ya, that's fine
#circuitpython-dev
slender iron
jaunty juniper
and the nightly links are 404 π€
slender iron
it hasn't changed much
ok, so pip install pysigrok-hardware-raspberrypi-pico
should be version 0.1.0
tidal kiln
ok to use venv?
slender iron
should also install pysigrok
yup
it should pull in pyserial too
but I may not have the deps set right
tidal kiln
$ python3 -m pip install pysigrok-hardware-raspberrypi-pico
Collecting pysigrok-hardware-raspberrypi-pico
Downloading pysigrok_hardware_raspberrypi_pico-0.1.0-py2.py3-none-any.whl (5.3 kB)
Installing collected packages: pysigrok-hardware-raspberrypi-pico
Successfully installed pysigrok-hardware-raspberrypi-pico-0.1.0
maybe not?
slender iron
ah ya...
tidal kiln
can just work around it for now. anything else other than pyserial?
jaunty juniper
U-ing π (-U)
tidal kiln
that worked. also grabbed click looks like
slender iron
pysigrok-cli -d raspberrypi-pico:conn=/dev/ttyACM2 -C GPIO16 --samples 1000 -c samplerate=1000000 -o test.sr
yup click is needed for pysigrok
pysigrok-cli --list-serial to find the right name
/dev/ttyACM2 - Pico - Board CDC is mine
the first command will capture to the test.sr file which you can then open in pulseview
pysigrok-cli --list-supported may be helpful too
tidal kiln
ModuleNotFoundError: No module named 'importlib_metadata'
what version of python?
tidal kiln
i could upgrade too. no reason i'm at 3.8 other than just haven't upgraded.
slender iron
I can add importlib-metadata as a dep and it should work back to 3.7
tidal kiln
ok. i'll hold off then. can be your test case for <3.10
slender iron
ok, pushed a new pysigrok version
tidal kiln
$ pysigrok-cli --list-serial
Available serial ports:
/dev/ttyS4 - n/a
/dev/ttyS0 - ttyS0
/dev/ttyACM0 - Pico - Board CDC
that sets min python to 3.7 and adds importlib dep
tidal kiln
just confirming it works with a manual install of metadata lib
tidal kiln
i'm going to just dump this venv and start over to test the deps.
slender iron
(should I make a pysigrok channel?)
tidal kiln
yep - agrees with dmesg π
jaunty juniper
it has the pyserial bug that we fixed 2 years ago but was never merged but I would use discotool to find the serial port anyway π
tidal kiln
ok, looks like the dep stuff is working ok now.
jaunty juniper
(where on mac the ports have the same name)
slender iron
let's move there
brazen hatch
With the help of the decoder I have deducted that yea pystack cannot be resizable.
The value is stored in bss.
Not sure what that is, but doesn't sound fun to play with.
And btw main wasn't happy either with getenv included.
It's up to you guys.
I am not putting any more time into this.
slender iron
Thanks for looking into it @brazen hatch
blissful pollen
I've been doing some tests on displayio speed ups tangentially to the GIF work I've been doing. Is there a pre-existing issue where I could post about my results? If not I can create a new one.
I couldn't find one search so just checking as I know jepler did some work previously looking at this.
jaunty juniper
so that is puzzling:
latest on ξ main [$] via C v14.0.0-clang via π v3.9.6
β― git status
fatal: cannot chdir to '../../../../../../ports/espressif/certificates/nina-fw': No such file or directory
fatal: 'git status --porcelain=2' failed in submodule lib/certificates/nina-fw
I'll just reclone, but what a weird error
twin sapphire
Any circuitpython devs looking for work? I need help with programming for my new toy project.
manic glacierBOT
blissful pollen
@onyx hinge I remember you doing some work on displayio speeds. Could you let me know where you posted it? Sorry I canβt recall and I wanted to compare it to what Iβve been looking at.
onyx hinge
@blissful pollen maybe https://github.com/adafruit/circuitpython/pull/7147
blissful pollen
Yes that was it I think. Somehow I missed it searching the PRs. Thanks!
analog bridge
Are the translations 4-bytes aligned on every board?
tulip sleet
Are the translations 4-bytes aligned on every board?
do you mean build-*/genhdr/compression.generated.h arrays?
analog bridge
the const compressed_string_t translation0... in the auto-generated build-*/py/translations-en_US.c
tulip sleet
I looked at the metro m4 express build, and it looks like it is packing those only on byte boundaries. From firmware.elf.map:
.rodata.translation0
0x00000000000600d8 0xd /tmp/firmware.elf.D3xoxR.ltrans0.ltrans.o
0x00000000000600d8 translation0
.rodata.translation1
0x00000000000600e5 0x7 /tmp/firmware.elf.D3xoxR.ltrans0.ltrans.o
0x00000000000600e5 translation1
.rodata.translation10
0x00000000000600ec 0xe /tmp/firmware.elf.D3xoxR.ltrans0.ltrans.o
0x00000000000600ec translation10
.rodata.translation100
0x00000000000600fa 0x9 /tmp/firmware.elf.D3xoxR.ltrans0.ltrans.o
0x00000000000600fa translation100
...
@analog bridge ^^
but the ones in an espressif build (I only have the 32n8 build built right now) do look like they are 4-byte aligned. Sounds like some compiler flag we don't know about
analog bridge
but the ones in an espressif build (I only have the 32n8 build built right now) ...
I think its ok, there is plenty of space free for now. I only looked at it because I am trying to estimate the translation size without performing the expensive link step in ci.
tulip sleet
it's a bit weird
tulip sleet
I think its ok, there is plenty of space free for now. I only looked at it becau...
I saw the 4-byte alignment on Metro ESP32-S2. It was being compiled with -O2. If I change that to -Os, it packs the structures tightly as with the Metro M4 I mentioned above.
tulip sleet
@slender iron @onyx hinge there is now an 8.0.x milestone, right now populated with the scorpio fix
slender iron
@onyx hinge I think the nrfutil thing is that the pc-ble-driver-py switched to "built distributions" from "source distributions" in the older version. I don't know why msys2 pip won't install a "built distribution" though
onyx hinge
Probably because there's not one? The msys one would not use the same file as a Windows one if there's compiled code in the mix.
slender iron
ah. I don't know what msys is...
cosmic turret
Hi, I am looking to load the Adafruit ESP32-S3 TFT( 4 MB Flash + 2 MB of PSRAM) with Micropython but i am not able to find the .bin file for it. Appreciate if anyone can share the file or link. Thanks
crimson ferry
they seem to have only the generic builds (other than the UM boards)... one of those might work https://micropython.org/download/?mcu=esp32s3
you could try the MicroPython Discord or forums (GutHub Discussions) for better support
onyx hinge
<@426200547024961539> https://github.com/adafruit/circuitpython/actions/runs/413...
and did nrf build any mingw64_nt_ wheels? (it appears not: https://pypi.org/project/pc-ble-driver-py/#files) -- this is why binary-only releases are bad.
