probably makes more sense to exist in ESP32SPI as it's equivalent of the ssl core module for the devices that have it. Then it could be situated so the user code could be very similar, differing only in imports I think.
I would mean a breaking change for requests if we enforce the deprecation / removal of set_socket. But I think it's more positive overall because then the usercode for ESP32SPI and built-in wifi will be more similar.
I guess maybe user code could be structured that way today even without moving FakeSSLContext though now that I think about it a bit more. So perhaps that is actually a different thing apart from moving it. If ESP32SPI is the only thing that use the FakeSSL stuff (to my knowledge it is, but perhaps there is something I'm unaware of) then I do think it's more fitting over in ESP32SPI.
manic glacierBOT
onyx hinge
I guess maybe user code could be structured that way today even without moving F...
I'm not sure if/how wiznet does ssl
lone axle
I'm not sure if/how wiznet does ssl
As far as I know it doesn't. All of the examples in that repo have http:// URLs specifically. I do think I tried it at one point, but it would have been a while ago, and I don't recall exactly how it behaves.
slender iron
<@252717193496756235> <@426200547024961539> I have made draft release notes for ...
Do we want to wait for a 9.x pre-release until 1.20 is merged with the mpy version we'll release with?
idle owl
Oooh hmm.
idle owl
Do we want to wait for a 9.x pre-release until 1.20 is merged with the mpy versi...
From my perspective, that sounds like a good idea.
mpy versions changing in the middle of an existing release seems like asking for support issues.
tulip sleet
Do we want to wait for a 9.x pre-release until 1.20 is merged with the mpy versi...
That could be a while; I was wondering whether I should call it 8.9.0-alpha.1 or 8.3.0-alpha.1 or something. We don't have to release either of those as stable.
IT would be nice to have a named release with the usb host and other new stuff in it
8.99.0-alpha.1
idle owl
Concur.
slender iron
but I'd also like to not distract from the mp merge
tulip sleet
The real reason I made release notes for an alpha is that I had to write up the summary lines for each 8.2.1 PR anyway.
so there is a lot of overlap between the x.x.x-alpha.1 and 8.2.1 release notes
I am about to finish the 8.2.1 release and didn't plan to do anything with the alpha release for several days, until we got more stuff merged. So we can cogitate and discuss this tomw or Monday.
thorny jay
maybe it doesn't
That's what I remember and the reason I waited so long before getting two modules (a Feather and a generic one).
At one point I was thinking you could use your magic and play the trick you did for the Pico W SSL support... but except for the technical challenge, I don't see the interest for Adafruit.
manic glacierBOT
thorny jay
That's what I remember and the reason I waited so long before getting two module...
One day I will play with those, waiting for that, I watch @lone axle stream that frequently review PR.
manic glacierBOT
manic glacierBOT
orchid basinBOT
spare jacinth
Any tag/recommendation for "mid-level" PRs/milestones i could help with?
Like, i dont want to cleanup or add type hints, but wont write the best code ever either
Im familiar with both py and C(not with the bindings between them but shouldnt be much trouble getting stuff to work i hope).
Available hw: RP2040, STM32H750 (is it supported?) and ESP32S3
slender iron
Any tag/recommendation for "mid-level" PRs/milestones i could help with?
Like, ...
Lemme look. The H750 may not be too bad to support
we have H743 support
Here is a pretty straightforward bug fix for stm: https://github.com/adafruit/circuitpython/issues/7615
spare jacinth
Will take a look, thx for the pointers
FWIW: i also have a few displays (will try and write a driver for STM's LTDC peripheral as soon as another screen arrives), and have been working on a TOML library lately, not sure if we want further func on getenv (supporting bools would be nice imo)
devout jolt
https://blog.adafruit.com/2023/07/25/circuitpython-8-2-1-released/
thank you, Dan!
slender iron
I think bool values could be nice but not sure if getenv would actually give you them in CPython
40 pin RGB display support is definitely on our radar
so it'd be cool to support
spare jacinth
My radar is 27 i think
slender iron
we're not actively working on it though
spare jacinth
RGB888(24) + a couple control signals
slender iron
I think this is the same, just more control signals: https://www.adafruit.com/search?q=40+pin+display
from the datasheet for https://www.adafruit.com/product/1596
spare jacinth
I think bool values could be nice but not sure if getenv would actually give you...
I dont see a reason not to, but will check in a min, as soon as i get back to the computer
My current parser allows un-quoted strings (foo = bar), which afaik arent supported on TOML spec, but seems easier to use 😛
I also allow both TOML (true) and python syntax (True) - read: case insensitive - for bools. Again out of the spec but seems useful
Differences, if documented and/or for QoL, doesnt sound bad imo
spare jacinth
But again, dunno how much of a compliant parser we want....
spare jacinth
I think thats the same i have bought, some extra pins for power and touch sensor
Will hopefully get something working at some point :p
midnight ember
In issues notes it says Espressif boards have ESP-IDF storage leaks and occasionally crash after extended WiFi use. What time frame is considered "extended use"? Weeks/Months? I've had my Adafruit ESP32-S3 feather pulling weather data from openweathermaps & local sensors then pushing it to adafruitIO via MQTT for weeks at a time without issue. Most I've run it (according to monotonic) is for 2.5 weeks before coming back to it for some more coding. It sits on my workbench running 24/7 updating every 15 minutes. I've not had a single issue with it since about CP 8.0 beta 3.
crimson ferry
In issues notes it says `Espressif boards have ESP-IDF storage leaks and occasio...
Yeah, if I do requests only, ESP projects will run virtually forever (though I do occasionally reload or reset in software if certain conditions arise). Several have been running since some time in 2022 (still on CP7). But I get spontaneous resets or hard faults on some projects where AP or HTTP Server or other wifi or socket stuff is thrown into the mix. More specialized issues with ESP-Now frequently resetting, or monitor init hard faults are probably separate issues.
proven garnet
@tulip sleet did you mean to tag me on the TT21100 issue?
I don't have a datasheet but suspect you meant to tag someone else most likely, unless I'm not understanding something.
I don't think I have the device either
tulip sleet
I don't have a datasheet but suspect you meant to tag someone else most likely, ...
I thought you wrote the library, but I was looking in the wrong place - sorry!
Scott wrote it
proven garnet
No worries! I just felt bad giving an entirely unhelpful answer of "no" 😅
meager fog
@crude blaze y0000000000ooooooooooooooooo...............
crude blaze
@meager fog hey!
crude blaze
yea I got time for a good while!
meager fog
ok so i think i just dont have the right dto setup
i cloned/ran the example for the icn on the pi4
crude blaze
yea sounds likely, i2c worked fine via the Pi? Or did you ran it on the samd?
meager fog
i did have to run it twice due to a CRC error (?) but i dont think thats important
root@devpi:/home/pi/Adafruit_CircuitPython_ICN6211# python3 examples/icn6211_simpletest.py
0x00 = 0xc1
0x01 = 0x62
0x02 = 0x11
0x03 = 0xff
0x08 = 0x00
0x09 = 0x10
0x10 = 0x40
0x11 = 0x88
0x14 = 0x83
0x20 = 0x20
0x21 = 0xe0
0x22 = 0x13
0x23 = 0x28
0x24 = 0x28
0x25 = 0x28
0x26 = 0x00
0x27 = 0x0a
0x28 = 0x0a
0x29 = 0x0a
0x2a = 0x01
0x31 = 0x78
0x32 = 0x04
0x33 = 0xff
0x34 = 0x80
0x35 = 0x08
0x36 = 0x28
0x51 = 0x20
0x56 = 0x92
0x63 = 0xff
0x64 = 0x00
0x65 = 0x00
0x66 = 0x00
0x67 = 0x00
0x68 = 0x00
0x69 = 0x03
0x6a = 0x00
0x6b = 0x31
0x7a = 0x3e
0x7b = 0xff
0x7c = 0xf1
0x7d = 0x01
0x80 = 0x43
0x81 = 0x0f
0x84 = 0x94
0x85 = 0xc9
0x86 = 0x28
0x87 = 0x00
0x90 = 0x05
0x91 = 0x0a
0x92 = 0x18
0x94 = 0x0d
0x95 = 0x04
0x96 = 0x64
0x97 = 0x00
0x99 = 0x05
0x9a = 0x96
0xb5 = 0xa0
0xb6 = 0x20
SoT error
SoT sync error
False control error
ECC error, single-bit
ECC error, multi-bit
CRC error
DSI Data type Not Recognized
root@devpi:/home/pi/Adafruit_CircuitPython_ICN6211# python3 examples/icn6211_simpletest.py
0x00 = 0xc1
0x01 = 0x62
0x02 = 0x11
0x03 = 0xff
0x08 = 0x00
0x09 = 0x10
0x10 = 0x40
0x11 = 0x88
0x14 = 0x83
0x20 = 0x20
0x21 = 0xe0
0x22 = 0x13
0x23 = 0x28
0x24 = 0x28
0x25 = 0x28
0x26 = 0x00
0x27 = 0x0a
0x28 = 0x0a
0x29 = 0x0a
0x2a = 0x01
0x31 = 0x78
0x32 = 0x04
0x33 = 0xff
0x34 = 0x80
0x35 = 0x08
0x36 = 0x28
0x51 = 0x20
0x56 = 0x92
0x63 = 0xff
0x64 = 0x00
0x65 = 0x00
0x66 = 0x00
0x67 = 0x00
0x68 = 0x00
0x69 = 0x03
0x6a = 0x00
0x6b = 0x31
0x7a = 0x3e
0x7b = 0xff
0x7c = 0xf1
0x7d = 0x01
0x80 = 0x00
0x81 = 0x00
0x84 = 0x94
0x85 = 0xc9
0x86 = 0x28
0x87 = 0x00
0x90 = 0x05
0x91 = 0x0a
0x92 = 0x18
0x94 = 0x0d
0x95 = 0x04
0x96 = 0x64
0x97 = 0x00
0x99 = 0x05
0x9a = 0x96
0xb5 = 0xa0
0xb6 = 0x20
im running it on the pi, since the samd21 is low on ram, and i figured the pi is easier to tweak
crude blaze
yea for sure! I have not tested the lib at all on MCU CPY actually so far
crude blaze
the errors I imagine are likely due to lack of DSI signal
orchid basinBOT
meager fog
i have the icn wired to main i2c bus and i just tweaked the i2c bus instantiation in the example
crude blaze
lemme look up the boot flag that I used last time to debug the device tree stuff, that gave a whole lot more error logs in the syslog
meager fog
dsi is just a plain dsi cable
i didnt change any boot flags, i just added the compiled icn.dto you gave me
root@devpi:/home/pi/Adafruit_CircuitPython_ICN6211# cat /boot/config.txt
# For more options and information see
# http://rpf.io/configtxt
# Some settings may impact device functionality. See link above for details
# uncomment if you get no picture on HDMI for a default "safe" mode
#hdmi_safe=1
# uncomment the following to adjust overscan. Use positive numbers if console
# goes off screen, and negative if there is too much border
#overscan_left=16
#overscan_right=16
#overscan_top=16
#overscan_bottom=16
# uncomment to force a console size. By default it will be display's size minus
# overscan.
#framebuffer_width=1280
#framebuffer_height=720
# uncomment if hdmi display is not detected and composite is being output
#hdmi_force_hotplug=1
# uncomment to force a specific HDMI mode (this will force VGA)
#hdmi_group=1
#hdmi_mode=1
# uncomment to force a HDMI mode rather than DVI. This can make audio work in
# DMT (computer monitor) modes
#hdmi_drive=2
# uncomment to increase signal to HDMI, if you have interference, blanking, or
# no display
#config_hdmi_boost=4
# uncomment for composite PAL
#sdtv_mode=2
#uncomment to overclock the arm. 700 MHz is the default.
