heh
#circuitpython-dev
gilded cradle
Oh, ok. I'm using the breakout. Could be the diff.
Anyway, thanks. I now have some more things to try.
crude blaze
sure thing!
if the hardware is different I would also maybe check if any I/O polarity like reset or so might be setup differently
gilded cradle
Gonna try wiring a 2.4" featherwing v2 first just to make sure nothing strange is going on
solar whale
Does anyone know if there a known limit to the number of characters that may be sent via the UART mode of the Bluefruit LE Connect app. Using the latest version of the IOS and McOS Desktop app, I can only send 20 characters max (18 + 2 EOL characters) With an old version of the MacOS desktop app, I am able to send longer messages. It does not throw any error and it echoes the long messages, but it does not seem to send them. I was just wondering if others are seeing the same thing.
tulip sleet
Does anyone know if there a known limit to the number of characters that may be ...
I see you opened an issue -- thanks. The app was updated last month but the changes are actually from March or so.
solar whale
Ah OK -- Are you seeing the same behavior?
Yikes, that is tough code to read through....
tulip sleet
Ah OK -- Are you seeing the same behavior?
I did not try it - I'll take your word for it 🙂
solar whale
I just put one of these together and stumbled across it https://learn.adafruit.com/bluetooth-controlled-matrix-led-sign-using-bluefruit-connect using the new nrf52840 matrix featherwing
crimson ferry
Is there a way at runtime to get the MicroPython version that's merged? I thought there was once, but I can't find anything.
onyx hinge
within a git checkout you can ```~/src/circuitpython$ git describe --tags --match "v1.*"
v1.21.0-19942-g7a4e7c36cd
inner orbit
Is this the correct place to come for some hints about compiling CP ?
I can compile for the Trinket M0 no problems. When I try to remove built-in modules to make space for something else, I get the following error: (I am following the instructions on the Adafruit page)
undefined reference to `usb_midi_module'
When I am using this in the .mk file for the Trinky:
CIRCUITPY_USB_MIDI = 0
stuck elbow
remember to do make clean
inner orbit
jedgar@jedgar-linux-imac:~/Documents/circuitpython/compiler/source/circuitpython/ports/atmel-samd$ make clean BOARD=trinket_m0
- Verbosity options: any combination of "steps commands rules", as
make V=...or env var BUILD_VERBOSE
rm -rf build-trinket_m0
Same output. And I know from maybe two years ago, this was working for me. But a new install, etc.
It also failed with some other modules like neopixel_write, rainbowio, etc
stuck elbow
the dependencies between modules are not automatically resolved, if there is another module that depends on usb_midi, you will need to disable it too
the line of the error may give you a hint about what module that is
inner orbit
Understood.
No line numbers, just undefined reference, so I think maybe something is not in the right place. But I can compile without that line and works fine. Who knows?
inner orbit
Looking or that now 🙂
for
Miracle of miracles, it just started working. Huh.
If it starts to fail I may come back with more information 🙂
Thank you for your help, sir.
crimson ferry
thanks tannewt and jepler, I was thinking along the lines of our sys.version string, which indicates a nominal CPython version (and sys.implementation includes , _mpy=262)
tulip sleet
thanks tannewt and jepler, I was thinking along the lines of our `sys.version` ...
did you see https://github.com/adafruit/circuitpython/pull/8808; of any interest or use to you?
tulip sleet
micropython added the platform module, which we haven't enabled yet
but it is merged in
crimson ferry
so maybe just wait for yet
tulip sleet
if you see something helpful that would be added by platform, that would be an incentive to make it work sooner
right now version-y stuff is spread out between sys.version*, os.uname(), and something else I forget
or maybe nothing else
crimson ferry
just looked at it, the platform string is board-specific, which we could truncate, but I was just thinking of a user-code way to know the main ancestors of the currently running code... versions of CircuitPython, MicroPython, CPython, mpy-cross
not critical at all, just could be informative
sys.implementation might be the 3rd you were thinling of
tulip sleet
yes, now I scrolled back here
manic glacierBOT
Here is a screenshot from the video on this page:
Let me know if I can provide more help :)
orchid basinBOT
manic glacierBOT
manic glacierBOT
proven garnet
Is str.partition() currently not enabled for all boards as the docs say? Relatedly (or alternatively), does anyone know if the GPS library is currently only viable for larger boards anyway?
tulip sleet
Is `str.partition()` currently not enabled for all boards as the docs say? Rela...
that's right, we were saving space on non-Express SAMD21 boards
if the library uses partition(), then it's never been used on tiny boards. It's a pretty big library
I don't see a call to partition() in the library
proven garnet
It would be added as part of an PR, so it sounds like it would then become an issue
proven garnet
it imports successfully on a Trinket M0, so I'd say avoid it
Sounds good, thanks!
gilded cradle
Hi @crude blaze. I got the display working on the 2.4". It turns out the filename of my overlay had characters it didn't like. I wasn't able to get touch working though. Was there some trick to it?
slender iron
@desert yarrow let's move here. I think ideally the microusb port would work as OTG where it can detect host or device use
that'd allow the CIRCUITPY use case to work still
desert yarrow
@slender iron - Sorry about the faux pas there. 🙂
desert yarrow
No because support was just added recently added that support the FTDI chips in there (very similar to a normal CDC device, but they require some special tweaks internally)
I created this in an effort to create a small RP2040-native USB library... if it can work, I'd like to see if I can access it from CP or MP or whatever possible.
manic glacierBOT
desert yarrow
I like TinyUSB, but it's been very hard to get running what we need (simple USB Host support for FTDI based USB Devices).
slender iron
in CP we'll only support TinyUSB
desert yarrow
I know, that's why I can't use it (yet?).
slender iron
that way the work applies across all ports
desert yarrow
Does CP allow for external libraries/modules, though?
For example, could I pull in custom code into CP through an import?
Or does CP have TinyUSB embedded always (ie - and you can't run it without TinyUSB)?
slender iron
you could maintain your own fork and not call it CP
you can run it without tinyusb but you really shouldn't
desert yarrow
I see, ok.
slender iron
get it work with tinyusb host and you can use esp32-Sx chips then too
they do otg as well
desert yarrow
Ok, sounds good. Ha Thach has done a great job with TinyUSB and the latest releases do support what I needed. They just hadn't been fully percolated through the other libraries yet. If the latest TinyUSB v0.16.0+ makes it's way into CP, then that support should just "show up" (for free-ish).