slender iron
yup
there are normal windows releases there
but I don't know my windows tooling options
onyx hinge
@slender iron in order to move on I'd recommed just not installing it and not trying to build that board on Windows. It's nice if building as many things as possible on windows works, but there's a clear barrier here and it's not affecting many boards
and put a documentation note if there's an obvious spot to hang it
@slender iron this doesn't affect building nrf boards generally ,does it? I didn't look at the wip
onyx hinge
send a request to nrf for a build of the package that installs on msys python π you never know, they might provide it
I wonder if we could build it using a github package location instead of pypi
but I don't really want to sink time into finding out
slender iron
they deprecated their python version of nrfutil anyway
so moving to the newer version is probably the best way to go
onyx hinge
pc-ble-driver is the new version of nrfutil?
tulip sleet
https://github.blog/changelog/2023-02-08-pull-request-merge-queue-public-beta/
I saw this merge queue thing; not sure if the CI needs to be updated or not.
idle owl
@tulip sleet Where are we with ESP32-S3 and LC709203? The tracking issue is still open, but we thought it was fixed.
tulip sleet
<@329766224093249548> Where are we with ESP32-S3 and LC709203? The tracking issu...
I revised the library to do retries. It worked for me, but one user said it didn't work for them.
tulip sleet
i was still trying (unsuccessfully) for an actual fix instead of doing retries as a workaround
it's also the case that the MAXwhatsis needs retries
idle owl
Hmm fair enough
proven garnet
I saw this merge queue thing; not sure if the CI needs to be updated or not.
Looking at it now, looks neat!
idle owl
i was still trying (unsuccessfully) for an actual fix instead of doing retries a...
Well if it works....
manic glacierBOT
slender iron
pc-ble-driver is the new version of nrfutil?
it is a dep
tidal kiln
did 8.0 do something to change the left hand offset on pyportal TFT? it's different now. text used to be indented a little more to the right.
tulip sleet
did 8.0 do something to change the left hand offset on pyportal TFT? it's differ...
may be due to status bar code; do you mean, compared to 7.3.3?
tidal kiln
may be due to status bar code; do you mean, compared to 7.3.3?
yep
tulip sleet
is the change good or bad? π
tidal kiln
bad-ish. interferes with pyportal case.
one sec. responding. will link forum thread...
tulip sleet
could be something in https://github.com/adafruit/circuitpython/pull/6473/files there was rework of the terminal area
@tidal kiln in the 7.3.3 version, would the blinka icon be covered up by the case frame?
tidal kiln
yes. a little.
tulip sleet
I think maybe the offset was corrected, without realizing the case frame did not go to the edge of the screen. There is probably a sw fix for this. I wonder whether we should just change it on the PyPortal, since we sell that case.
Could you open an issue?
tidal kiln
sure
tulip sleet
there isn't any trick like reversing the direction of the frame is there? :/
tulip sleet
ok, tnx, too bad
timid bolt
fyi, I updated my async branch with the following interesting changes:
https://github.com/adafruit/circuitpython/pull/7554
- Move the loop code to its own module.
- I copied the CPython code for the asyncio module and started deleting stuff until it worked on CP. FWIW, it's 28k compiled.
tulip sleet
fyi, I updated my async branch with the following interesting changes:
https://g...
- I copied the CPython code for the asyncio module and started deleting stuff until it worked on CP.
It's just temporary that you put it here, I assume, but you could fork https://github.com/adafruit/Adafruit_CircuitPython_asyncio and put it in a branch there. We wouldn't store a version in the CircuitPython source tree (even though that's what MicroPython did, and I should just delete that from our fork).
I don't really know why they put it in their source tree, since it's really optional.
interesting about _asynciomodule.c
timid bolt
Yeah, I didn't know where to put it. I should do it as a CP library as you say. I need to work out the workflow with frozen modules/git sub-modules, etc.
tulip sleet
i don't think we would freeze it, unless it was on a board that would always use it. It uses up space we want for something else. And we generally only freeze things because there's not enough RAM (e.g. on Circuit Playground Express, with only 32kB).
timid bolt
yeah, there is already a module called _asyncio. But I found that out at the very end. So I just changed the name of the module to _loop and kept the source code name as _asyncio. π
tulip sleet
we can clean that up later. _asyncio is a catchall for C stuff, and it's not to be used by the end user
anything with _ like that, like _bleio
we don't guarantee those things have a stable API that you should think is a good idea to use
could you stuff _loop in _asyncio or does it need to be separate for some reason?
timid bolt
I don't think it should necessarily be frozen either, but fwiw Adafruit_CircuitPython_asyncio currently is.
tulip sleet
stuff in frozen is not always frozen, it's just available to be frozen. It's only frozen on the boards that want to freeze it. I think some third-party board wanted to freeze it, but we wouldn't have recommended that
but their choice
neopixel si there so CPX can freeze it, for exampel
timid bolt
could you stuff _loop in _asyncio or does it need to be separate for some reason...
it is technically possible
tulip sleet
timid bolt
stuff in frozen is not _always_ frozen, it's just available to be frozen. It's o...
right, so if I'm editing a library that is under frozen, do I edit it in the circuitpython repo under frozen, or do I get it in the library's own repro and some how pull it into the circuitpython repo?
tulip sleet
you always edit itin the library's repo, and update the submodule commit when it's ready. Everything in frozen/ is a submodule (or maybe nearly eeverything, again maybe some third-paryt weirdness)
in the top-levle makefile there is make update-frozen-modules target which brings them all up to date
we do that before significant releases
MicroPython traditionally froze a few things, but mostly as .py, not .mpy. The pyboard has plenty of flash. The ESP8266 needed a lot of initialization setup. We were much more constrained by internal flash size space on the SAMD21, our first port, so we improved the tooling of frozen modules (FROZEN_MPY_DIRS instead of FROZEN_MPY_DIR). THen later the MPy folks introduced the idea of manifests, which we haven't taken in (yet?).
onyx hinge
just one board freezes in Adafruit_CircuitPython_asyncio but we also happen to use it during build-time for testing
tests/run-tests.py: base_path("../frozen/Adafruit_CircuitPython_asyncio"), run-tests uses this copy of asyncio preferentially
manic glacierBOT
manic glacierBOT
jaunty juniper
what is going on with the CI ??
apparently sh 2.0.0 was just published on pypi and changes how sh.contrib works ?
the error in the CI is:
Run python3 -u build_release_files.py
Traceback (most recent call last):
File "/home/runner/work/circuitpython/circuitpython/tools/build_release_files.py", line 30, in <module>
sha, version = build_info.get_version_info()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/runner/work/circuitpython/circuitpython/tools/build_board_info.py", line 61, in get_version_info
sha = git("rev-parse", "--short", "HEAD").stdout.decode("utf-8")
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'str' object has no attribute 'stdout'
Error: Process completed with exit code 1.
like sh.contrib.git(...) now returns a string ?
analog bridge
@tulip sleet what should the directories look like in #7534
tulip sleet
<@329766224093249548> what should the directories look like in #7534
something like mpy-cross/windows/... linux-amd64, linux-raspbian, linux-aarch64, maybe? or linux/amd64, with more subdirectories? I hadn't really thought about it in detail. I cleaned out a lot of the older builds, which helped, but the naming is still kind of confusing
the names don't alphabetize well, with "static" and "amd64" first rather than last in the filenames
maybe just changing the order of the modifiers would help enough
mpy-cross-<version>-<os>-<architecture>, maybe
mpy-cross-<os>-<architecture>-<version> is probalby better
"raspbian" is now an outdated name.