#arm_freq=800
# Uncomment some or all of these to enable the optional hardware interfaces
dtparam=i2c_arm=on
#dtparam=i2s=on
dtparam=spi=on
# Uncomment this to enable infrared communication.
#dtoverlay=gpio-ir,gpio_pin=17
#dtoverlay=gpio-ir-tx,gpio_pin=18
# Additional overlays and parameters are documented /boot/overlays/README
# Enable audio (loads snd_bcm2835)
dtparam=audio=on
# Automatically load overlays for detected cameras
start_x=1
# Automatically load overlays for detected DSI displays
#display_auto_detect=1
# Enable DRM VC4 V3D driver
#dtoverlay=vc4-kms-v3d
max_framebuffers=2
# Disable compensation for displays with overscan
disable_overscan=1
[cm4]
# Enable host mode on the 2711 built-in XHCI USB controller.
# This line should be removed if the legacy DWC2 controller is required
# (e.g. for USB device mode) or if USB support is not required.
otg_mode=1
[all]
[pi4]
#dtoverlay=vc4-fkms-v3d
# Run as fast as firmware / board allows
arm_boost=1
dtoverlay=vc4-kms-dsi-7inch
[all]
enable_uart=1
dtparam=i2c_arm_baudrate=100000
gpu_mem=128
e.g. removed #dtoverlay=vc4-fkms-v3d and added dtoverlay=vc4-kms-dsi-7inch
crude blaze
ah might be because the kms driver is not enabled, if you uncomment "#dtoverlay=vc4-kms-v3d" that might make it work already
meager fog
yeah i tried that but will do again
meager fog
#dtoverlay=vc4-fkms-v3d
# Run as fast as firmware / board allows
arm_boost=1
dtoverlay=vc4-kms-dsi-7inch
[all]
enable_uart=1
dtparam=i2c_arm_baudrate=100000
gpu_mem=128
root@devpi:/home/pi/Adafruit_CircuitPython_ICN6211# nano /boot/config.txt
root@devpi:/home/pi/Adafruit_CircuitPython_ICN6211# tail /boot/config.txt
# Run as fast as firmware / board allows
arm_boost=1
dtoverlay=vc4-kms-dsi-7inch
[all]
enable_uart=1
dtparam=i2c_arm_baudrate=100000
gpu_mem=128
otherwise I had not tested the official 7inch overlay in a recent OS version, lemme see how that plays out on my end if I use that
meager fog
i probably am using the very latest
crude blaze
in /boot/cmdline.txt I added loglevel=6 drm.debug=0xfwhich enables error logs for DSI on syslog
meager fog
ok let me do that now
ill reboot now, should i be looking at uart console or can i just check dmesg
i have a headless pi4 i ssh into
crude blaze
should show up in dmesg
manic glacierBOT
meager fog
hmm
crude blaze
ok the official 7inch overlay doesn't work for me either, then maybe try the overlay I had attached in the zip, compile and copy instructions in the .sh file
meager fog
ok
thats dmesg
ill swap the dto
i did have compilation warnings...
ICN6211.dts:58.5-15: Warning (reg_format): /fragment@1/__overlay__/panel_disp1@0:reg: property has invalid length (4 bytes) (#address-cells == 2, #size-cells == 1)
ICN6211.dts:71.5-15: Warning (reg_format): /fragment@1/__overlay__/reg_bridge@0:reg: property has invalid length (4 bytes) (#address-cells == 2, #size-cells == 1)
ICN6211.dtbo: Warning (pci_device_reg): Failed prerequisite 'reg_format'
ICN6211.dtbo: Warning (pci_device_bus_num): Failed prerequisite 'reg_format'
ICN6211.dtbo: Warning (simple_bus_reg): Failed prerequisite 'reg_format'
ICN6211.dtbo: Warning (i2c_bus_reg): Failed prerequisite 'reg_format'
ICN6211.dtbo: Warning (spi_bus_reg): Failed prerequisite 'reg_format'
ICN6211.dts:57.31-68.6: Warning (avoid_default_addr_size): /fragment@1/__overlay__/panel_disp1@0: Relying on default #address-cells value
ICN6211.dts:57.31-68.6: Warning (avoid_default_addr_size): /fragment@1/__overlay__/panel_disp1@0: Relying on default #size-cells value
ICN6211.dts:70.29-77.6: Warning (avoid_default_addr_size): /fragment@1/__overlay__/reg_bridge@0: Relying on default #address-cells value
ICN6211.dts:70.29-77.6: Warning (avoid_default_addr_size): /fragment@1/__overlay__/reg_bridge@0: Relying on default #size-cells value
ICN6211.dtbo: Warning (avoid_unnecessary_addr_size): Failed prerequisite 'avoid_default_addr_size'
ICN6211.dtbo: Warning (unique_unit_address): Failed prerequisite 'avoid_default_addr_size'
dtoverlay=vc4-fkms-v3d
# Run as fast as firmware / board allows
arm_boost=1
dtoverlay=ICN6211
ah yea but those are inherent to the PI's DT setup afaik I always get those with every DT file
as long as its just the warnings it should work
tulip sleet
@crude blaze @ladyada could you start a thread, since there is a lot of text here?
thanks
cunning orchid
Waiting on an unexpected extra layer bureaucracy for my approval. Should I refer to the Milestones in GitHub for roadmap? Or is there a dedicated roadmap somewhere?
slender iron
Waiting on an unexpected extra layer bureaucracy for my approval. Should I refer...
(not sure what approval you are referring to) Milestones in github are used for adafruit-funded prioritization. There isn't a separate roadmap.
we discuss it during the weekly meetings usually
we're just about to switch to 9.0.0 dev once the micropython merge bumps the mpy version number
we're going to update to esp-idf 5.0 then too
cunning orchid
(not sure what approval you are referring to) Milestones in github are used for ...
I'm waiting on official approval to contribute from my day job's Legal team (software engineering, so this is necessary for me to contribute). While I wait, I'm trying to plot out issues to prioritize and get my head wrapped around the intended direction, etc.
Good stuff.
slender iron
Ok, not beauracy I can help with then. 🙂
@cunning orchid any specific platforms or topics that interest you?
cunning orchid
<@478338815162056714> any specific platforms or topics that interest you?
I'm intending to dig into multicore support at some point but, think it will be better to start elsewhere. I've a few MCUs at my disposal (ESP32-S3, SAMD21, RP2040), and, an Arty Z7 FPGA (not well-versed in its configuration yet) so, should be able to test on a good chunk of archs.
Mainly, I'm looking to work on lower-level things as I've done a lot of Python and Go in my day jobs and really want to get closer to the hardware again and have been digging into communication protocol implementations, RISC-V, and FPGAs lately in my free time. Not ruling out porting Python code though.
manic glacierBOT
slender iron
I have a long term wish to support the zynqs.
I've also started a "I wish my SoC did X" wishlist
exposing the second RP2040 core for C code could be doable
stuck elbow
my list of "I should do it for circuitpython" things is now so long, that I just stopped adding to it
maybe when I'm retired...
slender iron
😄
cunning orchid
exposing the second RP2040 core for C code could be doable
After previous discussions, I've been contemplating approaches that would be applicable to RP2040s and other multicore. Definitely think that running low-level code makes the most sense (ex. Reserve x memory space for inter-proc comms, reserve y memory space for CPU1 processing, and run this.binary on CPU1). Would be fun to have the type of binary fairly agnostic, as long as it's compiled for the correct arch.
stuck elbow
I would love to see displayio running on the second core
slender iron
I would love to see displayio running on the second core
that'd only work if you aren't doing dvi on the second core
stuck elbow
of course
devout jolt
even a very dumb mechanism that loaded a .bin on boot into the second core and provided a small amount of RAM "scribble space" to talk between C-core and CirPy could be useful for a huge number of applications
slender iron
that's essentially what dvi and usb host do. no need to do it at boot though
cunning orchid
That's exactly what I'm thinking. For safety, I think it makes most sense to have two dedicated regions (one for said "scribble space" and one as a dedicated communication buffer). Not sure that safety could or should be enforced for the primary proc though.
slender iron
the cpu does have an MPU you can use to crash instead of allowing a memory write
cunning orchid
Oooh. Interesting.
While graceful failure seems preferable in many cases, that could be a good route.
slender iron
dvi and usb host use it to prevent rogue flash reads
cunning orchid
If that's the case, I'll definitely dig further into that as reusing the established approach makes most sense.
manic glacierBOT
slender iron
FYI I’m off and on today because Ari is home with a rash
cunning orchid
@slender iron Kinda curious from yesterday's chat. What direction were you thinking for Zynq ports? Blinka? Bare metal on hardcores? RTOS on hardcores? Some sort of RISC-V softcore?
slender iron
<@252717193496756235> Kinda curious from yesterday's chat. What direction were y...
Bare metal on hardcores is what I was thinking. It has hard usb as well. Then you could have custom peripherals in the fpga fabric that are managed by CP>
risc-v softcore would be cool eventually too. orangecrabs (w/ecp5) are probably best for full soft soc
CP runs bare metal on raspberry pi's w/cortex-a already
rancid pilot
Is there a way to print messages to the REPL from common-hal code?