Is there a timeline when TinyUSB v0.16+ will be in CP (maybe already?)
slender iron
there is a 0.16 PR open right now
desert yarrow
HOLY SMOKES@
desert yarrow
That would be the solution... v0.15.x did not support the FTDI chips as needed... but v0.16.x does!
slender iron
CP doesn't use the CDC level of host support though
we expose the lower level instead
desert yarrow
Ok
This patch has support for the 0xCD18 product id from vendor 0x0403 (FTDI)...
desert yarrow
Hrm...
slender iron
the expectation is that you'd write a driver in python
desert yarrow
Is there a reason why not, if it's there alongside HID?
slender iron
we use native hid host internally to input serial into CP
desert yarrow
Or... can I use the low-level USB Control transfer messages to talk to the device directly?
desert yarrow
Ah, there we go. The issue before was that the FTDI devices were simply ignored and they could never even enumerate, so I was dead in the water.
slender iron
though it could be interesting to make a pyserial api
desert yarrow
Now... they at least enumerate... and thus, I should be able to at least "talk to" them
slender iron
yup
slender iron
54 minutes ago: https://github.com/adafruit/circuitpython/pull/8809
you don't need it for the low level stuff though
it's only a change to the cdc host driver
desert yarrow
MERGED! Wow!
slender iron
@onyx hinge uploading multiple files to SD card via web workflow doesn't work because we stop the VM immediately and then do the reload wait
desert yarrow
The important thing is that at least, now, the device will be enumerated!!! Before it was just ignoring it because it didn't recognize the VID/PID or Class information. Whatever the reason, the device was just being ignored. Now, it should be recognized.
slender iron
I think it'd enumerate with older versions. It just wouldn't work with the native cdc host driver
desert yarrow
@slender iron - Now that it's landed, will a "current" build of CP include it, or is there a release date when it will be included. Or, can I build a new version right now? Or, should I wait for a nightly release?
slender iron
you can build now. absolute newest will have it once the CI runs
desert yarrow
Phenomenal! Thanks
slender iron
OTG isn't a priority for us because you can't use USB to edit code when using it as a host
desert yarrow
Is there a wifi-based editor for CP, so you wouldn't need the USB port?
desert yarrow
It is this? ==> https://code.circuitpython.org/ <==
slender iron
that works for BLE
you do http://circuitpython.local/ for local wifi
assuming you have mdns support in your os and router
otherwise you can use the ip address
desert yarrow
that's awesome
slender iron
👍
desert yarrow
btw, it's kinda fun watching the USB chatter back and forth between host and device... beautiful to see the little dance between the two! I'll checkout CP later when the CI build includes the latest goods.
crude blaze
Hi <@375455651050160130>. I got the display working on the 2.4". It turns out th...
nice!
@ touch, does it throw an error during boot in dmesg?
In the latest Kernel the driver should at least be available but things like rotation etc. will not work as that requires the driver changes I had submitted to Pi.
my overlay is currently assuming those changes to be in place but I think none of them should prevent the driver from loading at all
gilded cradle
nice!
@ touch, does it throw an error during boot in dmesg?
In the latest Kernel...
Yeah, I think the issue is because I was trying to use the one in your PR and it is expecting the driver changes.
crude blaze
ah yes, the ti-x-plate property is required by the old driver but I depricated that
if thats added it should work
gilded cradle
Oh right. Well, I'm going through and seeing what overlays I can get working for now. I got one for resistive 2.8 and capacitive 2.8 and display is working, but having issues with touch working intermittantly.
gilded cradle
ah yes, the ti-x-plate property is required by the old driver but I depricated t...
Ah, yeah. I think that's what the dmesg error indicated.
crude blaze
ah you figured it out, nice
Hm yea thats interesting you also saw that. What I observed were some display artifacts when touch was sending data. I figured likely EMI issues of my current messy jumper cable setup.
Do you have that on a proper HAT or are you also using jumper wires atm?
gilded cradle
I'm using a proper hat
crude blaze
I'm using a proper hat
does touch just bork out now and then or do you also get any sorts of weird display artifacts?
I'm not sure if I was seeing two seperate issue. Touch was def. a bit weird overall but I think there might also be a bug in wayfire there
gilded cradle
A little of both.
crude blaze
ah okay interesting
libinput was still showing the touch data coming in just fine but the window manager was just not reacting at times
gilded cradle
Interestingly, if I use the MIPI driver and the overlay at https://github.com/adafruit/Raspberry-Pi-Installer-Scripts/blob/main/overlays/touch-ft6236.dts it works perfectly.
Possibly it's a conflict with the ili9341 driver or something. I haven't tested that.
crude blaze
ah wait, are you using a TSC2007 or a focaltech cap touch display?
gilded cradle
Oh, for the TSC2007, I'm using a breadboard setup, but I'm also adding other PiTFTs and using HATs for all those.
crude blaze
ah okay! Yea then I wonder if it might actually be an EMI issue on the TSC2007 end. I did not have such issues with DSI displays that used focaltech cap touch
maybe worth waiting for the updated HAT then
gilded cradle
Oh, I never got touch working on the tsc2007. I switched over to the HATs and I did have issues on those.
crude blaze
ooh I see, then I misunderstood earlier
gilded cradle
No worries. Just trying to clarify.
crude blaze
so you had weird display glitches on those HATs too?
gilded cradle
yeah
crude blaze
huh!
gilded cradle
Well, not the display itself so much. Just the touch input. Like it sometimes leaves a tiny selection box before the cursor disappears.
crude blaze
ah, yea that I also experienced. That is definitely a bug in Wayfire but haven't been able to really reproduce to report to Pi Kernel
gilded cradle
Do you have any of the HATs like the capacitive one?
crude blaze
no unfortunately not. I only have DSI displays here with cap touch
gilded cradle
If so, here's an overlay that's doing that behavior:
crude blaze
eh RGB I mean, the ones for that upcoming DSI to RGB adapter
crude blaze
a bunch if not most of these sorta ones:
https://www.adafruit.com/product/5792
gilded cradle
This is what I was using: https://www.adafruit.com/product/2423
Yeah, I have a bunch of the RGB displays as well for when I was getting the Qualia up and going.
crude blaze
This is what I was using: https://www.adafruit.com/product/2423
ah yea no those I don't have around.
But yea if its mostly the Desktop behaving weirdly, I'm fairly certain that is something being buggy in Wayfire
gilded cradle
ah yea no those I don't have around.
But yea if its mostly the Desktop behaving ...
Probably
orchid basinBOT
midnight ember
@lone axle submitted a PR to CircuitPython_SoftKeyboard. Got a little bit done today. 🙂
lone axle
<@382939733107408897> submitted a PR to CircuitPython_SoftKeyboard. Got a littl...