I can put older stuff in "OLD/", and or I can rename the current files as well
analog bridge
cool... i am doing some tweaks and i can get it in
mpy-cross-<os>-<architecture>-<version> looks good
tulip sleet
`mpy-cross-<os>-<architecture>-<version>` looks good
sounds good, and I will clean up the current directory and write some script to rename according to your choices later
thanks for doing this
tulip sleet
@onyx hinge we need to backport https://github.com/adafruit/circuitpython/pull/7564 to 8.0.x, to prevent 8.0.x PR's from failing. I'll do this right now unless you already started.
onyx hinge
@tulip sleet do you want me to add it to that PR I just opened? it should be easy
tulip sleet
<@329766224093249548> do you want me to add it to that PR I just opened? it shou...
Sure! you can just cherry-pick the 7564 commit, you would need to resubmit anyway. tnx
onyx hinge
how's that look @tulip sleet ?
tulip sleet
how's that look <@329766224093249548> ?
successful builds! I will approve and merge when done.
onyx hinge
successful builds! I will approve and merge when done.
thank you Dan!
BTW will we merge 8.0.x to main or do I need to prepare versions of these bug fix PRs for main?
tulip sleet
thank you Dan!
BTW will we merge 8.0.x to main or do I need to prepare versions...
we will merge back to main regularly, if you want to do it, fine.
jaunty juniper
should I move #7563 to 8.0.x ?
tulip sleet
should I move #7563 to 8.0.x ?
that would be good. It's easier to merge from 8.0.x to main than vice versa
thanks
don't do it until I merge Jeff's PR, to pick up the sh 2.0.0 thing
I"m re-running failed jobs in 7563
jaunty juniper
ok
timid bolt
@onyx hinge what does the '0' signify in mp_obj_exception_initialize0?
onyx hinge
@timid bolt hm maybe I should rename it. Initially I wrote a function which received the "args" object (and was named without the "0"), then i noticed I was frequenty passing the empty argument tuple so I created the "0" variant that passed the empty arg tuple to mp_obj_exception_initialize. (this probably saves a few bytes of code, though I didn't measure) Then I ended up with the only caller to mp_obj_exception_initialize being mp_obj_exception_initialize0 so I removed mp_obj_exception_initialize.
at best the 0 is very unclear
but 0 meant "empty argument tuple", basically
timid bolt
general point: I've been bitten a lot by this "re-initialize after reset" problem too. It is somewhat of a bug farm. Some exploratory questions:
- do we really need soft reset?
- is there a more systematic way we could address the problem? (e.g., re-zeroing bss on soft reset)
slender iron
general point: I've been bitten a lot by this "re-initialize after reset" proble...
soft reset is meant to get back to a known state before rerunning python code. it isn't just bss. its also all of the peripherals
timid bolt
soft reset is meant to get back to a known state before rerunning python code. i...
yeah, it's more complicated than I thought. Indeed the exception object at the root of jepler's change is not even in bss. π€¦ββοΈ
just for argument's sake, what is the advantage of soft reset over hard reset? (assuming there is always only one vm)
tulip sleet
just for argument's sake, what is the advantage of soft reset over hard reset? (...
web workflow and ble workflow would be restarted after every reset: have to reconnect, etc. auto-reload would force a USB re-enumeration.
display would get cleared
any state you wanted to maintain across VM's would not be possible
timid bolt
ok makes sense, I wasn't familiar with those features
crimson ferry
any state you wanted to maintain across VM's would not be possible
any other state beyond sleep memory?
tulip sleet
i'm not sure sleep memory is maintained after a hard reset (button push)
it could vary
crimson ferry
I was just wondering whether other things besides sleep memory are designed to live across reloads
(other than what you listed above: workflow, display, USB)
blissful pollen
I was just wondering whether other things besides sleep memory are designed to ...
You could probably search the code for all the times a pin is set to never reset. ESP32 camera seems to do that
crimson ferry
I guess that would only apply across a reload, and not during a reset (?)
stuck elbow
nvm survives reset
crimson ferry
I'm trying to figure out what state survives reload, haven't found any consolidated docs about that
(re: reset... my assumption has always been that microcontroller.reset() is roughly equivalent to pushing the reset button, with caveats for deep sleep)
stuck elbow
there is a bit of state surviving, because you can reset to a bootloader
timid bolt
If this is related to my earlier question, by soft reset I mean ctrl-D. Afaik microcontroller.reset() does a hard reset.
crimson ferry
yes, reload vs. reset
it's a little clearer (to me) what survives reset vs what survives reload
crimson ferry
there is a bit of state surviving, because you can reset to a bootloader
I'd guess that uses rtc memory or similar, not flash
timid bolt
the stuff danh mentioned is for surviving reload (soft reset). Nothing really survives reset (hard reset) except non-volatile storage (i.e., the flash).
crimson ferry
microcontroller.on_next_reset survives reset (as deshipu noted)
I'm not sure about the onboard rtc across ports
jaunty juniper
it might be that that kind of reset is doing the action of going to bootloader, rather than resetting and remembering that it had to go to bootloader ?
crimson ferry
yeah, I don't know those mechanisms
jaunty juniper
things that are setup through supervisor will survive reload, the previous traceback is saved somewhere, set_next_code_file, etc.
crimson ferry
that's a good distinction (supervisor)
a bit confusingly though... ports/{port}/supervisor/port.c has a lot of the reset behavior
timid bolt
a bit confusingly though... ports/{port}/supervisor/port.c has a lot of the rese...
right, that "reset" is the reload π
Could someone point me to why there are only a fixed number of sockets? The only reason I could find is open_socket_objs in Socket.c, however I think this could be made into a dynamic array.
crimson ferry
afaict, it's a memory thing. the CYW43 driver defines 8 (but we don't really get 8 currently into userland - there's an issue for that), espressif has a compile flag
in espressif, sockets use onboard espidf memory, competing with pystack, monitor, AP connections, possible future ble features, etc (there are a number of related issues and PRs)
[outside the core, wiznet limits sockets to 4 (W5100S) or 8 (W5500), and ESP32SPI / Airlift has a limit of 10 in the NINA firmware]
crimson ferry
I'm not sure about the onboard rtc across ports
espressif does retain the onboard rtc time across microcontroller.reset() π
raspberrypi does not
timid bolt
I'm not familiar with the ESP ports. So does ESP-IDF == FreeRTOS + ESP SDK? Is that how the esp ports of CP run: with ESP-IDF on top of an RTOS?
crimson ferry
I think that's a fair characterization
tulip sleet
I'm not familiar with the ESP ports. So does ESP-IDF == FreeRTOS + ESP SDK? Is t...