Also I think the documentation is wrong here (not sure where to share it): https://docs.circuitpython.org/en/latest/shared-bindings/canio/index.html
The example says:
can = canio.CAN(board.CAN_RX, board.CAN_TX
But based on common-hal coding and also the documentation down below says:
class canio.CAN(tx: microcontroller.Pin, rx: microcontroller.Pin,
So the RX and TX pins are wrong in the example. Took me a couple of hours to realize what was wrong.
spare jacinth
Is there a way to print messages to the REPL from common-hal code?
No idea bout the name, but "print" has to be mapped to a C function on shared-bindings (or perhaps somewhere else)
You can probably call that one (or another func used inside that one) from your code
rancid pilot
thanks
mental nexus
tulip sleet
@onyx hinge does the MPy PIO impl also have a zillion arguments? or is more implied by the the "native" code?
... not quite as many: https://docs.micropython.org/en/latest/library/rp2.html#rp2.asm_pio
onyx hinge
@tulip sleet it seems to have fewer https://docs.micropython.org/en/latest/library/rp2.StateMachine.html#rp2.StateMachine.init
they're also spread out a bit differently it seems
tulip sleet
yah
onyx hinge
but for instance they omit the count of in/out/set pins in that signature, and leave it up to the coder to just make sure there's no error
it looks like there may be no fifo-joining, not sure
tulip sleet
i like our hand-holding better
meager fog
@crude blaze hihi im not sure if you're around again
crude blaze
@meager fog I am, currently editing some photos but have some time in a bit
manic glacierBOT
meager fog
@crude blaze np tag me when yr done! (if you wanna do another attack run)
lone sandalBOT
unreal wadi
Hi, I saw that this was merged lately: https://github.com/adafruit/circuitpython/pull/8155 Is there any example how to read a USB keyboard that's connected to a RP2040 board (RP Pico) ? (custom RX/TX pins)
unreal wadi
I have installed CircuitPython 8.2.1 and if I try this, it doesn't work:
spare jacinth
Did you compile/flash a fw with the feature enabled?
unreal wadi
No, just downloaded the "official" one. How do it do this?
spare jacinth
Dunno if the compiled is somewhere to be downloaded...
You could setup your machine to compile yourself, but that's trickier
Given it was merged on the main branch, i would expect the compiled to have this new code available
Did you download in the past few days?
unreal wadi
ah, ok ... found this one here: https://learn.adafruit.com/building-circuitpython/build-circuitpython
So at the end, it should be:
cd ports/raspberrypi
make BOARD=raspberry_pi_pico CIRCUITPY_USB_HOST=1
Am I right?
I downloaded it just now, but it seems not to be enabled. But I don't really know
spare jacinth
You can also look for images on the storage (S3 bucket) where adafruit stores them, no idea whether there will smthing there with that enabled
Command looks good to me, but not that im an expert
unreal wadi
I just installed this one here: adafruit-circuitpython-raspberry_pi_pico-en_US-20230728-75ba17e.uf2
There it seems to be enabled! Thanks for your support! 🖖🏻
spare jacinth
Cool 🤓
unreal wadi
hm, but seems not to work ... or I'm doing smth wrong 😅 I tried this here: https://github.com/adafruit/circuitpython/blob/main/tests/circuitpython-manual/usb/device_info.py (slightly changed)
But it will not recognize my USB keyboard. Also tried to swap GP4 / GP5 ... don't know. Maybe I'm to early trying this out. 😄
unreal wadi
hm but seems not to work or I m doing
rancid pilot
Is it possible that I2S audio is broken on the Teensy MicroMod board?
tulip sleet
Is it possible that I2S audio is broken on the Teensy MicroMod board?
We relied on a community member for that port; we don't have one to test.
you could test I2S on a regular Teensy as well
rancid pilot
the wrong pin is listed in the pin config file
i think
so you cannot init the audiobusio 😦
tulip sleet
i"d compare it with the equivalent regular Teensy. Are you able to make your own build to test? We would welcome a PR.
rancid pilot
I can build, and i can test. But i don't know how to make PRs.
tulip sleet
https://learn.adafruit.com/contribute-to-circuitpython-with-git-and-github. But if this is a one-character ( ! sounds like), you can use the web interface in github to edit the file and make a PR.
rancid pilot
This file is generated by a python script, so even if i'd edit the next python file update would just undo my changes?
python3 /home/qute/gen_peripherals_data.py /home/qute/mcux-soc-svd /home/qute/Downloads/ MIMXRT1062
this would overwrite it
onyx hinge
@rancid pilot @slender iron would know more but from what I recall, the chip maker does not permit distribution of the "svd files" that this script uses as input, which is why the results are committed instead of the original files.
tulip sleet
does the same error appear in the Teensy 4.0 or 4.1 pins.c file?
and if it's the same, does I2S not work on those boards?
I wonder if this is some kind of pin-mapping confusion on the micromod carrier board
onyx hinge
it is hard to tell from the screenshots @rancid pilot pasted (please use text when possible!) but I assumed the content they were talking about was in a file like ports/mimxrt10xx/peripherals/mimxrt10xx/MIMXRT1062/pins.c which would depend on the mcu chip, not a circuitpython board definition.
tulip sleet
i.e. look at all the 1062 boards
got it; the q would still apply about other 1062 boards.
if the base 1062 pin map is in error that would carry over
rancid pilot
I could test, I have all possible teensy boards, but I feel like I should just use smtg else that works 😦
tulip sleet
well, I think it would be a quick test just to see if you could create an I2SOut on a Teensy 4.x using the same pins.
it sounds like the same issue would occur
onyx hinge
i.MX RT1060 Processor Reference Manual, Rev. 2, 12/2019 page 313 lists GPIO_AD_B1_14 as a possible alternative for SAI1_TX_BCLK and does not list GPIO_AD_B1_11.
and the generated periph.c seems to match this
tulip sleet
from the board pins.c:
{MP_OBJ_NEW_QSTR(MP_QSTR_MM_I2S_CLK), MP_ROM_PTR(&pin_GPIO_AD_B1_11)}, // D21
AD_B1_11 can be used as SAI1_RX_BCLK
tulip sleet
The teensy41 board pins.c does not list I2S-specific pins
tulip sleet
@onyx hinge so are you saying the generated pins.c (not the board pins.c) does appear wrong? Perhaps there was an SVD error and we should update the submodule
tulip sleet
or maybe there is a pin validation error in common-ah
onyx hinge
At least with the implementation in circuitpython, a "TX" clock has to be used to transmit audio data via i2s
there may be some register level configuration that lets an "RX" clock be used to transmit data but if so CircuitPython doesn't implement it
tulip sleet
const mcu_periph_obj_t *bclk_periph = find_pin_function(mcu_i2s_tx_bclk_list, bit_clock, &instance, MP_QSTR_bit_clock);
in I2SOut.c for mimxrt
rancid pilot
D4 is accepted (BCLK2) in the diagram
but the MicroMod has designated pins, and silkscreened I2S pins, which are wrong :/
onyx hinge
that diagram might represent what works in some other environment
in circuitpython, the pin has to be a "TX BCLK", but AD_B1_11 is an "RX BCLK".
rancid pilot
Thanks for the help. I'll figure smtg out. Maybe there is an Adafruit board with i2s and can bus with better circuitpython support.
onyx hinge
I don't think we've implemented canio on any mimxrt10xx microcontrollers yet
rancid pilot
I just did 🙂 But something is wrong with my CAN transciver chip. I think. Bad chinese fake or...
tulip sleet
This is what MicroPython allows for I2S pins: SCK are allowed clock pins:
#define I2S_GPIO_MAP \
{ \
I2S_GPIO(1, SCK, RX, GPIO_AD_B1_11, IOMUXC_GPIO_AD_B1_11_SAI1_RX_BCLK), /* pin 21 */ \
I2S_GPIO(1, WS, RX, GPIO_AD_B1_10, IOMUXC_GPIO_AD_B1_10_SAI1_RX_SYNC), /* pin 20 */ \
I2S_GPIO(1, SD, RX, GPIO_AD_B1_12, IOMUXC_GPIO_AD_B1_12_SAI1_RX_DATA00), /* pin 38 */ \
I2S_GPIO(1, SD, RX, GPIO_B1_00, IOMUXC_GPIO_B1_00_SAI1_RX_DATA00), /* pin 8 */ \
I2S_GPIO(1, SCK, TX, GPIO_AD_B1_14, IOMUXC_GPIO_AD_B1_14_SAI1_TX_BCLK), /* pin 26 */ \
I2S_GPIO(1, SCK, TX, GPIO_B1_02, IOMUXC_GPIO_B1_02_SAI1_TX_BCLK), /* pin 36 */ \
I2S_GPIO(1, WS, TX, GPIO_AD_B1_15, IOMUXC_GPIO_AD_B1_15_SAI1_TX_SYNC), /* pin 27 */ \
I2S_GPIO(1, WS, TX, GPIO_B1_03, IOMUXC_GPIO_B1_03_SAI1_TX_SYNC), /* pin 37 */ \
I2S_GPIO(1, SD, TX, GPIO_AD_B1_13, IOMUXC_GPIO_AD_B1_13_SAI1_TX_DATA00), /* pin 39 */ \
I2S_GPIO(1, SD, TX, GPIO_B1_01, IOMUXC_GPIO_B1_01_SAI1_TX_DATA00), /* pin 7 */ \
I2S_GPIO(1, MCK, TX, GPIO_AD_B1_09, IOMUXC_GPIO_AD_B1_09_SAI1_MCLK), /* pin 23 */ \
I2S_GPIO(2, SCK, TX, GPIO_EMC_06, IOMUXC_GPIO_EMC_06_SAI2_TX_BCLK), /* pin 4 */ \
I2S_GPIO(2, WS, TX, GPIO_EMC_05, IOMUXC_GPIO_EMC_05_SAI2_TX_SYNC), /* pin 3 */ \
I2S_GPIO(2, SD, TX, GPIO_EMC_04, IOMUXC_GPIO_EMC_04_SAI2_TX_DATA), /* pin 2 */ \
I2S_GPIO(2, MCK, TX, GPIO_EMC_07, IOMUXC_GPIO_EMC_07_SAI2_MCLK) /* pin 33 */ \
}
but does it allow using an RX clock when transmitting audio?
every source agrees AD_B1_11 is a transmitting clock and AD_B1_14 is a receiving clock, from what I saw.
tulip sleet
the MicroPython code appears not to care
hmm, it might check futher down
@rancid pilot ESP32-S2 and -S3 support canio. I think someone else contributed that port, or did you do it @onyx hinge?
onyx hinge
saiConfig.syncMode = kSAI_ModeAsync;
SAI_RxSetConfig(self->i2s_inst, &saiConfig);
saiConfig.bitClock.bclkSrcSwap = true;
saiConfig.syncMode = kSAI_ModeSync;
SAI_TxSetConfig(self->i2s_inst, &saiConfig);
```yeah it looks like there's some register setting that allows using an RX clock when transmitting data. We'd be happy to see this implemented in circuitpython, but as it's not needed for adafruit hardware we probably wouldn't prioritize it.is the pin you want accessible on the micromod carrier?
rancid pilot
Mine is here. I don't know how to make a PR and when testing and doing logic analyser checks the signals appear to stop at me CAN bus transcivere IC so I can't test. : https://github.com/adafruit/circuitpython/issues/8151
the I2S pins are on the top
tulip sleet
yeah but if the pin you want is available elsewhere then you don't have to use the labeled ones
rancid pilot
🙂
onyx hinge
I went ahead and entered an issue about supporting RX BCLK but it's not likely to be a priority-- community help to implement it would be appreciated. ^
meager fog
@crude blaze ok heres the latest draft for rev B - i am gonna at the TSC2007 but im kinda wondering now how does the raspi 7" handle CTP IRQ... (?) kinda wish they'd added another GPIO or two to the DSI connector 😕
slender iron
<@182615154557583360> <@252717193496756235> would know more but from what I reca...