Nice! Thank you. That is merged 🎉
manic glacierBOT
lone axle
Not exactly CircuitPython but still cool to see. Linux kernel is getting support for seesaw gamepads built-in https://www.phoronix.com/news/Linux-6.8-Adafruit-Seesaw
inner orbit
Can someone enlighten me as to where to put .py files I want compiled into my Trinket M0 custom firmware?
A very small function that I would like embedded to save some "disk" space.
tulip sleet
Can someone enlighten me as to where to put .py files I want compiled into my Tr...
you "freeze" modules that you are adding: https://learn.adafruit.com/building-circuitpython/adding-frozen-modules
Note that recent changes increased the available space on CIRCUITPY on main: https://github.com/adafruit/circuitpython/pull/8567
manic glacierBOT
inner orbit
Note that recent changes increased the available space on CIRCUITPY on `main`: <...
I am using some frozen modules, per the Adafruit instructions. I have a custom .py file I need to have compiled into the build.
idle owl
@slender iron I picked up a 7.3" Inky Frame. I think you added support for the 5.x" version. Is the 7.3" version supported now? If not, how difficult is it to add support for it since the other one works? I have never touched display stuff in the core.
slender iron
<@252717193496756235> I picked up a 7.3" Inky Frame. I think you added support f...
I think there is a pr open to add it. If not, then you’ll need to copy the init code from the python driver for it
manic glacierBOT
manic glacierBOT
idle owl
I think there is a pr open to add it. If not, then you’ll need to copy the init ...
Ah! Ok I'll look into it once I have it in hand. Thanks!
lone sandalBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
keen eagle
I've been doing some CircuitPython builds and noticed something odd I wanted to check in about
When I build from main I see __dict__ when I tab to autocomplete from an object in the REPL:
Adafruit CircuitPython 9.0.0-alpha.6-26-g84c4f62278 on 2024-01-21; Raspberry Pi Pico W with rp2040
>>> import wifi
>>> wifi.
AuthMode Monitor Network Packet
Radio __dict__ radio
When I build 9.0.0-alpha.6 I don't see that
Adafruit CircuitPython 9.0.0-alpha.6 on 2024-01-21; Raspberry Pi Pico W with rp2040
>>> import wifi
>>> wifi.
AuthMode Monitor Network Packet
Radio radio
is that a debugging thing enabled between releases or something unintentional? (I understand what __dict__ is, just surprised to see it visible here)
manic glacierBOT
onyx hinge
just a proof of concept at this point: UVC 'video camera' support is made possible by recent contributions to TinyUSB, and I got the very most rudimentary level of support working at the same time as circuitpython.
hidden rain
Try to control an E-S MOTOR 37SG-3650BL motor with PWM. It seems that the required voltage is 12 VDC. It runs regardless immediately without pwm. The PWMIO when set to True turns it off. Any ideas why?
random junco
I'm in the planning stages for a podcast episode all about CircuitPython 9. Looking at the release notes for MicroPython 1.20 and 1.21 that have been merged into CP9, is it fair to say these are "under the hood" improvements that most end users won't notice? Or is there a big feature or update I'm missing?
tulip sleet
I'm in the planning stages for a podcast episode all about CircuitPython 9. Loo...
What we took are mostly fixes to the Python VM. Any board-level changes reflected in those release notes are not in code we use.
The "split heap" code from MicroPython now enables us to use heap allocation outside the VM, which was very awkward before: any dynamic storage allocation could only be done once, before the VM started. The heap then used the remaining RAM. Now there is an "outside" heap which can be used while the VM is running, and that storage will not be GC'd. So various storage allocations that used to be static can now be dynamic, such as stuff needed for USB setup.
We also removed the "long-lived storage" scheme that was added a long time ago to reduce fragmentation. In that scheme, storage that we expected to live a very long time (mostly allocations for compiled bytecode) was allocated at one end of the heap, and shorter-lived storage (like the temp storage used during compilation) was allocated at the other end. Part of the scheme involved moving allocated objects and adjusting pointers. The moving caused some inherent but obscure problems where objects' identities seemed to change.
Now long-lived storage is gone, because it was not very compatible with the split-heap scheme. This may cause some projects on small-RAM boards like SAMD21 not to work any more due to increaed fragmentation. We have some ideas for doing something like long-lived storage in a different way that wouldn't involve moving objects, but that won't be in 9.
random junco
Thank you Dan! I really appreciate it
tulip sleet
Scott and I can talk about this more if you need more background.
random junco
sounds good, I'll let you know if I have any follow-up questions
thorny dove
Correction: Board: Adafruit Feather ESP32-S2 TFT
onyx hinge
There's no existing support for "YUV" images in CircuitPython displayio is there?
tulip sleet
There's no existing support for "YUV" images in CircuitPython displayio is there...
not that i see
onyx hinge
the top half of the frame was set to red by assigning to a memoryview from Python
onyx hinge
Hello everyone and particularly <@&356864093652516868> -- we'll have our weekly meeting in a little over 75 minutes from now. Please take time to make sure the notes document is up to date, then join us in the CircuitPython voice channel at 11am Pacific / 2pm US Eastern if you can.
Notes doc: https://docs.google.com/document/d/1hvExFQWMsV78bORs1uBxon_bncchrHPV96r4cAWEhcA/edit?usp=sharing
candid sun
i'm going to sit out the meeting today to power thru some guide tasks but wanted to extend a group hug
turbid radish
Hi @idle owl !
turbid radish
Do send your tidbits to cpnews@adafruit.com, they really help
gilded cradle
Yep, I can read the Blinka section
lone axle
am good to go for libraries 👍
random junco
noted!
midnight ember
hope you are ari are doing better this week ❤️
slender iron
silver tapir
Also, if somebody needs me to work on native languages from your country, I'd be happy to. (US is kinda complex as it has way to many, but maybe I can sort it out by lang family)
This would be part of the unicode CLDR standard, so it should work without extra hardware, but native in Windows, MacOS, Linux, Android, etc.
midnight ember
amazing updates Jepler. that sounds really amazing.
mortal kernel
midnight ember
@lone axle cardboard? If you give me some dimensions I can design a 3D print for you.
silver tapir
There was an amazing luthier in the US, Dr Carleen Hutchins, that de-mystified violin making, and also created a family of 8 sizes of violins that is way better than a typical string quartet. A group that plays them is the "Hutchins Consort" (.org), if you live in the US I urge you a lot to find a concert and go (or at least to watch them in youtube).
Since they are custom sizes, nobody makes strings for them anymore, so this family of instruments might go extinct. So I'm currently ordering materials and building a CNC for winding the flatwounds.