I'd say FreeRTOS is an integral part of the ESP-IDF. They have modified it for their own needs as well.
timid bolt
was the presence of FreeRTOS useful or interesting when porting CP to Espressif?
tulip sleet
was the presence of FreeRTOS useful or interesting when porting CP to Espressif?
we took advantage of the task mechanism
but only in a limited way, for some monitoring tasks. THere is one major CPy task
we also had to find and fix bugs due to some assumptions ESP-IDF had inadvertantly made about what ran in which task and on which CPU (on the dual-CPU chips). Scott found those at least twice, I think.
they were PR'd upstream
onyx hinge
(I wonder if there is some invalid content or filesystem condition that causes an exception to be thrown outside the vm or otherwise create an unexpected failure condition; that could land you in a boot loop or otherwise 'bricked'-ish device)
(we SHOULD be working at a level without the possibility of an nlr exception but you never know)
brazen hatch
Mood.
It still ain't booting, but at least I am able to debug it.
I am trying to have it boot with a dynamic alloc instead of the bss thingy that it was.
Once that is done, the rest should be easy.
though I do not quite understand how the original allocation worked so it's all a guestimation.
brazen hatch
rp2_common still seems to expect _pystack to be in bss
manic glacierBOT
These changes fix adafruit/circuitpython#5947. I've tested on both the zero2w and the pi4b (The 4b didn't exhibit the original issue) and the boards now behave properly with 1 to 30 pixels. I've run a 30 pixel animation on the zero2w for 8+ hours without a hang.
Original corrupted rgb transmission:
Early Fix, not using transmission queue:
solar whale
Is it now recommended to use "settings.toml" to set all of the things we used to do via "secrets"? for example. to access my AIO username and key should I set them in settings.toml and then retrieve them with os.getenv?
Is the "settings.toml" file processed for all boards?
languid whale
so it's official, pi cow has bluetooth now
Supported by CP?
stuck elbow
Supported by CP?
no, it just appeared, give it some time
languid whale
no, it just appeared, give it some time
Like esp32Sx? π
stuck elbow
yes
crimson ferry
@solar whale Unofficial answer: yes I believe so, though there are so many examples out there with secrets.py. My take: settings.toml is useful for pre-boot.py values and for the ease of os.getenv(). But it doesn't bring in Python objects, so it can be more work, especially for more complex data. My personal strategy is to stick to pre-boot.py use, and put the rest in a .py file (or download it).
teal thorn
If I want to add support for another Arduino board, what would be the recommended source for VID/PID numbers? The MKR Vidor is very similar to MKR Zero. Fairly straightforward port, just need new IDs. Should I request from PID.codes?
tulip sleet
If I want to add support for another Arduino board, what would be the recommende...
For MKR Zero, we assigned Adafruit PID's. I'm not sure that Arduino has a mechanism for asking for additional PIDs, but I don't know whether we would continue that or not. You could submit a pull request and we can discuss it there. Will there be differences besides the name in the board definition?
teal thorn
The SD card pins become JTAG pins. Might eventually include different libraries for FPGA.
teal thorn
For MKR Zero, we assigned Adafruit PID's. I'm not sure that Arduino has a mechan...
The battery sense pin moved
tulip sleet
certainly noticeably different, got it
manic glacierBOT
manic glacierBOT
tender canopy
Hi, im having issue with BLINKA_U2IF vs BLINKA_FT232H, epd gives error on U2IF - SPI write_readinto Not implemented
example based on
So works fine on FT322H but not PICO
GC9A01A round display works but only without RESET line defined (No picture with reset)
SDD1327 without reset no good picture showing at all
ruby walrus
Is there a way to see the human readable source code for the built in libraries in Circuitpython? I am building an MP3 player on a pico and I would like to improve the bit rate and other things that might require me to view the source code for built in libraries such as audiobusio and audiocore. I did not see any of the built in libraries in the Python Source Bundle.
wraith crow
@ruby walrus I suspect the audio libraries you're looking at are not frozen python libraries but implemented in C++.
ruby walrus
C++ is fine. Do you know how I can look at the code.
Is been my experience that most other languages let you inspect the contents of all their libraries.
wraith crow
I think the modules you want to see are in this directory: https://github.com/adafruit/circuitpython/tree/main/shared-bindings
There doesn't seem to be many of the Adafruit folks on tonight, but if you have questions or want specific help this is the right place to ask about those modules.
The board specific parts of what you're looking for are probably in this directory: https://github.com/adafruit/circuitpython/tree/main/ports/raspberrypi/common-hal
blissful pollen
C++ is fine. Do you know how I can look at the code.
Is been my experience that...
I do believe all the mp3 stuff is built in. You can find most of it: https://github.com/adafruit/circuitpython/tree/main/shared-bindings/audiomp3
and
https://github.com/adafruit/circuitpython/tree/main/shared-module/audiomp3
Some other audio library information is port specific so you would find it in:
https://github.com/adafruit/circuitpython/tree/main/ports/PORT/common-hal (replace port with the specific family)
manic glacierBOT
slender iron
Hey <@&356864093652516868> , the pinned message says the meeting is today but the calendar says tomorrow (due to Lincoln's brithday in the US). Any objections to doing it today (in 2ish hours) and ignoring the calendar? I suspect that's what we all expected.
blissful pollen
I thought it was today so no problems here
lone axle
Same here, today is good for me.
slender iron
π
storm minnow
Wait, today is a holiday?
proven garnet
I wish every Monday was a holiday
Oh, @lone axle - the CI update to specify the Python version is ready if you want to use that for your PyGameDisplay library.
storm minnow
not sure where we get the holiday data from
Evidently somewhere that doesnβt totally keep up; pretty sure no individual presidential birthdays are βcelebratedβ anymore because we have Presidentβs Day (next week) to celebrate them all at once
thorny jay
I wish every Monday was a holiday
I felt that most US Holidays are on Monday's... or you (USA people) really have a lot of holidays (I don't believe that).
slender iron
ya, next week's meeting is shifted
thorny jay
Hey <@&356864093652516868> , the pinned message says the meeting is today but th...
I will not make it to the meeting, and if I have notes it will be done in less that 10 minutes on a phone. π
Don't worry for me, go ahead.
slender iron
You've got time.
lone axle
Oh, <@382939733107408897> - the CI update to specify the Python version is ready...
I am interested in that Can you point me toward how to set it up? Or want to PR there with the change for it?
And thank you for working on that configurability!
storm minnow
I felt that most US Holidays are on Monday's... or you (USA people) really have ...