SVD files you can distribute but there are other descriptive files that you can't distribute. You can get them for your own use though.
rancid pilot
I am sorry to ask so many questions. Is I2CTarget (aka I2C peripheral, aka I2C slave) in a working condition? I am using board "Adafruit QT Py ESP32-S3 No PSRAM" as the slave. I am waiting for requests. Nothing is connected at all and I keep getting a lot of requests. Like circuitpython is spamming itself with non-existent requests. When I connect a master and do an I2C scan I can see that the slave is alive and responding. When I try to read from the slave I get garbage as response.
slender iron
I am sorry to ask so many questions. Is I2CTarget (aka I2C peripheral, aka I2C s...
We don't often use it so it could be broken.
meager fog
timonsku ok heres the latest draft for
spare jacinth
looking into #8178 (beware im not too familiar with CircuitPy architecture)
im seeing that main.c has a function named run_code_py (assuming this is the early code that triggers executing code.py)....which in turn calls common_hal_release_displays, but only inside #if CIRCUITPY_ALARM (plus some other C-checks inside this pre-processor-check)
if i understand correctly, "alarm" is used to go into sleep mode until some condition is met and then re-execute the whole code.py file
ideas so far
- small addition: shouldn't harm calling
releaseon an#elseblock... i can see most ports have the feature enabled by default, tho - should i dig further to understand the entire function and try to automatically release on very first pass at the code(maybe just add a
static bool) and/or some other states, or does that sound bad?
tulip sleet
looking into #8178 (beware im not too familiar with CircuitPy architecture)
im ...
In general, we don't want to automatically release: the display is kept alive across VM instantiations so one can see errors. The idea of #8178 is to give a better error if the user code wants to release the display, but not to always release the display.
spare jacinth
Ok thx, will take a closer look, didnt see any code that would detect unreleased on a quick glance, but i can probably find a way 🤓
tulip sleet
The error happens because someone is trying to create a display, but there already is one (hence "too many display busses"). The error is uninformative, because the user is not necessarily trying to create a second display, but simply to use one. If you read the thread referenced in that issue, I think it describes the error case pretty well, esp see Neradoc's reply
I was just thinking of changing the text of the error
spare jacinth
But my quick glance lead to the understanding that this error is the same as actually (trying) creating too many displays, ie: no way to tell both events apart
So, straight up changing the error message wouldnt be nice for the other event
tulip sleet
I think you can't determine the user's intent at the time, we can just offer an extra hint, making the text longer
paraphrasing: "too many display buses -- did you forget to release the current display?". That's a bit too long, but that's the idea
spare jacinth
alright, yeah, makes sense
spare jacinth
I think you can't determine the user's intent at the time, we can just offer an ...
Good enough? "Too many displays. Forgot displayio.release_displays() ?"
How would i go about updating(spanish) and/or invalidating(other langs) translations?
tulip sleet
Good enough? `"Too many displays. Forgot displayio.release_displays() ?"`
How ...
It happens automatically.
let me look at the the msg...
tulip sleet
Good enough? `"Too many displays. Forgot displayio.release_displays() ?"`
How ...
The specific error is Too many display busses, so Too many display busses; need display.release_displays() ? or '; forgot`. Either is probably fine. What would you say in Spanish?
spare jacinth
My doubt right now is... Is this even related to busses (hardware lines), or amount of display "objects"(probably RAM concerns)? I mean, error is not bad but maybe isnt 100% technically correct (plus a user may not know, and doesnt need to know, what a bus is)
Thats the only reason i removed "busses"
Is it hardcoded on the translations or whatever, making it hard to change the text?
tulip sleet
There is already a "too many displays" error, which gets raised under different cirucmstances. The error specifically happens when you try to create a new display bus in the process of creating a new display.
spare jacinth
Oh... guess i didnt look far enough in the chain of functions that get executed to get down here 🙃
tulip sleet
you can search for "Too many displays" and "Too many display busses"; i think you did already
the error is changed in the .c code, not in the .pot file, which is generated automatically. But I think I am telling you something you know already.
spare jacinth
yeah, i figured that out, some magic going on to find strings and re-gen those
tulip sleet
yeah, we look for translate(...)
spare jacinth
(but i was looking at the code without "busses" on it, how dumb am i hahahaha)
tulip sleet
the discord link in the issue has some specific failing code in it
spare jacinth
Cant get pre-commit to like my changes, tried both make translate(as suggested by error msg) and make msgmerge(after digging Makefile), but still not going through
spare jacinth
Translations..........................................................Failed
- hook id: translations
- files were modified by this hook
find extmod lib main.c ports/atmel-samd ports/cxd56 ports/espressif ports/mimxrt10xx ports/nrf ports/raspberrypi ports/stm py shared-bindings shared-module supervisor -type d \( -path "ports/*/build-*" -o -path "ports/*/build" -o -path ports/atmel-samd/asf4 -o -path ports/cxd56/spresense-exported-sdk -o -path ports/espressif/esp-idf -o -path ports/mimxrt10xx/sdk -o -path ports/raspberrypi/sdk -o -path ports/stm/st_driver -o -path lib/tinyusb -o -path lib/lwip -o -path extmod/ulab/circuitpython -o -path extmod/ulab/micropython \) -prune -o -type f \( -iname "*.c" -o -iname "*.h" \) -print | (LC_ALL=C sort) | xgettext -f- -L C -s --add-location=file --keyword=translate --keyword=MP_ERROR_TEXT -o circuitpython.pot.tmp -p locale
python3 tools/check_translations.py locale/circuitpython.pot.tmp locale/circuitpython.pot; status=$?; rm -f locale/circuitpython.pot.tmp; exit $status
Checking locale/circuitpython.pot
Missing message id. Please run `make translate`
{'Too many display busses; forgot display.release_displays() ?'}
make: *** [Makefile:256: check-translate] Error 255
find extmod lib main.c ports/atmel-samd ports/cxd56 ports/espressif ports/mimxrt10xx ports/nrf ports/raspberrypi ports/stm py shared-bindings shared-module supervisor -type d \( -path "ports/*/build-*" -o -path "ports/*/build" -o -path ports/atmel-samd/asf4 -o -path ports/cxd56/spresense-exported-sdk -o -path ports/espressif/esp-idf -o -path ports/mimxrt10xx/sdk -o -path ports/raspberrypi/sdk -o -path ports/stm/st_driver -o -path lib/tinyusb -o -path lib/lwip -o -path extmod/ulab/circuitpython -o -path extmod/ulab/micropython \) -prune -o -type f \( -iname "*.c" -o -iname "*.h" \) -print | (LC_ALL=C sort) | xgettext -f- -L C -s --add-location=file --keyword=translate --keyword=MP_ERROR_TEXT -o - | sed -e '/"POT-Creation-Date: /d' > locale/circuitpython.pot
what happens when you run make translate?
spare jacinth
wait, am i dumb and im just missing a git commit locale?
tulip sleet
[my possible suggestion above is not feasible, never mind, has to do with using qstrs in message]
yes, you have to add the modified circuitpython.pot to the commit
spare jacinth
will also add to the PR an extra output line to the command making the suggestions (if i find where it comes from found!) to also tell about the file to be committed, thx
tulip sleet
@stuck elbow thanks for making this sweep; do you have privs to remove the labels in libraries and adafruit/circuitpython (or only the latter)? I can make executive decisions for thecircircuitpython repo
stuck elbow
<@351136009922019331> thanks for making this sweep; do you have privs to remove ...
I do, but I don't feel like I have the authority to make the decision in most cases
tulip sleet
i will trail behind and remove; i agree with all the choices so far
stuck elbow
there was only a handful, so I'm pretty much done in the main repo
the libraries don't seem as bad
spare jacinth
i messed up @tulip sleet 😅
display vs displayio
tulip sleet
np, i'll approve another one; I should have seen that
spare jacinth
omw
tulip sleet
I will stop the builds so as not to waste resources
spare jacinth
translated latest (11) texts to spanish on weblate, are they automatically added to the PR or do i have to click some button that i didnt see?
onyx hinge
@spare jacinth weblate I think uses a scheduled task to make a new PR; you don't have to do anything.
spare jacinth
guess i will need to compile a new board config, so asking here instead of #help
i have bought and am trying to setup a Luatos ESP32C3, however the code for it (https://circuitpython.org/board/luatos_core_esp32c3/) is for a prior version without UART/USB converter
looks like both manually running esptool, and using the web tool have actually flashed the firmware, and i even got the popup on the webtool to configure credentials (tho it didnt close, perhaps due to errors?)
right now i cant get to it by neither mu/thonny/putty (REPL) nor web workflow (files?)
is this a missing setting (letting CircuitPy about the UART/USB) or just part of C3 port not being completely done?
buoyant lagoon
../../shared-module/displayio/Bitmap.c:219:28: error: comparison of integer expressions of different signedness: ‘uint32_t’ {aka ‘long unsigned int’} and ‘int’ [-Werror=sign-compare]
219 | for (uint32_t i = 0; i < self->stride * self->height; i++) {
and now i'm down to just compiler warnings, my distro is fairly strict by default
39 uint16_t height;
41 uint16_t stride; // uint32_t's
but the error doesnt make much sense, its already unsigned!?
spare jacinth
u16 * u16 wont fit on another u16, seems like its getting casted to i32(int) automatically
what if you: for (uint32_t i = 0; i < (uint32_t)self->stride * self->height; i++) {
buoyant lagoon
i would say thats a bug, it should have become a u32
i32 is also smaller then u32, so it might still truncate
[nix-shell:~/apps/rpi/circuitpython/ports/raspberrypi]$ make -j 16 BOARD=raspberry_pi_pico
Use make V=1, make V=2 or set BUILD_VERBOSE similarly in your environment to increase build verbosity.