So sure, if you have a weird size instrument, or maybe you need lighter or heavier strings that you find commercially, even if it's a guitar/bass/ukulele, let me know and I'll to my best.
lone axle
<@382939733107408897> cardboard? If you give me some dimensions I can design a ...
Thank you, I'm excited to do cardboard for this one though. I've got a steady source of boxes from a regular purchase that I'm looking to build into projects.
The final dimensions will be in the guide along with the files for my decals if folks want to use the same size box and re-use the decals, or design a new enclosure for it. Would be great if anyone wants to design and share different variations of it.
midnight ember
Cardboard is a great media to use especially for learn guides as everyone has some and might not have a 3D printer.
I like the idea for a unifying method that is consistent no matter what the board or connection method is.
lone axle
I think it makes sense to release it in the different libraries all at once (and update the frozen ones in the core as well).
idle owl
@onyx hinge I have another obligation in a bit here. I'd appreciate it if you could pull in my topic next, and then go back to Justin for their second topic.
midnight ember
Is the amount of psram specified in the board config or is that just flash storage?
lone axle
Does OnDiskBitmap need to store it in RAM?
wraith crow
The Inkyframe does have an SD card I think though
midnight ember
you can also make the image smaller and scale it x2 or x3
sdcard is a much better storage idea to get around the low storage on the pico.
yeah no speed peformance but you'll have a ton more storage to work with
wraith crow
Speed isn't really an issue for e-ink 😄
lone axle
pimoroni does have some micropython samples here: https://github.com/pimoroni/pimoroni-pico/tree/main/micropython/examples/inky_frame it won't work with circuitpython or displayio but maybe some good inspiration.
midnight ember
Good luck Kattni! Keep us updated.
short tendon
What would be a good m0 + airlift example to make sure I'm good on space for ConnectionManager?
midnight ember
large e-inks sounds great for calendars, good project.
i think most m0 with airlifts at the time were M0 feathers with Airlift featherwings?
was the original pyportal m0 or m4?
lone axle
What would be a good `m0 + airlift` example to make sure I'm good on space for `...
Honestly I wouldn't worry too much about m0 + airlift. I don't think there are any projects with it (it would really tough to squeeze it in). M4 + airlift would be my primary concern on the airlift side. To make sure that pyportals and matrix portals continue to work as expected.
short tendon
the m4 + airlift and the m4 matrixpanel are my current low-end test boards
midnight ember
yeah at this point it's a little unrealistic to expect an m0 with airlift to have enough since the language has grown so much. m4 does seem like a better least common denominator.
idle owl
Thank you for waiting, Justin!
lone axle
fwiw I believe the stubs can work with VSCode as well (Anecdotally I think there may be more VSCode users than PyCharm in the community)
midnight ember
How/where would the stubs be hosted?
slender iron
yup, that's what I was thinking
midnight ember
if/elif x 400
lone axle
How/where would the stubs be hosted?
They're hosted in Pypi so installable via pip install commands. We have existing ones there today but it just doesn't cover board specifically today.
midnight ember
ah that would be nice
lone axle
Yep, I would be in favor of the default being a combination of all possible pins, then if they want they can run the command (or set env variable) to a specific one.
midnight ember
for devs that have a lot of boards and change multiple boards daily it would be nice to have a quick way to switch between stubs per board.
idle owl
I need to head out. Thanks again, all! Appreciate the tips.
midnight ember
and needs to be updatable for adding new boards that are released. pip upgrade sounds like an easy way to do that.
that would be nice. i do use nvm for an api project to store a token. knowing how much nvm is available is useful.
lone axle
Thanks for hosting Jeff. Hope everyone has a nice week!
midnight ember
Thank you @onyx hinge for hosting and everyone for participating.
errant grail
Thank you!
short tendon
Thank you everyone!
onyx hinge
slender iron
thanks for hosting jeff!
onyx hinge
It's a good time 🙂 I don't even stress out about it anymore.
midnight ember
Amazing
slender iron
Yep, I would be in favor of the default being a combination of all possible pins...
I'd do a minimal set. Otherwise people will expect it to work on their board
lone axle
I'd do a minimal set. Otherwise people will expect it to work on their board
I'm not sure a minimal set would end up containing anything though. Since someboards use board.GPIO#, some board.D# and board.A#, and some board.IO#. I think there may not be any pin names that are actually shared and valid across every device. I can't say that I've actually crosschecked it though.
slender iron
ya, true
short tendon
Could leave the default with nothing...
slender iron
or a SET_BOARD_ENV or something
short tendon
I wonder what would happen if you raised an exception in the stub
brazen hatch
Oh wow! It looks amazing. Easy virtual monitor! I will absolutely work that into ljinux once my display stack is finalized and this is merged.
tame creek
i think there's a regression in the 9.x bundle? i can't get macropad examples that use the display to work anymore. macropad_simpletest_display.py fails somewhere deep in adafruit_display_text/bitmap_label.py
brazen hatch
Also if this worked on espressif we could use it with espcamera for an easy webcam
CircuitPython webcam anyone?
lone axle
i think there's a regression in the 9.x bundle? i can't get macropad examples th...
If you're on github as well can you open an issue on the macropad library please? If you have the stack trace handy you can post it along with the issue. I can try to reproduce and dig into it a bit later this week if no one beats me to it.
tame creek
If you're on github as well can you open an issue on the macropad library please...
i think it's a change to the coordinate validators in bitmaptools, actually. i'll try to open an issue with a stack trace
should i open an issue against circuitpython or the macropad library?
lone axle
should i open an issue against circuitpython or the macropad library?
I would start with the macropad library since it's the highest level thing trying to use it. It can always be moved after the fact as appropriate.