I think most fixed holidays are indeed Mondays, lol. Or the βobservedβ usually is (like, Christmas is on a Sunday, the βobservedβ (and thus day off) is Monday)
proven garnet
I am interested in that Can you point me toward how to set it up? Or want to PR ...
Yup! There's a new note at the bottom of the README for the actions that details how to use it: https://github.com/adafruit/workflows-circuitpython-libs
blissful pollen
@lone axle in case you were wondering I did get the party parrot gif going. Long story short every frame have to blank the background (plus the alignment issues). There was an issue on the AnimatedGif library talking about it. You can solve it but memory or CPU time go up a lot
proven garnet
You can use that python-version argument for the build and both release actions
lone axle
Yup! There's a new note at the bottom of the README for the actions that detail...
Nice! Thank you! 
lone axle
<@382939733107408897> in case you were wondering I did get the party parrot gif ...
Oh, good find, it does make sense the transparency would add lot of complexity vs. being able to repaint all pixels each frame.
I wonder how hard it would be to make a script to "fill in" the transparency with some color on each frame.
blissful pollen
It's not even the transparency (which should work). It is the disposal method. If set to "2" the decoder is supposed to draw the GIF and then restore the background before drawing the next frame.
If you overwrite that pixel in the next frame - doesn't matter. But if the next pixel is now transparent the previous pixel matters. But now you have to buffer almost 3 frames of data
blissful pollen
I wonder how hard it would be to make a script to "fill in" the transparency wit...
It would be pretty simple if wanted. The code almost checks/skips transparent pixels. Just have to add a replace transparent with feature.
lone axle
Oh interesting, I was thinking of modifying the source gif. If it could be an argument in code to replace it with another color that would be very convenient too.
blissful pollen
You can modify the source GIF too I think. ezgif I think has an option like that
errant grail
Does the GIF use a displayio.Palette? If so, it could be manipulated as well.
blissful pollen
No it doesn't. Most GIFs change the palette frame to frame.
It would be possible to add a new mode to not preconvert and have a bitmap with a palette but not sure on the speed of that.
As I poke deeper into displayio there are a lot of speed vs memory type choices. Been thinking of ways more of the choice could be given to the user, without bloating the code too much
errant grail
Perhaps disable the palette extraction phase after the GIFβs initial frame palette is determined, keeping only a single palette? Could that reduce some of the cycles needed per frame?
blissful pollen
The GIF palette itself changes frame to frame so the code would have to update the displayio.Palette object every frame still.
But I'll admit I didn't look too closely at that processing
errant grail
Hmm. I wonder how often the frame palette content actually changes during an animated GIF. Donβt think that any Iβve created have changes frame to frame. Will have to take a look.
idle owl
@blissful pollen Is the In the Weeds topic yours? (There's no name on it. π )
blissful pollen
<@228322543507865601> Is the In the Weeds topic yours? (There's no name on it. οΏ½...
Yes, sorry I thought I put my name there but it is Monday π
idle owl
No worries! We're not there yet anyway π
idle owl
I need to miss today's meeting. Would have had to miss it tomorrow as well if we moved it, so nothing lost. π
idle owl
I need to miss today's meeting. Would have had to miss it tomorrow as well if we...
@slender iron FYI. You'll need to read Libraries, or see if someone else is interested. I made a note in the doc about it needing to be read.
slender iron
I can. No problem
idle owl
Thanks
slender iron
brb
manic glacierBOT
Well renamed the variable onto that branch, it's not like a name change is going to affect the fact it won't bo- HOLDUP.
Why did it work now?????
It also erased the internal filesystem for some reason..
stuck elbow
Hmm. I wonder how often the frame palette content actually changes during an ani...
there are two palettes, a global one and a per frame one
thorny jay
Listening only. (is someone speaking yet? I only get beep from discord when someone join)
tulip sleet
nobody talking yet
thorny jay
Perfect thanks
lone axle
πΌ
midnight ember
good luck trying to silence a cat
analog bridge
ya, I haven't spoken for quite a while π
still zephyr
same
turbid radish
Please subscribe at www.adafruitdaily.com
thorny jay
That Tulip CC looks pretty much like my #CP2021 and #CP2023, but in MP rather than CP.
gilded cradle
Just a moment...
go ahead and read it
I think I fixed audio. Discord changed sources on me.
onyx hinge
onyx hinge
seems that David's jiggler moves in an octagonal pattern: list = [(0,1), (1,1), (1,0), (1,-1), (0,-1), (-1,-1), (-1,0), (-1,1)]
(note: pylint probably wouldn't like using list as the name of a variable, since it is replacing a built-in)
blissful pollen
I've also had to build in a screwdriver slot to pry parts apart before
onyx hinge
my parts never fit that exactly. π¬
midnight ember
circuit python plant guides coming? i'd love to see that.
blissful pollen
midnight ember
excited for the auto installer from circuitpython.org, awesome stuff Melissa.
gilded cradle
Thanks
midnight ember
CI efficiency updates nice!
thorny jay
circuit python plant guides coming? i'd love to see that.
I am surprised the red from RGB are the right "frequency" for plant... There are specialized LED for that. I guess this can be checked with spectrometer.
Keep multi language for the "release" and alpha and beta, ...
" randomly add one language at each PR"
"randomly" could be a selected sequences.
slender iron
midnight ember
That kind of policy sounds aimed more towards small/tiny projects and not major projects. Seems geared to avoid excessive forking of popular smaller projects.
midnight ember
Yeah that's a big decision that's an Adafruit Folks and company level thing.
analog bridge
Thanks! π
midnight ember
Thanks for hosting @slender iron β€οΈ
lone axle
Thanks Scott for hosting!
ember iris
Thank you all! That last topic was properly in the weeds! Hope everyone has a great week
midnight ember
Great discussions, lot of cool stuff happening in 2023. π
digital shoreBOT
midnight ember
Looking forward to playing with pysigrok someday.
idle owl
I am surprised the red from RGB are the right "frequency" for plant... There are...
The wavelength of the red and blue LEDs on the RGB DotStars peaks within the desired range for plants. I figured it was worth trying.
idle owl
circuit python plant guides coming? i'd love to see that.
There's a plan for a guide, anyway. π
ornate breach
The wavelength of the red and blue LEDs on the RGB DotStars peaks within the des...
Is this the same for other RGB that Adafruit carries ?
slender iron
Here is the notes document for next Tuesday's CircuitPython Weekly meeting. It is 24 hours later than normal at 11am Pacific / 2pm Eastern on Tuesday the 21st here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if youβll be attending the meeting - itβs super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and weβll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1CmvwiKPFPOibcQyH-xomxXlqLq3m5JrL8zcmr3qE5Qw/edit?usp=sharing
proud ember
I just learned that CircuitPython code can be updated over Wi-FI on supported devices. (Thatβs awesome BTW!). The guide says it needs to be connected to a network with internet access to work. Are there plans to have a workflow where the CP device acts as an access point so the updates are as portable as from a USB cable?