QSTR not updated
/nix/store/7bir7rgpr9i95ky49cxpg9mk1x28il24-arm-none-eabi-binutils-2.35.1/bin/arm-none-eabi-ld: warning: -z relro ignored
/nix/store/7bir7rgpr9i95ky49cxpg9mk1x28il24-arm-none-eabi-binutils-2.35.1/bin/arm-none-eabi-ld: cannot find -lc_nano
collect2: error: ld returned 1 exit status
make: *** [Makefile:456: build-raspberry_pi_pico/firmware.elf] Error 1
@spare jacinth the cast does fix the compile error, now its on to a link error!
spare jacinth
i32 is also smaller then u32, so it might still truncate
i have no idea what the "rules" are for type promotions inside the compiler's logic...
there's likely a good reason to default to i32 instead of u32, anyway i it works it works 😛
buoyant lagoon
@tulip sleet any idea about this new link error?
tulip sleet
<@329766224093249548> any idea about this new link error?
we recommend you download the ARM toolchain specified in the guide rather than use what your distro provides. I will look at it.
tulip sleet
I don't know, but I don't think so -- the binaries are built to work on a lot of different platforms
buoyant lagoon
the toolchain ive loaded, is known to work on the pico
tulip sleet
what version of gcc are you using
buoyant lagoon
$ arm-none-eabi-gcc --version
arm-none-eabi-gcc (GCC) 10.2.0
i cant find c_nano in the source anywhere
tulip sleet
the toolchain ive loaded, is known to work on the pico
that is similar to what we ask you to use here: https://learn.adafruit.com/building-circuitpython/linux#install-build-tools-on-ubuntu-2986713 See the "ARM Cortex-M toolchain" link
buoyant lagoon
../../ports/mimxrt10xx/sdk/tools/cmake_toolchain_files/mcux_config.cmake: set(TARGET_LINK_SYSTEM_LIBRARIES "-lgcc -lc_nano -lm" PARENT_SCOPE)
only in another port, and git grep is being stupid and wont show it
tulip sleet
we use ag a lot for searching
this does build without errors on the standard setup. I can't explain why yours is different.
buoyant lagoon
do you know where c_nano.a is normally coming from?
tulip sleet
do you know where `c_nano.a` is normally coming from?
it's in the toolchain we're using: gcc-arm-none-eabi-10-2020-q4-major/arm-none-eabi/lib/libc_nano.a
buoyant lagoon
ahhh
my toolchain comes with newlib, but most of the things i build provide their own libc
simplest option then, docker!
that will fix my missing /lib and let me just apt-get install whatever the directions say
tulip sleet
I just unpack the toolchain and put it on my PATH. It does not require a system install
tulip sleet
See the link above and read the paragaph for explanation
Most people use Ubuntu 22.04. Scott (tannewt) uses arch
but the guide is written for 22.04
buoyant lagoon
[clever@amd-nixos:~/apps/rpi/circuitpython]$ gcc-arm-10.3-2021.07-x86_64-aarch64-none-elf/bin/aarch64-none-elf-gcc
bash: gcc-arm-10.3-2021.07-x86_64-aarch64-none-elf/bin/aarch64-none-elf-gcc: cannot execute: required file not found
yep, fail, just like expected
it expects /lib64/ld-linux-x86-64.so.2 to exist
[clever@amd-nixos:~/apps/rpi/circuitpython]$ ls /lib
ls: cannot access '/lib': No such file or directory
[clever@amd-nixos:~/apps/rpi/circuitpython]$ ls /lib64
ls: cannot access '/lib64': No such file or directory
moved from other channel:
make remove-all-submodules at the top level will remove them all cleanly.
small comet
I took a look at https://github.com/adafruit/circuitpython/issues/8119
The naive fix seemed to be to add a reference to the circuitpython_typing online repo at:
https://github.com/adafruit/circuitpython/blob/8050b07a687277cb277e4c13de83cfdc076682d5/conf.py#L444
Unfortunately, there's some long-standing Sphinx issue, which doesn't
support the current mechanism that the adafruit-typing module uses
(https://github.com/sphinx-doc/sphinx/issues/8934)
The symptom: the Sphinx html generator looks for a class definition
for (say) circuitpython_typing.ReadableBuffer but the online repository
tags it as data (https://pastebin.com/21ZjBX3g) so it never gets resolved.
Am I missing a trick here?
onyx hinge
I took a look at https://github.com/adafruit/circuitpython/issues/8119
The naiv...
I don't have an instant answer for that, but that sounds like good research -- can you add what you've learned to the issue as well?
small comet
Sg - fwiw, I saw there was some 'good first issue' cleanup going on, and this one seemed to get into the weeds; so maybe worth reviewing for that tag
spare jacinth
guess i will need to compile a new board config, so asking here instead of #help...
little bump
small comet
@spare jacinth fwiw there's some ongoing upstream idea https://github.com/micropython/micropython/issues/6170 and I'm guessing the CP maintainers may have some thoughts on how best to merge this with that
spare jacinth
good thing is: regardless of how items are inserted, this method should still work (unless the definition of the structure holding the data changes)
and we were lacking it anyway to have better coverage of CPython's API
small comet
Right - I was thinking (because of the new duplication of the method pointers) if it might become more tricky to merge any upstream changes, but (thankfully!) someone more experienced will no doubt figure this out 🙂
@spare jacinth (I also found https://github.com/adafruit/circuitpython/blob/main/tests/basics/ordereddict1.py handy to add new tests for regression, maybe that could be another way to flag any breaking future changes... good luck 🙂 )
spare jacinth
im not sure what that file is used for, tho... other tests seem to have a .exp to match their output, but this one doesnt (?)
small comet
Yea - it's confusing (but better documented in micropython.) The way it seems to work is that for anything without a .exp suffix, it runs cpython first, and compares with micro/circuitpython output
spare jacinth
oh, makes sense, lets see if i can run the test (and check its output)
gloomy cosmos
I have an issue that I posted over at general help circuitpython chaneel, but I doubt anyone is going to be able to help me over there. Would someone mind taking a look at #help-with-circuitpython message. I'll pretty sure there's an issue in the sockpool lib. I can also file an issue on GitHub if that works better for you folks.
small comet
fyi @onyx hinge - not a high priority; I tried to find the least unappealing workaround while I was looking at it.
manic glacierBOT
spare jacinth
TIL, thx again Dan :)
light lichen
Is this a place (channel) for getting assistance adding support for a board? specifically waveshare esp32-s3-pico
Specifically troubleshooting getting it to run on the board after compiling (or configuring correctly to get to run on the board)
lone sandalBOT
manic glacierBOT
small comet
(just a quick sanity check) on the esp32, is it feasible to implement a useful countio without a dedicated pulse counter peripheral? (I was rueing the lack of countio on an esp32c3, and realized it doesn't have such a peripheral.)
onyx hinge
probably would require creative use of some existing peripheral. maybe rmt?
small comet
Will read up more on it - thx for the pointer!
This is an ultra-naive question - is pulseio already built on the rmt peripheral on the ESP32? (So one might creatively do any countio equivalent entirely in python?)
onyx hinge
no, looks like the existing implementation is using PCNT
small comet
no, looks like the existing implementation is using PCNT
Oh, ok! (I was confused by https://github.com/adafruit/circuitpython/blob/55c82f45daf797af7de6d1edf86082a21f9df7d7/ports/espressif/common-hal/pulseio/PulseIn.c#L119 - I'll continue poking around with it - this is mostly for my weekend entertainment, so non-critical for sure. thx again for the tips!)
onyx hinge
I was looking at countio/Counter.c, sorry for any confusion
manic glacierBOT
warm stump
A MiniDisc!? (a line from scooby-doo where velma finds a minidisc on the floor as part of an investigation, can't find it now)
Trying to explain the many features of this contraption and adding to it is the fun part. It's essentially just a disk drive but has otherwise completely redundant physicality details to commentate.
The main slide tray has 2 prod sticks which prod the minidisc out of the slot so it falls on the floor in amusing fashion.
It obviously also has a spindle motor for the actual disc which has decals.
There's a driving ring which looks complex and all it does is moves an axle 1 unit upwards to lock the minidisc in place.
Then it has (and why not) a Pneumatic Laser Arm with no air in it.
```I did consider while adding each feature that if you add a circuitpython device to this (or a motor to each motor and a pump to each pump)
and if the spinning disc were actually made out of actual optical glass & polycarbonate substrate, then it might actually work as a storage device -_(\
digital shoreBOT
adafruit
spare jacinth
there was a website with a list of issues from (all?) repos, but i cant find it... can someone share the link for me to bookmark?
would like to try and fix some 🙂
spare jacinth
ha, my keywords "adafruit contributing" should have been "circuitpython" instead
ended up with links to docs for how to make a PR and the like
thx 😄
onyx hinge
@tulip sleet if you have a bit of time a review of https://github.com/adafruit/circuitpython/pull/8218 would be appreciated! -- it also goes with a related change in a submodule, see comment from last week.
tulip sleet
<@329766224093249548> if you have a bit of time a review of https://github.com/a...
sure! reviewed
onyx hinge
The thing that is not great about remove_prefix is it's not exceptional if the expected prefix isn't present, but it is still a more explicit way to do the slice. @tulip sleet I think I addressed everything except the function names, which I am not sure your suggestions are better than my originals but I'll change it if you have a strong opinion.
tulip sleet
The thing that is not _great_ about remove_prefix is it's not exceptional if the...
I am interested in detecting if the prefix is wrong, since it could change out from under the code. Ohterwise it will fail in some mysterious way. So however is the easiest to do that would be good (replace failure or whatever).