Earlier I noticed this PR in the core: https://github.com/adafruit/circuitpython/pull/8823 that is dealing with bitmaptools validation. If you've got the opportunity to it might be worth testing the build from this PR to see if it happens to be the same issue / fix.
short tendon
So I'll try more again later this week, but I can't get another pyi to import into the base boards stub. Even excluding using an env. Tried:
importlib.import_module("boards.board_other")
from boards.board_other import *
import boards.board_other
And I added the file boards/board_other/__init__py
Doing import board.boards.board_other as board works fine (so I know the file is good)
onyx hinge
well drat 😕
short tendon
I'll still play with it. Will need to dig into pyi more first
onyx hinge
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1UgruSyeclpC_fzDze8qYEjEB7RMIq15InBfu18dC9Es/edit?usp=sharing
tame creek
also, 9.x on macropad sometimes crashes to safe mode with a hard fault after disconnecting from within tio, etc. not sure how reproducible this is yet
onyx hinge
https://typing.readthedocs.io/en/latest/source/stubs.html#version-and-platform-checks maybe only version checking conditions are supported, not that you'd gotten to the stage of trying the conditional parts
spare jacinth
So I'll try more again later this week, but I can't get another `pyi` to import ...
im no expert, at all... but afaik .pyi aren't actually imported but tools like mypy, IDEs and whatnot scan for info in file.pyi before file.py when they see import file
it might be a matter of how <whatever tool you are using> is handling relative imports or whatever... and thus maybe a manner of configuration, and not actual code (?)
onyx hinge
https://typing.readthedocs.io/en/latest/source/stubs.html#imports 'from ... import *' syntax is supposed to be explicitly supported though ('re-exported'). Less surprising that __import__ or importlib.import_module would not be.
so re-writing a single board.pyi stubs file is seeming more likely as the workable solution. you could just write it directly into the project directory and it would take precedence, right?
import specific_board as board
else:
import board
```otherwise maybe you go down the road of writing this or similar in code.py, though I'm not enthusedif import_module doesn't work, then it would be a huge if/elif block. And that's not scalable (IMO). I think either:
- telling users to do the
import specific_board as boardabove - a command line method to replace
board_stub/__init__.pyiwith a specific one
slender iron
if `import_module` doesn't work, then it would be a huge `if/elif` block. And th...
giant if block is ok if it is autogenerated
is it true that env variables can change its behavior?
short tendon
Not certain yet. Was focusing on importing first
short tendon
And when I had them in there, it found them anyway. I had tried:
import os
if os.getenv("board_id", None):
if os.getenv("board_id") == "board_a":
AA: microcontroller.Pin
elif os.getenv("board_id") == "board_b":
BB: microcontroller.Pin
and both AA and BB were defined, so switched to the loading of files first
tulip sleet
I'll still play with it. Will need to dig into `pyi` more first
I asked a q about .pyi in stackoverflow which might be relevant here. The replies corrected some misimpressions I had about .pyi: https://stackoverflow.com/questions/71106529/cannot-import-pyi-file-in-stubs-package
short tendon
I asked a q about .pyi in stackoverflow which _might_ be relevant here. The repl...
Ahh, so we might not be able to import a sub-pyi file
tulip sleet
yes, that was my impression -- thinking that the file is going to be executed by a type-checker is wrong, apparnetly
short tendon
So what @onyx hinge said with:
if STATIC_TYPING:
import specific_board as board
else:
import board
might be what we can do
That, or a shell command:
set_board_stub specific_board
I can try
lone sandalBOT
short tendon
@slender iron, well maybe....
slender iron
👍
short tendon
Dang it. even in a py it still imports if it's in an if block. So the full logic is still not there
I'm guessing since it's typing it's trying to be as complete as possible
I think this might be the closest we can come:
It could also be a good first step...
proud veldt
not sure whoelse besides @mortal kernel is working on ble for the pico w, but i found this and I hope it might be of some inspiration/help https://www.youtube.com/watch?v=RnpVAqOmc8E
proven garnet
I think this might be the closest we can come:
I'm just catching up here, I have a love/hate relationship with .pyi files for all of these reasons. Have stub-only packages been discussed? If you make their generation and PyPI upload automated, you could make it so that each virtual environment could download the typing stub for a single board.
tulip sleet
i think there was worry about 500 packages, one for each baord
proven garnet
You're limited to one per environment since the type checker will try to find a single match for board, but you could just pip install the typesheds
proven garnet
i think there was worry about 500 packages, one for each baord
That makes sense. The lack of logic functionality in typesheds is both understandable and frustrating.
manic glacierBOT
proven garnet
I know that for the CircuitPython VS Code extension you explicitly set the board, so I assume it works like the proposed environment variable that was mentioned prior in that context. I don't remember how much that setting affects things though (type checking? importable modules?).
Sorry if I'm going over treaded territory
short tendon
So making them all would be easy, there are 2 downsides:
- would pypi rate-limit us?
- if you wanted to see if your code worked for 2 different boards, you would need to uninstall one stub and install the other
proven garnet
Not sure about the rate limiting, it is weird that the answer isn't in an FAQ though I suppose there isn't a usual need to upload like this.
tulip sleet
I see this as more something that might be done in the editor extension rather than just by installing packages. Also, for instance, suppose you want to write generic code rather than for a specific board, or for a set of boards. It seems like a lot of work when you just want to check the pin names or supported features for a board
slender iron
interesting to see CP 8 was released at the start of february 2023
manic glacierBOT
tulip sleet
<@329766224093249548> want to pair program today?
something like that, I have a bunch of questions too about native subclassing. could not figure out what was going on in that issue. Are you going to eat lunch soon? I have some links to fix in a Learn Guide, that will take 10-15 minutes
slender iron
yup, gonna eat now and then I'll be free
tulip sleet
great, ping me when you're back
slender iron
@tulip sleet done eating. are you ready?
tulip sleet
yes, finished the link surgery
manic glacierBOT
short tendon
Not sure who is best to ask, so will just throw it into the ether...
I have a laundry list of things that I would enjoy doing with/for CP.
Things like:
- Connection manager
- Adding things to cookiecutter (tox related)
- Creating and updating some libraries (OAuth, generic time fetcher, config library, etc)
- adding test coverage to libraries
- and more!
Although I can ask random questions here or take up the next 10 weeks of in-the-weeds...
I thought I'd ask if there is a better way?
I will give as much to this as I can, and that is accepted. Right now it's kinda a thing at a time and would love to super-power that.
Let me know.
slender iron
Asking questions here is probably the most expedient way of going about it
short tendon
Sounds good. Was hoping for something that might be a little faster. And didn't annoy people with constant questions of "Do you want this"
slender iron
I'd much rather you ask before doing something than just doing it
🙂
gotta run now. thanks for your help!
short tendon
Oh I agree - don't want to waste anybody's time. Just trying to adjust my personal time and how much I can give to working on Adafruit work, vs other stuff
keen eagle
Oh I agree - don't want to waste anybody's time. Just trying to adjust my person...
I have the same thing - I don't want helping to be a burden 😉
short tendon
So my issue is more that I have time to give, and Adafuit is the first bucket I want to give it too. But if I can't pre-fill that with a task-list, it will go elsewhere. I have a bunch of ideas and partial libraries I've written for my own things and just trying to figure out what to do.
manic glacierBOT
short tendon
Back to the stubs thing. Although not sure why they come up double... but this is pretty cool (IMO)
If I had an easy way to do a video, I'd show code errors when switching boards. I like the idea of an env var, but this super quick switching is super cool...