Sorry if this is the wrong place to ask
slender iron
@proud ember we are hearing folks want that. I generally dislike using AP mode but it shouldn't be too hard
It can work on a network without internet but it doesn't currently support if the ESP is the AP
midnight ember
Are you talking about firmware updates or workflow? They're easy to confuse the two.
slender iron
this is the best place to ask
proud ember
Are you talking about firmware updates or workflow? They're easy to confuse the ...
I was thinking about updating main.py. Not sure what the right term is
proud ember
<@259138024556855296> we are hearing folks want that. I generally dislike using ...
Ah, good to know! The use case Iβm thinking of is being able to update the code in an underground lab (with poor network connectivity) where things have to be sealed, so things with USB ports sometimes arenβt allowed.
midnight ember
That would be part of the wifi workflow which I thought could run locally without the need for an internet connection.
slender iron
only /code/ needs internet
web workflow won't start the AP automatically but might be on that interface already. I haven't tried it
midnight ember
is web workflow and wifi workflow 2 different things?
i have yet to try either :/
slender iron
no, I think of them as the same
jaunty juniper
I don't think "wifi workflow" is a thing
midnight ember
ohhh
slender iron
its web workflow but only over wifi now π
midnight ember
got it
slender iron
k, lunch time for me
jaunty juniper
it is possible to make your own custom way to update files automatically or manually by running a server on AP that would replace a local file that main imports for example
I want to see a library for that, that people can just setup and call in a loop
midnight ember
could you setup an esp to both act as the server and AP? that would be interesting.
jaunty juniper
yeah python code, like httpserver, should work in client or AP mode
midnight ember
entire local network on a chip, neat.
sick creek
maybe you can
jaunty juniper
a person made a MP script called ugit that syncs the drive with the contents of a git repo, but I'm trying to make it easier to use and configurable
https://github.com/Neradoc/ugit-for-circuitpython/tree/circuitpython
also testing AP mode is really really annoying π
crimson ferry
could you setup an esp to both act as the server and AP? that would be interesti...
The CircuitPython AP channel is limited to the channel the station previously used to connect to the external AP. (if the station connects first)
you can have a device act as wifi AP and wifi station simultaneously, and as network client and network server simultaneously (note that there are some limitations on Pico W https://github.com/adafruit/circuitpython/pull/7101#pullrequestreview-1188988867)
proud ember
also testing AP mode is really really annoying π
Can you elaborate? I might be doing this at some point
jaunty juniper
well, you need a computer that connects to the AP to do the tests, the AP goes away when the board reloads since it's not implemented in settings.toml (yet), which might force you to reselect it manually every time, and your computer might not have internet access during that unless you have also an ethernet cable or something
but that's just when testing AP mode, most of the dev time you can be in station mode
crimson ferry
and the CircuitPython IP address is 192.168.4.1, and that may conflict with your LAN, so you may not be able to have Ethernet enabled on your computer
brazen hatch
I still plan on better implementing picow wifi stuff, but I really am out of free time.
It's buried down in my tasks list right besides the esp32-c3 serial issue
crimson ferry
realizing we may be limited by the cyw43 driver
brazen hatch
yea I plan on reading the source directly to figure out what to do
rpi docs really do not help here
crimson ferry
my sense is that the Pico W SDK is not as mature and granular as espressif
brazen hatch
Eh, it's still not even a year since picow was released
I presume in 2-3yrs it will be equally as mature.
The problem is than no other boards may utillise it.
rp2040's are used everywhere, but that wifi chip, only on picow
crimson ferry
there is an open issue for web workflow over AP https://github.com/adafruit/circuitpython/issues/6795
is that right though? you can buy the wifi module, and I read the license on the driver and it allows other use
jaunty juniper
but does anybody ?
crimson ferry
not that I know of
jaunty juniper
Father RP2040 W when ? π¬
crimson ferry
that would be nice
brazen hatch
Father RP2040 W when ? π¬
father π
crimson ferry
brazen hatch
the granddad of picow
crimson ferry
web workflow won't start the AP automatically but might be on that interface alr...
I tried it and coulnd't get it to work, maybe at least due to the AP starting after the web workflow bind
jaunty juniper
well now I have to leave the typo in, thanks π
brazen hatch
jokes aside, a esp01 + sdcard hat for the pico form factor would be very powerful
I may make it in the summer
slender iron
thanks for trying it @crimson ferry. it doesn't seem like it'd be that much work. π
crimson ferry
it might work if starting AP was a settings.toml
jaunty juniper
having a way to start/stop the web workflow manually (that sticks on reloads) could allow for network hopping, and also double as a security feature (start the workflow on button press)
ah but that would also require making wifi connection stick during reloads too
brazen hatch
well if setting.toml stuff run on every reload it really shouldn't be an issue
jaunty juniper
(which, I think is a feature we should have btw, settings.toml is too static)
crimson ferry
if web workflow is on, I think the connection lives across reload
KeyboardInterrupt:
Code done running.
Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 8.0.0-rc.1 on 2023-01-30; Adafruit QT Py ESP32S2 with ESP32S2
>>> import wifi
>>> wifi.radio.ipv4_address
192.168.6.179
what I'm saying is, if you wifi.radio.connect() to a different ssid, I don't think it sticks on reload ?
crimson ferry
ah, right
brazen hatch
sounds like something in need of a rewrite to me.
It feels like it should connect exactly the same way as setting.toml wifi stuff does
init'ing connection and web_workflow
there is no reason or benefit for the code to be different.
well, maybe there is actually.. not all ports that have wifi have web-workflow
but that could be solved by a simple #else
crimson ferry
you mean non-native esp32spi?
brazen hatch
Oh I was talking about the wifi module connections?
crimson ferry
I was wondering which wifi port didn't have web workflow
brazen hatch
Oh idk. I just assumed there are some
jaunty juniper
the thing is, in CP it's unusual to make things in python be persistent between reloads, but wifi access is worth it IMO, and is basically already there, it just needs python access to the feature
brazen hatch
just how picow was a bit ago
brazen hatch
the thing is, in CP it's unusual to make things in python be persistent between ...
sounds like a free pr to me
if only I had time
idle owl
Is this the same for other RGB that Adafruit carries ?
No idea, I only checked the DotStars. This is because they're the only ones that Adafruit carries that have fully-white LEDs, vs RGBW, which isn't enough brightness on the white for it to work, as far as I know. So if I'm already using the white DotStars, I might as well be using the RGB ones as it will be easier to wire up.
idle owl
Is this the same for other RGB that Adafruit carries ?