Re names, don't have to change. I was just trying to indicate this is very specific to 0's and 1's, as opposed to using $(filter or something like that. That's why I asked for an example in a comment.
onyx hinge
OK. does a comment referring the interested reader to a use below "cut it", or do you want the example right there?
tulip sleet
OK. does a comment referring the interested reader to a use below "cut it", or d...
the current new comment is great!
onyx hinge
OK now the prefix is checked
tulip sleet
i have to go out for 45 mins or so; will pre-approve
onyx hinge
thanks dan, appreciated as always
slender iron
<@&356864093652516868> The meeting is in an hour and half today. See the pinned message for the notes doc. Talk to you soon!
midnight ember
According to my ISP shouldn’t expect my internet to be back online for a couple hours.
onyx hinge
enjoy your time on the desert island
blissful pollen
Didn't update the notes but just here lurking today while I review exceedingly exciting documents...
onyx hinge
I'm going to assume there may be an element of sarcasm in that statement
blissful pollen
maybe 🙂
onyx hinge
it's pretty good after turning david up to 200% in the right click menu
digital ibex
I haven't messed around with SynthIO yet. I've gotta get on that!
turbid radish
Date is now permanently every Monday morning US Eastern time
idle owl
You can also email circuitpythonday (@) adafruit.com with your own events for CircuitPython Day!
onyx hinge
thank you new & infrequent contributors! ❤️ 
digital ibex
I'd love to help review where I can, can someone help me get set up after the meeting?
onyx hinge
We will want another 8.2.x release soon, due to the Adafruit IO certificate change. There's an open (I think*) PR for it, but not an Issue.
slender iron
I'd love to help review where I can, can someone help me get set up after the me...
Yup!
tulip sleet
We will want another 8.2.x release soon, due to the Adafruit IO certificate chan...
If there is nothing else urgent, I can do this tonight or tomorrow, since it's a blocking issue for many people
onyx hinge
@tulip sleet brent did check out the build and confirmed it can connect to IO. but if the actual certificates in nina-fw changed during your troubleshooting session on friday/the weekend I can update the PR again.
tulip sleet
<@329766224093249548> brent did check out the build and confirmed it can connect...
I am talking to brent about whether that's true or not
royal rover
Is there a list of libraries available for Blinka
digital ibex
Is there a list of libraries available for Blinka
Yup! https://docs.circuitpython.org/projects/bundle/en/latest/drivers.html should be at least some of them
Edit: oops, I read blinka as CircuitPython! Most of them should work though
onyx hinge
@royal rover you can search pypi based on package name https://pypi.org/search/?q=adafruit-circuitpython-&o= though a small number of packages that are not expected to work on blinka/desktop python are published on pypi anyway.
[and based on David's hug reports I second the HR to scottyp humanIthink and tyeth for bringing the issue up over here on the discord; I didn't directly see that part of the process]
thorny jay
Certificate are a pain...
onyx hinge
nice to hear from you @thorny jay !
thorny jay
Micro Dot
onyx hinge
If that's what I think it is, the display is about 0.3" (7.62mm) wide, the same as a classic DIP IC
thank you everyone, have a great week!
idle owl
Thank you, everyone!
onyx hinge
And remember
You can also email circuitpythonday (@) adafruit.com with your own events for CircuitPython Day!
we'd love to have your participation!
digital ibex
Bye everyone! Have a wonderful week!
onyx hinge
August 7
turbid radish
Thanks Scott
lone axle
Thanks for hosting Scott, have a great week everyone!
thorny jay
If that's what I think it is, the display is about 0.3" (7.62mm) wide, the same ...
Like this:
slender iron
thorny jay
The "problem" with (most) Pimoroni product is that they have a Python driver for the Raspberry Pi SBC and a driver (maybe C) for thier own version of MicroPython. When a compatible CP library exist, it is possible to quickly reuse their hardware, else you are on your own.
devout jolt
https://github.com/electro-smith
The Daisy platform is amazing. Can’t wait to listen to the meeting to see why it came up
slender iron
The Daisy platform is amazing. Can’t wait to listen to the meeting to see why it...
it is unrelated to the meeting. I found it via a video rec on the adafruit account as I was doing post-meeting stuff
slender iron
@turbid radish @idle owl I don't need to do versions anymore right? only the meeting links and py dev of the week
devout jolt
it is unrelated to the meeting. I found it via a video rec on the adafruit accou...
Nice. Daisy is used in several commercial products now. They’ve got a great ecosystem if you need real-time audio processing & generation. I just learned recently you can pretty easily load Pd or Max patches from your computer onto a Daisy. That’s a bit mind-boggling
slender iron
I watched the pocket piano 201 video and then was recommended one on chompi. My natural tendancy is to figure out what the guts are. 🙂
@digital ibex I've got time now if you want to chat about reviewing
digital ibex
Ok, should I hop in the voice channel?
slender iron
sure, that may be fastest
crimson ferry
"Clone failed: A submodule points to a commit which does not exist." 😦
slender iron
does it say which one?
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1vJq1BU6TdWuTJVSCUKHbE4F6m4ZoIcEMsjPUotB5LZg/edit?usp=sharing
crimson ferry
it doesn't say which one, trying again in case it was a fluke
slender iron
it can happen if a commit is in a different branch/repo than the one listed in .gitmodules
orchid basinBOT
onyx hinge
https://blog.adafruit.com/2023/07/31/circuitpython-8-2-2-released/
thank you!
manic glacierBOT
manic glacierBOT
midnight ember
I don't know which one of you or all devs who decided not to wipe nvm during a circuit python install but huge hug to you. 🤗 I had an API token stored in NVM, installed CP 8.2.2 and the token was still there and picked up right where it left off. Have no idea how or why that worked but thank you for allowing that possibility to exist. 
tulip sleet
I don't know which one of you or all devs who decided not to wipe nvm during a c...
we never want to destroy people's work. And that's the way the UF2 bootloader worked anyway 🙂
slender iron
(teensy's bootloader wipes all of flash)
midnight ember
@slender iron I don't have any teensy's... would that apply to the iMX?
slender iron
nope, just teensies
midnight ember
awesome 👍
manic glacierBOT
onyx hinge
@stuck elbow are you able to share your streaming mjpeg code with me?
onyx hinge
thank you!
onyx hinge
OK but there's STILL the problem that bitmaptools.blit doesn't appear in the documentation that I can see! https://docs.circuitpython.org/en/latest/shared-bindings/bitmaptools/index.html
aha! the stub needs to be dedented 1 level
no that's not the only problem
onyx hinge
oh the other problem was my own making. yay.
cunning orchid
Still waiting for the bureaucracy here before I can really materially contribute (my boss is ex-Cisco and wants to get all liability ducks in a row and thoroughly documented). Thinking that I may try putting together a little guide for navigating such waters. Maybe subsection of contributing? Or Adafruit learning?
cunning orchid
That's where I was thinking, yeah. There is already some mention of the possibility of needing approval but, a bit sparse on details.
There's been mention of a waiver document of some sort, so, once that's figured out, I'll do a write-up and maybe see if there are documents/guidelines from the FSF to help folks in startups and other companies that aren't necessarily FOSS-averse but FOSS-inexperienced.
slender iron
There's been mention of a waiver document of some sort, so, once that's figured ...
Some open source projects will have you agree to a contributors license agreement (CLA): https://en.wikipedia.org/wiki/Contributor_License_Agreement
cunning orchid
Thanks! I'm wondering as well if you folks have had any thought yet on project governance considerations with the CRA heading for adoption in the EU. (Maybe more of a weekly meeting question).
slender iron
I'm not aware of what the CRA is. (I'm in the US)
cunning orchid
I'm in the US too but it is looking like global open-source may be impacted (intentional in the committee version, pending on negotiation w/ the current rotating executive to finalize). It is the Cyber Resilience Act. Much of it seems rather sensitive but, currently it's looking likely to "left-shift" CE security/process certification liability for open-source software and hardware to the projects themselves, if they are to be available in the EU.
slender iron
I take the THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND portion of the license literally. Any support I give is a bonus.
@onyx hinge I had to delete my .git/modules/ports/espressif/esp32-camera/ folders after deinit. sync doesn't seem to update the tracking branch
cunning orchid
Agreed. That's a pretty central concept in most FOSS licenses. We won't know for certain how it stands until Sept, but the committee draft intent seems to be to explicitly nullify such terms. Here's an Apache Foundation article: https://news.apache.org/foundation/entry/save-open-source-the-impending-tragedy-of-the-cyber-resilience-act
onyx hinge
If/when that becomes law I'm sure Adafruit will chek with lawyers to determine the response. Until then it's not clear why we'd do anything but the status quo.
cunning orchid
If/when that becomes law I'm sure Adafruit will chek with lawyers to determine t...
I've been in a lot of liability and licensing discussions in my day job as of late so, might be a bit hyper-aware at the moment.
onyx hinge
I appreciate you have good intent bringing it to our attention 🙂
spare jacinth
stuff like control+arrow not working on REPL is an issue with app connecting to the serial port (Mu currently, but im pretty sure Thonny did the same), or that's simply not handled by the code on fw?
seems like it is the later and there's a good reason (code complexity?) for it, but if it feels somewhat feasible to implement i could poke at it 🙂
that should probably go towards MicoPy tho, i assume
slender iron
I don't know what control+arrow is meant to do 🙂
spare jacinth
on right/left, move entire words instead of single chars
on up/down, most likely nothing special 😛
slender iron
There is somewhere where we interpret the special vt100 codes
I know the display spot but not the repl one
spare jacinth
funny enough, shift+control+left/right does select entire words, lol
so, the logic is there already, hopefully not too coupled with select-mode
spare jacinth
yeah, definitely some terminal stuff in play too, putty doesnt select on whole words
so... sadly, something with no fix :(
slender iron
I'm sure you could fix it. It just won't be easy 🙂
spare jacinth
support for using key as kwarg added, and also realized i had no KeyError(to match CPython), but output is ```py
d.move_to_end("foo")
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
KeyError:
```copied from the only place i found that raises aKeyError(in my case:if (!found) mp_raise_type(&mp_type_KeyError);)
so.. it is not returning the key (?), how do i set it up? 🙃
stuck elbow
a quick grep shows mp_raise_type_arg(&mp_type_KeyError, field_q);
there is also mp_raise_msg_varg(&mp_type_KeyError, MP_ERROR_TEXT("pop from empty %q"), MP_QSTR_set);
spare jacinth
need to work on searching skills, thx that should be it
stuck elbow
I use ag, aka the silversearcher
spare jacinth
i use Ctrl+Shift+F, aka VSCode's builtin lol
small comet
Still waiting for the bureaucracy here before I can really materially contribute...