And when I made the comments on the call yesterday about adding things into the doc-string? This is why
Not pulling Frozen Modules and Avalible, but would....
orchid basinBOT
manic glacierBOT
manic glacierBOT
The 20ms minimum is enforced by the underlying nRF software, and is part of the the BLE standard.
In this file, it seems to be supported
slender iron
If I had an easy way to do a video, I'd show code errors when switching boards. ...
ya, looks neat!
short tendon
Do we feel this is the way to go then?
slender iron
Do we feel this is the way to go then?
good question. the try except is a bit clunky. the env variable didn't work though right?
short tendon
Let me try that
Not in a way that I can get it to. I also tried:
from typing import TYPE_CHECKING
if TYPE_CHECKING:
import board.matrixportal_m4 as board
else:
import board
And TYPE_CHECKING is False in the editor
slender iron
it doesn't work with os.env?
manic glacierBOT
short tendon
it doesn't work with os.env?
Let me try again. I couldn't get the editor to pick up the value. Since it would be more of a run time thing...
short tendon
good question. the try except is a bit clunky. the env variable didn't work thou...
Nope. Even with ones that I know exist like Path it won't work. Realizes that the if TYPE_CHECKING works (think my editor got confused, after restarting it it worked - did try restarting for getenv), but that errors on a board...
slender iron
kk. Don't use pycharm so don't have experience with it
short tendon
does it work in VS Code?
slender iron
I don't use vs code either
short tendon
Mu?
slender iron
sublime text
short tendon
ahhhh. I do a lot in UltraEdit. sometimes simple is good
short tendon
@slender iron - do you feel we can move forward with creating the stub board files and figure out how best to use them in each IDE while that's moving along?
slender iron
I think you should do what's best for you
short tendon
Do we want to add them to circuitpython-stubs?
slender iron
Do we want to add them to `circuitpython-stubs`?
That's ok with me
short tendon
Cool
proven garnet
Cool
If you're doing one for VS Code, check out the CircuitPython extension for VS Code. For what it's worth I believe it already grabs board definitions and bubble library info. Not sure how its capabilities align with your goal, but I think I remember seeing it open source and also permissively licensed (MIT, I think) if you decide to do something else.
short tendon
If you're doing one for VS Code, check out the CircuitPython extension for VS Co...
I use VS Code, but more as tracking multiple projects and more like a text editor. Does it also support stubs?
proven garnet
I use VS Code, but more as tracking multiple projects and more like a text edito...
Yes, it support stubs, but I don't remember exactly how the extension does it, but you basically select which board you're using, and I believe it updates its autocomplete for board pin definitions
I can look into that though, to make sure that selecteing the board really does change pin definitions and it isn't just a general catch-all selection of pins
short tendon
I can look into that though, to make sure that selecteing the board really does ...
If you have time, that would be great
midnight ember
previously the only way to display a circuit python display to OBS was with the Pico DVI. Hopefully UVC will allow live streaming of a display with displayio to OBS? FoamyGuy could output the display instead of having a webcam pointed at the display. I know it's really early but the prospect of the capability is really exciting.
slender iron
previously the only way to display a circuit python display to OBS was with the ...
yup! it is exciting
@onyx hinge please re-review: https://github.com/adafruit/circuitpython/pull/8699
onyx hinge
@slender iron no, not asking for specific action around the flash ioctl thing. but we would end up with crashy-crashy problems if we took a flash implementation from micropython without modifying it, and sadly the compiler won't give us an error to help out
slender iron
thanks for doing the review!
slender iron
@onyx hinge build is green after consolidating the enum
onyx hinge
slender iron
very cool!
onyx hinge
colors are swapped but I know where that is
onyx hinge
@slender iron or are those colors right? The blue looks quite blue but the purple's correct.
slender iron
it looks ok to me. try the display ruler 🙂
onyx hinge
and choosing the opposite "swap bytes in word" setting is clearly wrong (green blinka)
onyx hinge
oh, okay
well boo. it works on macropad but on qtpy esp32s3 it fails to start if I have the UVC-configuring line in boot.py.
well I guess I'll be getting out the big debug guns tomorrow 🙂
slender iron
well I guess I'll be getting out the big debug guns tomorrow 🙂
very curious if imx works too
proven garnet
If you have time, that would be great
Yes, the extension does handle pin definitions
proven garnet
It generates stubs for them upon building the extension and supplying them with it. The build CI parses the C file pin defs and generates pyi files for each board.
The extension allows you to select different boards, and doing so changes Pylance (Python analysis) to use those generated and supplied stub files, as well as other things like giving the CircuitPython std lib modules that it also parses precedence over CPython's ones.
In summary, stubs are built when the extension is built on GitHub, and selecting the board reorganizes the order of where autocomplete looks to give precedence to those stubs.
proven garnet
For what it's worth, if my understanding is correct about the process, it means that pin stubs are not generated regularly but only on extension updates. The library autocomplete works with the same Pylance method, but instead downloads the most recent bundle from within the extension, so that IS up to date.
short tendon
Cool. I'll take a look. I'm also building from pins.c and other places
tulip sleet
@slender iron @onyx hinge I need 28 bytes in ru translations for a couple of M0 basic boards that use a lot of pins. There's not much left to squeeze, HOWEVER, if I turn on MICROPY_ERROR_REPORTING_TERSE, which shortens a bunch of messages in py/, then I gain back about 1300 bytes. These terser messages are not bad: they mostly leave out ancillary stuff, like number of values expected instead of just wrong number, or an additional type name for explication.
just to give you an idea of other headroom, trinket m0 ru has <300 bytes free, because it has many fewer pins.
These builds have already been pruned to an inch of their life. I checked for accidental double precision, etc.
these terse messages are already in the translation files, they just might not have been used previously
onyx hinge
I think it's worth considering
tulip sleet
i will do this in the morning; it needs a little extra mechanism to control from the Makefiles
thorny dove
Question: I am using an Adafruit Feather ESP32-S2 TFT flashed with CPY v9.... (see below). I tried to follow the move of board.DISPLAY to displaybus.DisplayBus, however I can't set root_group. See the REPL trial below: Adafruit CircuitPython 9.0.0-alpha.6-29-g5f318c37bb on 2024-01-19; Adafruit Feather ESP32-S2 TFT with ESP32S2. See image of REPL output https://imgur.com/a/RzygVaL
stuck elbow
Question: I am using an Adafruit Feather ESP32-S2 TFT flashed with CPY v9.... (s...
you are trying to set it on a class, you need an instance
thorny dove
you are trying to set it on a class, you need an instance
Thank you. I defined: display2 = busdisplay.BusDisplay. Next defined a displayio.Group(), then couldn't set display2.root_group.
stuck elbow
yes, display2 is a class, you never created an instance of it
by the way, it's better to ask this kind of questions in #help-with-circuitpython
thorny dove
ok, thanks
thorny dove
@stuck elbow I flashed the latest. Used board.DISPLAY. It runs OK
lone sandalBOT
short tendon
If anybody has a moment to look at this draft PR: https://github.com/adafruit/Adafruit_CircuitPython_Requests/pull/152 which stems from this issue: https://github.com/adafruit/Adafruit_CircuitPython_AdafruitIO/issues/115.