I verified things with research and LED datasheets. I'm sure you can do the same for other LEDs. Unless the datasheet is useless. π
ornate breach
Ah okay, cool! Iβll dig up the sk6812 data sheet as thatβs what I have laying around
timid bolt
web workflow (aka supervisor) and "code.py"/repl (aka python vm) are strangely independent consumers of the same wifi resource. So it would be weird if their wifi settings are in conflict. The supervisor gets its wifi settings from settings.toml at hard reset. The vm gets its settings from whatever is passed to wifi.radio.connect. I think the most predictable behavior is if the vm has to agree with the supervisor. So if web workflow is turned off, the vm can connect to anything. However if web workflow is on, the vm can only connect to the same wifi as in settings.toml (which is a no-op) and connecting to any other wifi is an error.
slender iron
this is cool: https://github.com/community/community/discussions/16925
@tulip sleet I think it could be a "how to port" guide
and contain links to the test examples
though we should have them in the cp repo
tulip sleet
not much there; I started doing it one way (more board-specific), but was asked to make it more general
slender iron
interesting that its very test structured
I wonder if we could automate it by pairing with a known good port/board
tulip sleet
interesting that its very test structured
the idea is that it would be a set of tests to help bring up a new board, to make it easy to test the functionality
slender iron
or I could use this pysigrok code and have it logic analyze on the other side
tulip sleet
I thought I was going to write a test for each and every pins, etc, but was asked to test basic functionality instead (does I2C work, etc.?). At least I think that's what I remember
crimson ferry
after a UF2 is loaded (espressif), time.localtime() seems to be randomized / undefined (rather than 2000 01 01, as after a reset), and can be outside the representable limits of time.mktime() - is this expected?
jaunty juniper
when it restarts after dropping a UF2 ? I wonder what could possibly cause that !
crimson ferry
yes, and I don't recall that happening until recently (it started triggering an exception due to some time processing I do in boot.py)
jaunty juniper
struct_time(tm_year=1943, tm_mon=1, tm_mday=27, tm_hour=3, tm_min=46, tm_sec=46, tm_wday=2, tm_yday=27, tm_isdst=-1)
that is unexpected !
crimson ferry
I don't think it used to do that
crimson ferry
espressif keeps time across microcontroller.reset, so some execution path is different from that
ruby walrus
<@1074401019250868357> I suspect the audio libraries you're looking at are not f...
Thanks for the help
onyx hinge
https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/check-your-code#workaround-for-pre-commit-issues-on-ubuntu-and-debian-3086216 @tulip sleet I think that they eventually put in a workaround for this within setuptools, it might be worth checking whether the section is still necessary .. or maybe it's better to leave it because someone might still have an affected version on some system they're using
tulip sleet
That is neat -- circuitpython/tools/convert_release_notes.py may want to recognize this.
halcyon breach
Is there any relationship between the circuitpython and micropython i2s systems? Seems like they were developed at about the same time?
tulip sleet
Is there any relationship between the circuitpython and micropython i2s systems?...
one might have cribbed from the other, but generally the peripheral-level API and C code is different on CircuitPython and MicroPython
halcyon breach
I was trying to understand how to extend the circuitpython implementation to do 32-bit-per-sample i2s, it's tricky
tulip sleet
which chip?
we have enough trouble getting non-glitchy audio in some circumstances. Higher-resolution audio did not seem like a high priority
are you trying to do high-fidelity, or are you using I2S for some non-audio data purpose?
there are clicks and pops sometimes at start and stop; we would like to address all this, but it probably requires a revamp of audio handling in general
halcyon breach
I noticed the micropython version has its own 256-sample DMA buffer, and the circuitpython one seems to be a little more abstracted. All I wanted to do was to bit-shift the audio to do volume control. Using the RP2040 and soon the ESP32C3 plus the MAX98357A DAC. Since the ARM chips tend to have a single-cycle multiplier, you could multiply the samples before copying them into a DMA buffer to do all the necessary bit shifts or overall volume control. (I opened an issue to propose this)
crimson ferry
anyone know what this may indicate?```
boot.py output:
...
it's not something I'm explicitly printing
onyx hinge
@crimson ferry ```c
#ifdef CIRCUITPY_BOOT_OUTPUT_FILE
if (boot_output != NULL) {
// Ensure boot_out.txt is capped at 1 filesystem block and ends with a newline
if (len + boot_output->len > 508) {
vstr_add_str(boot_output, "...\n");
thanks!
onyx hinge
if there are good reasons the cap can probably be increased.
crimson ferry
didn't have \n
I don't have a strong preference, the cap is easy enough to work around at that level, can write to another file
onyx hinge
also note how it doesn't try to print the partial message, so if you're printing just one large thing that comes to a single call in mp_hal_stdout_tx_strn it'll just get turned to "...", not the first ~500 character and then "...".
crimson ferry
I was bringing some data forward from safemode.py
manic glacierBOT
orchid basinBOT
tulip sleet
also note how it doesn't try to print the partial message, so if you're printing...
we should fix this
onyx hinge
so that print("x" * 1024) in boot.py will print ~500 "x" then "...\n"?
if (len + boot_output->len > 508) {
+ vstr_add_strn(boot_output, str, ??? arithmetic ???);
vstr_add_str(boot_output, "...\n");
yes, somethign like that, or maybe even say [truncated due to length].
perhaps I even meant to do that when I wrote that code
tulip sleet
maybe you did π
yup, it was you https://github.com/adafruit/circuitpython/pull/5536
onyx hinge
did we discuss the selection of the 1-block maximum? I'm not actually sure the motivation.
lethal abyss
hello everyone, it's been a long time since I tried some custom circuit python.
I'm currently exploring a way to produce a dedicated port for the Lilygo T-embed ESP32-S3
The build setup documentation has a dedicated chapter for espressif (even if it's strange to see it after all other chapters)
My issue is that during the esp-idf/install.sh step, I encounter an error:
ERROR: This script was called from a virtual environment, can not create a virtual environment again
I always work on my python projects from venv python and I don't know if there is a known way to install esp-idf anyway
tulip sleet
@onyx hinge my [truncated due to length] suggestion was a mild one, and makes for extra code and translations. Maybe it's not worth it?
@onyx hinge different topic: https://forums.adafruit.com/viewtopic.php?t=198806 are ESP32-CAM and ESP32-EYE different things? It's a strange error
analog bridge
hello everyone, it's been a long time since I tried some custom circuit python.
...
idf creates its own venv which is activated when you do source esp-idf/export.sh
lethal abyss
`idf` creates its own `venv` which is activated when you do `source esp-idf/expo...
so I have to exit my venv, use this one then install again the general build requirements. Maybe a quick note at the beginning of the setup documentation will help.
Thanks
onyx hinge
@tulip sleet yes esp32-eye and esp32-cam are different products. someone tried to add support for esp32-cam but we never got the camera to work iirc. closed PR https://github.com/adafruit/circuitpython/pull/6827
I forget, it seemed like the esp32-eye had some limitations too but I am confident the camera worked at a minimum basic level.
orchid basinBOT
onyx hinge
@tulip sleet because Q2's gate is pulled up to 3V3, does that cause it to conduct or not to conduct by default when CAM_PWR is not otherwise driven?