(Just my 0.02) depending on each company's culture/lawyercats of course - it may be helpful to start with a set of licences (rather than specific projects). and see if it's possible to pre-approve contributing in one's personal capacity to repos only governed by those licences.
jaunty juniper
@spare jacinth interpreting ctrl-arrow into the propre code is likely a PC-side thing. Mu does things on its side like sending the right amount of "move left/right" when clicking in the "command line".
But there is extended (while still limited) emacs keys support that can be enabled with MICROPY_REPL_EMACS_KEYS and MICROPY_REPL_EMACS_EXTRA_WORDS_MOVE
https://github.com/adafruit/circuitpython/blob/main/shared/readline/readline.c
don't ask me how I ended up in this section of the code a long time ago
spare jacinth
getting this when trying to compile py ../../extmod/ulab/code/numpy/vector.c: In function ‘vector_sinc1’: ../../extmod/ulab/code/numpy/vector.c:787:20: error: conversion from ‘mp_float_t’ {aka ‘double’} to ‘float’ may change value [-Werror=float-conversion] 787 | if (fpclassify(x) == FP_ZERO) { | ^ adding -Wno-float-conversion didnt work as Makefile contains -Wfloat-conversion, does this happen to somebody else?
commenting out on Makefile does workaround, ofc
(WSL with Ubuntu22)
Update: Fun, test gets skipped hahaha
small comet
getting this when trying to compile ```py
../../extmod/ulab/code/numpy/vector.c:...
https://github.com/v923z/micropython-ulab/pull/636 was for a similar issue, wonder if this is a new one or somehow didn't get merged (edited to refer to the fix)
onyx hinge
we probably didn't update ulab again since the fix went in?
slender iron
I'm thinking of making adafruit_usb_host_descriptors and adafruit_usb_host_mass_storage libraries. Any alternatives I should consider?
One is helpers for reading and parsing descriptors, the other is for interfacing over mass storage.
spare jacinth
getting this when trying to compile ```py
../../extmod/ulab/code/numpy/vector.c:...
any idea why it gets ignored? and/or how to prevent it?
seems like --target esp32 and/or building the variant/coverage unix port (did it even get used?) got the test to run, but it wont pass checks because of different format, shall i rather make a long .exp for one of existing folders, or make a new test? well, i can perhaps print the keys instead of whole dict
small comet
(I had the impression a good place to make any changes would be in the upstream repo) but I've also forgotten what I actually poked at around this.
manic glacierBOT
manic glacierBOT
spare jacinth
How complex is that to implement? ⤴️
(No previous experience with PIO)
tulip sleet
How complex is that to implement? ⤴️
(No previous experience with PIO)
You'd need to understnd to understand the feature. I don't think it is a lot of code.
midnight ember
Basic Fitbit API example for Adafruit_Requests library pull request added this morning. https://github.com/adafruit/Adafruit_CircuitPython_Requests/pull/140
onyx hinge
@slender iron @tulip sleet OK I was mistaken in part about ulab. The specific ref that circuitpython pointed at was NOT in the history of their main branch, possibly because the PR was resolved with "squash and merge". Their main branch has "38caf84 Fix -Wunused-variable diagnostic when !ULAB_SUPPORTS_COMPLEX (#631)" while CP was pointing at "< 3728d22 (HEAD) Fix -Wunused-variable diagnostic when !ULAB_SUPPORTS_COMPLEX. So it's interesting that this worked for me & for the CI system but not for others including scott.
tulip sleet
makes sense, do you want to PR to repoint it?
tulip sleet
@onyx hinge there is a pre-commit failure in 8256 ⬆️
onyx hinge
oops, I'll fix it
the "translations" check is not run if only the submodule extmod/ulab is changed, so it wasn't caught locally. pre-commit run --all does though
onyx hinge
ah did an engineer ever fix a problem without making more? 😓
@tulip sleet since it also fixes a build problem (that I introduced 😩), your review on https://github.com/adafruit/circuitpython/pull/8252 would be appreciated.
tulip sleet
<@329766224093249548> since it also fixes a build problem (that I introduced 😩)...
I am behind on PR's 🙂
turbid radish
I'm unable to get the number of users we have on the server, / serverinfo isn't showing it
onyx hinge
?membercount
@turbid radish apparently I don't either
@turbid radish on the discord desktop client, click the "v" icon next to "adafruit" above the listing of channels. then click "server settings" and "members". the screen will show you the current count.
spare jacinth
if you have the rights, seems like there's a "members" tab on server's settings
onyx hinge
"jinx" I think they say.
onyx hinge
sigh, feeling a bit distressed that I apparently messed up the build system pretty good with a change I didn't even think would have an impact on it. well, it'll get put right soon. if I understood the problem now.
gilded cradle
@lone axle I have some questions when you get a chance about bitmaptools and displayio.
lone axle
<@382939733107408897> I have some questions when you get a chance about bitmapto...
I am around for a bit now. Whats up?
lone axle
I'm pretty familiar with some of it's functions but less so with others.
gilded cradle
Well, I've been trying to use the alphablend function and I can sort of get it to work, but because palettes are separate to bitmaps, the colors aren't correct. Any ideas?
lone axle
That is one of the ones I haven't used yet. But my interpretation of whats mentioned in the colorspace docs here: https://docs.circuitpython.org/en/latest/shared-bindings/bitmaptools/index.html#bitmaptools.alphablend
The colorspace of the bitmaps. They must all have the same colorspace. Only the following colorspaces are permitted: L8, RGB565, RGB565_SWAPPED, BGR565 and BGR565_SWAPPED.
means it may not work for indexed bitmaps with different palettes. I think they would need to have the same palette. This mentions using one of those listed colorspaces, but I would expect if you happen to have 2 different indexed bitmaps with the same palette even if it's not specifically one of these full color spaces it could still work with them.
My best idea would be to try non-indexed BMPs and see if it makes a difference, or if possible try to merge the palettes for your two images into one so that they each can use the same palette.
gilded cradle
Ok, well I was pulling the bitmaps in with imageload (which only works with indexed bitmaps) as it doesn't work with OnDiskBitmaps. Any ideas how to pull in an non-indexed bitmap?
lone axle
Hmm, that is a good point. In my head I had mistakenly had that restriction backwards. I was thinking OnDisk was only indexed, and imageload was indexed or rgb.
gilded cradle
Yeah, I think ondisk only works with 24-bit bmps and imageload only works with 8-bit bmps.
8-bit of course means indexed.
lone axle
It looks like when it was added originally the test case was using directio rather than displayio to get it onto the display: https://github.com/adafruit/circuitpython/pull/5544 I wasn't aware of directio module before, though I think I recall something about GIFs using a similar technique.
gilded cradle
Huh, I haven't seen directio. I'll have to look at that
Oh, ok. I think directio is for camera data
He mentioned something like that
lone axle
I am curious about alphablend after poking through the documentation a bit. I'll try out the merged palette approach in the next few days and see if I can get a "simpletest" that works with displayio. If I'm understanding things correctly it would mean that it's restricted to only working on images whose combined palettes occupy less than 256 colors.
gilded cradle
It still may be useful
According to the pr you linked, it should work with RGB565 16-bit images
Also, I may take a look at what gifio might be able to do
Anyway, thanks for chatting. I think you helped. I am interested to check out your simpletest when you get it working.
onyx hinge
alphablend is about combining two input images colors proportionally, like output[0] = 50% * input1[0] + 50% * input2[0]. I'm not sure how that makes sense with palette images, because that RGB value you get by combining two pixels might not be anywhere in the palette.
it's not about combining image according to the transparency information one or the other of those images has
gilded cradle
Yeah. I'm working on a way to get the input bitmaps to be 16-bit in which case, the palettes become the same for the images, which is why it works.
onyx hinge
I don't think it will work for palette images, even if they're 16 bits.
they have to be RGB565 or one of the variations
gilded cradle
Right
My initial strategy is to try and load an ondiskbitmap, convert to (RGB565) 16-bit with color converter and if necessary copy pixel by pixel into a regular Bitmap.
onyx hinge
OK.
gilded cradle
If that works, I can try and optimize it from there
gilded cradle
Fortunately the images needed for these matrices are fairly small.
midnight ember
@turbid radish Are you looking for total members or members currently online?
Online Members
spare jacinth
fixed, or at least improved, the list-parsing bug
would love some feedback for improvement, as the lib is pretty much feature-complete from my end (i dont feel like adding support for dates, line breaks inside lists, nor multiline -aka triple quoted- strings)
slender iron
first time I've seen this: ```
#0 0x00002dea in FLEXSPI_ReadBlocking (base=0x402a8000,
buffer=0x20007a9c "C", size=1) at sdk/drivers/flexspi/fsl_flexspi.c:760
#1 0x00002fc2 in FLEXSPI_TransferBlocking (base=base@entry=0x402a8000,
xfer=xfer@entry=0x20007aa0) at sdk/drivers/flexspi/fsl_flexspi.c:876
#2 0x00002a24 in flexspi_nor_wait_bus_busy (base=base@entry=0x402a8000)
at supervisor/flexspi_nor_flash_ops.c:62
#3 0x00002a86 in flexspi_nor_flash_erase_sector (base=base@entry=0x402a8000,
address=2097152) at supervisor/flexspi_nor_flash_ops.c:103
#4 0x00002924 in port_internal_flash_flush ()
at supervisor/internal_flash.c:133
#5 0x000027c4 in supervisor_flash_flush ()
at ../../supervisor/shared/flash.c:137
#6 0x00002816 in filesystem_flush ()
at ../../supervisor/shared/filesystem.c:172
#7 0x000029ce in reset () at reset.c:34
#8 0x00002696 in reset_cpu () at supervisor/port.c:490
#9 0x00002804 in reset_into_safe_mode (
reason=reason@entry=SAFE_MODE_HARD_FAULT)
at ../../supervisor/shared/safe_mode.c:139
#10 0x000026ec in HardFault_Handler () at supervisor/port.c:636
#11 <signal handler called>
#12 0x00003da0 in tu_fifo_write (f=f@entry=0x20008004,
flash flush from hard fault
onyx hinge
hmmmm maybe for some safe mode resets you'd want to do that but probably not hard fault.
slender iron
higher in the stack it was also reading from flash
no idea how it faulted
wish I had trace going 🙂
it did corrupt my fs
manic glacierBOT
lone sandalBOT
orchid basinBOT
midnight ember
@turbid radish It's much easier to find if using a mobile device than desktop discord app. The UI's are not the same and show more server info on the mobile app.
hold down on the Adafruit channel icon and a menu popout with more options appears
which displays online members and total members
There is also an adafruit_requests example where you can get the info with a microcontroller. 😉 https://github.com/adafruit/Adafruit_CircuitPython_Requests/blob/main/examples/requests_api_discord.py
but that one requires a data scraping method detailed here: https://github.com/lorenz234/Discord-Data-Scraping
clear halo
Damien on Embedded.fm:
https://embedded.fm/episodes/456
random junco
@onyx hinge May I please DM you when you have a moment? I have a question for you regarding CircuitPython Day
onyx hinge
<@426200547024961539> May I please DM you when you have a moment? I have a ques...