Just want to make sure the method is sound before I add tests
lone sandalBOT
slender iron
If anybody has a moment to look at this draft PR: https://github.com/adafruit/Ad...
I'd suggest un-marking it as draft if you want a review. I tend to ignore them
tulip sleet
@slender iron I am going to make draft release notes for a 9.0.0 alpha or beta release, which brings up the question, alpha or beta?
maybe we are ready for beta
slender iron
beta works for me
manic glacierBOT
tulip sleet
@onyx hinge are you working on any near-term fixes that I might put into beta.0? I think I could hold off on the terse messages because the beta.0 build might just fit due to the version string being shorter. Or we'll just have a couple of broken ru builds
onyx hinge
@tulip sleet there's the initially mistitled and still untested https://github.com/adafruit/circuitpython/pull/8847 for rgbmatrix. I don't think floppsy board support or uvc are important right this second.
tulip sleet
<@329766224093249548> there's the initially mistitled and still untested https:/...
that's easy to test, and we can give the one user a personal build for their project
onyx hinge
I wasn't intending to do much work today so I won't be testing it right now
tulip sleet
ok, so not urgent. I think I will just do beta.0 this morning.
that will get rid of a large backlog. Can do 8.2.10 in a week or two or less
onyx hinge
great, and thank you!
short tendon
Will there be an 8.2.10 release? I saw the PR into 9 for Fix subclassing dict which would force code to support 9+ if they wanted to use that
tulip sleet
Will there be an `8.2.10` release? I saw the PR into 9 for `Fix subclassing dict...
There will be an 8.2.10. I don't think we will backport the subclassing dict fix because it's large. The bug has been present since 5.something, so it's not been a huge blocker. It does prevent microdot from being ported to 8.x, I think.
short tendon
Okay. There are a few things I personally have that I've had to do some crazy work-arounds to get to work. Was there more than 1 pr? because it looked pretty small...
tulip sleet
though one could always wrap dict.
we are hoping to get 9.0.0 final out relatively soon
short tendon
You can wrap it, but dict(wrapped_dict_class) doesn't work
tulip sleet
I think you could target your connectionmanager to 9.0.0
i mean you make a duck-type dict-ish thing
short tendon
connectionmanager doesn't need it. I have a slew of other things sitting on a back burner. The main one is a Config library that loads from json files, and it works in the most part by exposing things like __getitem__, but there is strange behavior around passing it into a dict constructor.
I'll try it on the new 9.0.0 beta and see if it works
orchid basinBOT
short tendon
@tulip sleet should I open a bug for this:
class Test:
def __init__(self, data={}):
self._data = data
def __repr__(self):
if hasattr(self._data, "__repr__"):
return self._data.__repr__()
# dict.__repr__ does not exist in CircutPython
return str(self._data)
def __delitem__(self, key):
if key not in self._data:
raise KeyError(key)
del self._data[key]
def __getitem__(self, key):
if key not in self._data:
raise KeyError(key)
return self._data[key]
def __setitem__(self, key, value):
self._data[key] = value
def items(self):
return self._data.items()
def keys(self):
return self._data.keys()
In CPython this works:
In [26]: x = Test()
In [27]: x['a'] = 1
In [28]: x
Out[28]: {'a': 1}
In [29]: dict(x)
Out[29]: {'a': 1}
In CP 9.0.0-beta.0 it doesn't (also doesn't in 8.2.9):
>>> x = Test()
>>> x['a'] = 1
>>> x
{'a': 1}
>>> dict(x)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "<stdin>", line 19, in __getitem__
KeyError: 0
When adding logging CPython it calls keys and iterates over those. in CP it seems to start at 0 and work up from there
tulip sleet
<@329766224093249548> should I open a bug for this:
```
class Test:
def __init...
Try it in MicroPython also
short tendon
Try it in MicroPython also
Will do shortly
short tendon
@tulip sleet looks like it happens in both micropython 1.21.0 and 1.22.1
tulip sleet
<@329766224093249548> looks like it happens in both micropython `1.21.0` and `1....
not necessarily covered in https://docs.micropython.org/en/latest/genrst/index.html#, or it may just be a bug. I'd search the MicroPython issues and open a new one if necessary.
subclassing builtin types is known to be flaky in a number of ways; whether this is a real bug or just a limitation I am not sure
short tendon
since you can subclass dict in CP9, I can replace all the self._data with super() calls. But then of course it doesn't work on 8.x
Does Adafruit have anything like LTS for versions?
tulip sleet
Does Adafruit have anything like LTS for versions?
When we do a GA release, the last n-1 release is still available, and we give a pointer to the last bundle for the n-1 release. There may be some overlap when the n release is released as well.
short tendon
Gotcha. Makes sense
tulip sleet
Gotcha. Makes sense
we have talked about version-specific bundles where the library versions included are specific to a particular release, like 1.1.0 for 8.x and 2.0.0 for 9.x, but we haven't done that
short tendon
Oh that would be cool. (Could even match the CP version...)
tulip sleet
so for now we just write version checking code in the lib itself if it's needed and take it out later
short tendon
Makes sense. Would be cool if mpy-cross could automatically leave out the other bits
manic glacierBOT
onyx hinge
set up CIRCUITPY.IMG as /dev/loop0
mounted /org/freedesktop/UDisks2/block_devices/loop0 on /media/jepler/CIRCUITPY
jepler@bert:~/src/circuitpython/ports/posix$ cat /media/jepler/CIRCUITPY/boot_out.txt
Adafruit CircuitPython 9.0.0-alpha.6-48-ga17d2ac0db-dirty on 2024-01-27; posix with amd64
Board ID:default
I considered whether POSIX CircuitPython would use FAT or POSIX VFS implementation; I still think doing POSIX VFS would be good (access files directly on the hosting computer) but FAT was easy..
so, now, it arbitrarily allocates a 512KiB file "CIRCUITPY.IMG" which can be "loop mounted" on supporting operating systems like linux.
one reason I want to switch to POSIX VFS is because I want to run tests from this thing, not upstream's 'unix' port.
because it will not be such a hack to enable stuff in shared-modules for this port, meaning we can test more of it
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
idle owl
Oof. I'm trying to copy files to a Pico W, running 9 beta 0. I already nuked it and tried installing CP again. I can copy .mpy files fine, but anything else fails with an I/O error, both using Finder and cp via command line. Anything else being a .bmp, or a .py file. I also can't save code.py evidently.
Going to try 8.
random junco
Oof. I'm trying to copy files to a Pico W, running 9 beta 0. I already nuked it ...
are you on Sonoma and using the remount script?
slender iron
@onyx hinge I'm interested in getting CP running in an arm cortex simulator environment too
idle owl
are you on Sonoma and using the remount script?
Oh right. Bleach. I thought that only applied to writing updates to files. I am on Sonoma, yes, not using the remount script. Will go find it.
random junco
Oh right. Bleach. I thought that only applied to writing updates to files. I am ...
todbot's update works well as you can pass it an arguement to it: https://github.com/adafruit/circuitpython/issues/8449#issuecomment-1874682118
idle owl
Oh nice. Thank you!
Ok, now it tells me I don't have enough free space.
This is the bmp that comes with the simpletest file for the display library.
I can edit code.py at least.
devout jolt
This is the bmp that comes with the simpletest file for the display library.
be sure to do dotclean /Volumes/CIRCUITPY every once in a while, because it seems like dot-files are back now. Those can take up space
slender iron
I wonder how much flash space we'd save if we split dict tables into keys followed by values
slender iron
right now QSTR only dicts store a two black bytes for each 16bit qstr key
idle owl
I switched to trying to use the SD card, but it says Mount point directory missing. The pin for CS is correct, as well as SPI. I have no idea how to troubleshoot SD cards.
Docs mention if two things are using SPI, do the card first. It already was first. Code isn't even getting to init'ing the display right now.
The MicroPython example to use the SD card looks about the same as my code, so it's not that I'm doing that part wrong, I would think.
Maybe the card is formatted wrong? I did FAT. I remember there being some SD card formatter app, but I can't find it now.
slender iron
I switched to trying to use the SD card, but it says `Mount point directory miss...
You need to make a directory where you want to mount the SD card
I just changed this so it's a breaking change in 9.0
idle owl
Hah ok. Like mkdir from command line on CIRCUITPY?
slender iron
ya, or just in finder
idle owl
Finder isn't playing nice. But, understood. Thank you!
Huzzah!
It's upside down, but it displayed the image!
onyx hinge
wahoo
onyx hinge
<@426200547024961539> I'm interested in getting CP running in an arm cortex simu...
that'd be cool
slender iron
that'd be cool
I was looking at renode a little bit for this. I think zephyr is using it for their testing
idle owl
Oof. The PR is too old to have build artifacts anymore. I guess I'll pull it and build my own to test it.
slender iron
<@&356864093652516868> Just a last minute reminder that our meeting is in just over ten minutes in the CircuitPython voice chat. I look forward to chatting with you all!
onyx hinge
I totally got distracted, thanks for the ping 🙂
blissful pollen
lurking today, no notes
onyx hinge
I'm just working on a PR to add a placeholder CIRUCITPY/sd because I think that'll help folks
lone axle
I've worked with ones that were maybe 1/4 of that size, which required full PCs to drive the display. To see it that large and driven by a microcontroller is insane
short tendon
Liz that's pretty cool. If you are curious, you should look at https://github.com/boxpet/pygeomag, it's a project I did to adjust for magnetic declination
turbid radish
Thanks all!
candid sun
Liz that's pretty cool. If you are curious, you should look at https://github.co...
ohh very cool!
onyx hinge
thank you scott!
gilded cradle
Thanks
candid sun
thanks for hosting @slender iron
lone axle
Thanks for hosting Scott! Have a great week everyone!
turbid radish
Please do get your #CircuitPython2024 contributions in
ember iris
Thanks all! Have a great week everyone!
short tendon
ohh very cool!
Feel fee to ask any questions. It's an area I'm passionate about
onyx hinge
@slender iron what #ifdef turns on ble workflow? is it just CIRCUITPY_BLEIO?
slender iron
idle owl
Wheeeeeee building CP for the first time on this machine. -j22 That was the fastest build I've ever done.
ornate breach
How fast did it build?
idle owl
I didn't think to time it until after the fact, but it was probably 30-40 seconds?
ornate breach
CP usually takes 30-45 seconds on my Intel MBP with 8GB of RAM
idle owl
I'm terrible at timing, tbh.
Could have been less.
I have... an embarrassing amount of RAM in this thing.
ornate breach
I have 64GB of RAM in my desktop lol
I rarely do anything that justifies that amount haha
idle owl
I tanked the entire thing, and it still took as long as yours. That's odd to me. I wonder what's slowing it down.
Same. This is the first time I've tanked any of it.
ornate breach
Tbf, building in macOS terminal seems to go pretty faster for me in a lot of areas where it’s just slow on windows
idle owl
Ok, so the build loaded, the display works. I have no idea how I'm supposed to use these objects for the LEDs and buttons, though....
ornate breach
Aside from random things like forgetting to export a path
idle owl
Fair enough
ornate breach
objects can usually be accessed through the board class
idle owl
Right. That's where they are. But I don't understand what to do with them.
ornate breach
Is this a board you designed?
idle owl
I think the buttons are using a shift register.... I guess the keyboard shared module does that?
No, it's the Pimoroni Inky Frame 7.3". I got this board def from an existing PR.
ornate breach
Oh okay
idle owl
I'm testing it to see if it's working so I can get it updated and the CP folks can move forward with it
ornate breach
I would assume the keyboard shared module should handle button presses
idle owl
Ok, I'll see if the pins match what that module wants.
ornate breach
👍🏻
ornate breach
I hear ya there lol… it’s been so long since I’ve contributed to circuitpython. These days I just use the eval board builds for circuitpython if I hadn’t already submitted a board before lol
idle owl
So it's reading them as buttons 7-3, but hey it's a start!
slender iron
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1TwKXc8VJyR-Zju5QWouz7psBYjrVBhrkBT4DtXMLVlQ/edit?usp=sharing
idle owl
I've lit up an LED!
slender iron
idle owl
So two things. At least right this moment. One: The display needs to be rotated. Two: The buttons are reading weird numbers. I don't know how to go about fixing either.