@random junco DM away but if it needs a detailed answer I won't get right back to you.
orchid basinBOT
manic glacierBOT
small comet
(just curious) I saw an adafruit_debug_i2c wrapper library but the comparable adafruit_debug_spi library seems to have stopped development. Is there any new thinking around such wrapper libraries, or just lack of time/priority to move that forward?
onyx hinge
@small comet probably mostly lack of time. If offered it sounds like something we'd be happy to add to the community bundle!
small comet
<@1130377638523977739> probably mostly lack of time. If offered it sounds like s...
🙂 thx for the context.
@onyx hinge just confirming, as https://github.com/adafruit/Adafruit_CircuitPython_Debug_SPI seemed intended to be an adafruit bundle library (like the i2c library). Not that I'm throwing my hat in the ring 🙂 but do you think it would it be preferable to make contributions as pull requests there, or do you have in mind having this be a new repo outside the Adafruit organization altogether?
onyx hinge
It looks like no Adafruit person ever actually wrote any runnable code in there, it's all just updates of infrastructural stuff.
If Adafruit is not planning to maintain it, it's better that it go in the community bundle.
Because there's no runnable code I don't think anything's lost by starting fresh.
small comet
Ack - thx for extra context! (Awkwardly, it's more red tape work-wise to moonlight on a new project vs contributing to an existing adafruit one, but that's a whole other story 🤷♂️ )
thorny jay
I don't understand why the Picopad is not available in a stable release: https://circuitpython.org/board/pajenicko_picopad/
I think it was there in time for 8.2.2 (maybe 8.2.1), but that was only for patching the certificate, so nothing was backported.
Or maybe backporting board only occurs on minor version and it will be in 8.3?
proven garnet
@onyx hinge I just submitted a PR against the unified workflow repo for the issue you tagged me in!
onyx hinge
<@426200547024961539> I just submitted a PR against the unified workflow repo fo...
that was fast
any idea why it worked previously, or appeared to?
proven garnet
No, I don't, which is the weird part. I actually would have expected it to fail up until now actually, based on the information you linked
I've started using what I PR'd exclusively for this reason though, and for our purposes makes the most sense anyways. I'll probably do some digging into why we only just hit this issue, or at least didn't originally.
onyx hinge
oh! pip install . installs this library to a place on pythonpath; you removed that in favor of .. other commands.
successful run ```
Successfully installed adafruit-circuitpython-pioasm-0.0.0+auto.0
============================= test session starts ==============================
platform linux -- Python 3.11.4, pytest-7.4.0, pluggy-1.2.0
rootdir: /home/runner/work/Adafruit_CircuitPython_PIOASM/Adafruit_CircuitPython_PIOASM
collected 15 items
tests/test_mov.py ... [ 20%]```
notice how it has just installed the library
[also noted on the issue] @proven garnet
proven garnet
OHHHHH
Confidently wrong about how confident I was about not introducing a breaking change 😂
That makes sense though, thank you for discovering and explaining that!
small comet
@proven garnet (was curious about your PR because I was curious about testing; a driveby fyi - wondering if the pyhron here might be a typo. https://github.com/adafruit/workflows-circuitpython-libs/pull/30/files#diff-5881ec64b771c47aba005f7a992b077a0d509f0acc5671b7576b1eb4714603f5L70
proven garnet
<@135415127879974912> (was curious about your PR because I was curious about tes...
😁 Thanks for catching that!
My personal belief is that python -m pytest is best since it always includes the library you mean to test, though it is interesting to see how the change @onyx hinge mentioned affected it. Makes a lot of sense, and something I should be more mindful of when working with the CI files!
small comet
And I learnt something about how the CI is put together by watching from the peanut gallery, so win-win 🙂
onyx hinge
@proven garnet can you add a comment about why so we don't mess it up again accidentally?
# this includes the current directory in sys.path so that the tests can find the module being tested or similar
proven garnet
`# this includes the current directory in sys.path so that the tests can find th...
Sure thing!
Updated, please let me know if more details is needed
onyx hinge
nah that'll hopefully make someone at least check before changing it.
tests passed in that other PR now. yay. thanks and have a great rest of your weekend.
well, actually, it failed, just later. argh
actions don't work with python 3.9, but that was something I'd changed trying to figure out the problem
plush citrus
any chance I could ask for the new synthio to supply the elementary waveforms out-of-the-box? understand that memory is limited, but something like synthio.sine_wave, synthio.sawtooth, etc. would be quite helpful
plush citrus
also I submitted a PR for NAU7802 to enable polling rate selection
onyx hinge
any chance I could ask for the new `synthio` to supply the elementary waveforms ...
probably not, but someday I hope we'll have a pure-Python library to help generate them. It's just a few lines of code, but horrible to have to copy & paste them everywhere.
thorny jay
I have something to say about CircuitPython.org nRF warning:
If I check an NRF board from Adafruit ( like https://circuitpython.org/board/circuitplayground_bluefruit/ ) there is a warning about UF2 bootloader version 0.61+ and See Update UF2 Bootloader below.
If I go on a non Adafruit board ( like https://circuitpython.org/board/Seeed_XIAO_nRF52840_Sense/ ) I have the same warning but there is no section "Update UF2 Bootloader" below.
My Bootloader was the following:
UF2 Bootloader 0.6.2-12-g459adc9-dirty lib/nrfx (v2.0.0) lib/tinyusb (0.10.1-293-gaf8e5a90) lib/uf2 (remotes/origin/configupdate-9-gadbb8c7)
Model: Seeed XIAO nRF52840
Board-ID: Seeed_XIAO_nRF52840_Sense
Date: Nov 30 2021
SoftDevice: S140 7.3.0
I went ahead as it does not cost to give it a try... and it worked.
So I don't know what the "problem" is... is it misleading warning? is it absence of bootloader section? or no problem at all?
thorny jay
I have something to say about CircuitPython.org nRF warning:
If I check an NRF ...
Ok 0.6.2 is > 0.6.1 so I did not have the issue. So the only thing is the absence of section and the reference to a section that maybe does not exist for non-adafruit board.
thorny jay
Ok 0.6.2 is > 0.6.1 so I did not have the issue. So the only thing is the absenc...
I don't know what is the criteria to display or not the section Update UF2 Bootloader but here are the nRF board without it:
- https://circuitpython.org/board/microbit_v2/
- https://circuitpython.org/board/Seeed_XIAO_nRF52840_Sense/
- https://circuitpython.org/board/makerdiary_nrf52840_mdk/
- https://circuitpython.org/board/ssci_isp1807_dev_board/
- https://circuitpython.org/board/tinkeringtech_scoutmakes_azul/
- https://circuitpython.org/board/teknikio_bluebird/
- https://circuitpython.org/board/simmel/
- https://circuitpython.org/board/electronut_labs_blip/
- https://circuitpython.org/board/sparkfun_nrf52840_mini/
- https://circuitpython.org/board/challenger_840/
- https://circuitpython.org/board/ssci_isp1807_micro_board/
- https://circuitpython.org/board/ADM_B_NRF52840_1/
- https://circuitpython.org/board/espruino_banglejs2/
- https://circuitpython.org/board/makerdiary_nrf52840_connectkit/
manic glacierBOT
midnight ember
I don't know what is the criteria to display or not the section `Update UF2 Boot...
yup, they all say See Update UF2 Bootloader below. but the section is missing. That could easily cause confusion and frustration.
onyx hinge
that section is present when the condition is true: {% if bootloader_version and bootloader_id %}
It's missing because the metadata of those non-Adafruit boards is incomplete, and/or they do not use the UF2 bootloader.
Feel free to submit patches to circuitpython-org to improve the metadata for third party boards.
tulip sleet
I made the See Update UF2 Bootloader below. I dind't know there were so many third-party boards that didn't have that setion. I was addressing several forum posts that didn't realize their boards needed a bootloader update.
manic glacierBOT
misty garnet
not clear on the current status, does UART block on a write until all bytes are transmitted? If not, are we going to get a flush() command that does? Or is this complicated because different MCUs handle things differently?
onyx hinge
https://daniel.feldroy.com/posts/2023-08-pypi-project-urls-cheatsheet hum it's just come to my attention that it's possible to add a discord link to a pypi page. not sure it's worth the trouble to do so, we'd have to patch all the libraries.
slender iron
not clear on the current status, does UART block on a write until all bytes are ...
I think the intention is that it blocks until everything is sent. There may be cases that queueing to the hardware is enough to unblock
onyx hinge
I think the intention is that it blocks until everything is sent. There may be c...
I think the minimum guarantee may be: it has been submitted to a hardware FIFO or other hidden queue so the passed-in buffer can be freely modified. but I'm not sure we guarantee the bits have actually 'gone out'. For SPI I hope it's stronger than for UART, since the SPI FIFO has to be fully drained before /CS may be deasserted and there's not an explicit 'flush'...
slender iron
I know we had issues with PIO and shutting it down because the OSR was actually flushed.
@tulip sleet heads up you are scheduled to host today's meeting
tulip sleet
<@&356864093652516868> Reminder of Monday meeting, in 2 hours, 2pm US ET / 11am US PT. Add you notes to https://docs.google.com/document/d/1vJq1BU6TdWuTJVSCUKHbE4F6m4ZoIcEMsjPUotB5LZg/edit?usp=sharing
I had this notice stashed to send now
thanks for the reminder, though, sometimes i don't remember!
cobalt grail
So that would be <t:1691431200:F>
Just a friendly reminder Discord do support local time stamps - https://c.r74n.com/discord/timestamps
tulip sleet
Just a friendly reminder Discord do support local time stamps - https://c.r74n.c...
I had no idea, thanks!
cobalt grail
There are other generators out there that might suit you better though.
slender iron
last I looked, there was no way to do recurring times