(that's from the esp-cam's datasheet)
there's no sw control of the camera's actual powerdown or reset pins π¦ just this control over its regulators
tulip sleet
i get confused by FET schematic symbols, looking...
onyx hinge
me too, thank you
https://www.vishay.com/docs/70611/70611.pdf it seems it has to be pulled LOW to conduct
tulip sleet
phone call...
onyx hinge
np
I think that's consistent with how the esp32-camera library treats the power down pin ```c
if (config->pin_pwdn >= 0) {
ESP_LOGD(TAG, "Resetting camera by power down line");
gpio_config_t conf = { 0 };
conf.pin_bit_mask = 1LL << config->pin_pwdn;
conf.mode = GPIO_MODE_OUTPUT;
gpio_config(&conf);
// carefull, logic is inverted compared to reset pin
gpio_set_level(config->pin_pwdn, 1);
vTaskDelay(10 / portTICK_PERIOD_MS);
gpio_set_level(config->pin_pwdn, 0);
vTaskDelay(10 / portTICK_PERIOD_MS);
}
this is esp-cam related too: https://forums.adafruit.com/viewtopic.php?p=960484#p960484
onyx hinge
yeah that is the same author as the thread dan led with -- did they double post it?
Here's what I have so far for a reply.
The esp32-eye and esp32-cam are different products, the eye is from espressif and the cam is from AI-Thinker (and possibly other vendors). It might load to work the bin file for a different board but it also might not.
A community member also tried to get the camera to work on the esp32-cam with a new board definition but failed with the same error: https://github.com/adafruit/circuitpython/pull/6827 -- we were unable to resolve their problem.
I notice that in your example, the "external clock" pin (clock input into the camera module) is not specified. Unless the camera is fitted with its own clock source this is required for the camera to operate. However, since some camera modules are fitted with their own clock sources (such as the Adafruit ov5640 breakout) we do not treat failure to specify the pin as an error.
Another possible problem is power and reset control of the camera. In the schematic, you will see that CAM_PWR is connected to p-channel mosfet Q2. This pin must be pulled LOW in order for Q2 to conduct and supply power to the camera module. This GPIO should be specified as the powerdown_pin= in the camera constructor.
orchid basinBOT
tulip sleet
@onyx hinge I think your interpretation is correct. There are few clear explanations but this is one: https://circuitjournal.com/which-mosfet-should-you-use-with-arduino
onyx hinge
OK, posted
tulip sleet
@slender iron @onyx hinge I am going to do an 8.0.1 release. It has the Scorpio fix, i2CDIspaly deinit fix, exception chaining fix, sh module 2.0.0 fix, and Spresense SDK update. I don't see any other outstanding PR's to include. Sound ok to you?
onyx hinge
that sounds like a good round-up of stuff!
slender iron
We should probably start doing spdx in the core files
slender iron
@analog bridge is your CI PR ready?
timid bolt
Re https://github.com/adafruit/circuitpython/issues/7333
It is because of code missing MICROPY_PY_LWIP_ENTER/EXIT blocks.
The is a lot of code missing these blocks. In lieu of a better solution, should I go nuts adding these blocks everywhere they are required?
the particular code responsible for the issue is mdns_server_construct, but inspecting elsewhere there are a lot of other places (e.g. in socketpool).
tulip sleet
@onyx hinge ^^
@timid bolt I think so, but Jeff has the last word on this. Could you paste a comment saying the above to the issue?
onyx hinge
sounds great, I love it when bugs are fixed.
crimson ferry
@timid bolt is it missing mainly in bind-listen-accept? I can do client sockets 'til the cows come home with no safe mode
timid bolt
<@1065334137243836517> is it missing mainly in bind-listen-accept? I can do clie...
Not particularly. The bug is not deterministic. Most of the time everything will be fine.
qq: why do we use the "raw" lwip interface, instead of the higher-level and thread-safe "socket" api?
crimson ferry
@timid bolt sorry I thought you were talking about the espressif issue... ignore my "client sockets" comment
(7333 rp hardfaults is very reproducible; 7459 espressif hardfaults is sporadic)
tulip sleet
qq: why do we use the "raw" lwip interface, instead of the higher-level and thre...
In rp2040? The code was mostly taken from the MicroPython impl
slender iron
qq: why do we use the "raw" lwip interface, instead of the higher-level and thre...
I think the higher level API requires an RTOS
manic glacierBOT
onyx hinge
qq: why do we use the "raw" lwip interface, instead of the higher-level and thre...
The socket API requires freertos too
At least that's what I recall
crimson ferry
analog bridge
@slender iron @onyx hinge @tulip sleet I asked scott to enable merge queue but on further reading the docs, I am not sure it will work as expected with the current ci/repo configuration and I need to test it first. Please disable it before merging any PRs.
brazen hatch
Uhh so you guys.. I made the whole settings.toml logic..
it works
it's just that..
getenv doesn't work
even after boot
int's don't work??
Ah nvm filesystem was corrupt
Now I only need to test stackless via settings.toml and we are good.
Yea stackless isn't working by just not allocating..
I will have to figure that out
But yea let's first get the commits up and the branch changed.
onyx hinge
<@252717193496756235> <@426200547024961539> <@329766224093249548> I asked scott ...
@slender iron @tulip sleet I have turned the setting back off for the main branch. (it's under branch protection rules, it took me a bit to find it)
tulip sleet
@onyx hinge re the MicroPython ticks_diff email I sent. Deeper in the discussion is this: https://github.com/orgs/micropython/discussions/10723#discussioncomment-4977540 where it's pointed out that the more complicated diff is for differences < 32 bits, which is true for our case. So never mind, I believe, though it's interesting.
analog bridge
Are the non-release builds wiped from S3 bucket on a new release?
onyx hinge
Are the non-release builds wiped from S3 bucket on a new release?
@tulip sleet probably knows
tulip sleet
<@329766224093249548> probably knows
@analog bridge I do it by hand with aws s3. I didn't mean to delete the very latest PR builds last night: I mistyped a glob pattern. I usually try to keep a couple of weeks of PR releases around so that people can do simple bisecting.
I have thought of automating it, but it's easy, and I can apply some judgment to the current situation.
onyx hinge
<@426200547024961539> re the MicroPython ticks_diff email I sent. Deeper in the ...
[huh I haven't really used github discussions before. on an initial look I'm finding it REAL HARD to differentiate between the text of a reply, and quoted text, due to the presence of those vertical bars to indicate them]
tulip sleet
you mean the leftmost vertical stripe is confusing?
@timid bolt TaskGroup PR for MicroPython: https://github.com/micropython/micropython/pull/8791 I am not sure we would have the async with problem that is being discussed: we don't use "native emitters"
onyx hinge
you mean the leftmost vertical stripe is confusing?
yeah there are a bunch of vertical things of differing thicknesses, and some of them indicate quoting and some don't
tulip sleet
so a strip that starts at an avatar pic is a reply. You could open an issue or a discussion on GitHub about this. They seem receptive to fixing various UI things.
could be a different color or a dashed line, etc
the whole background is diff color now, so not sure of the purpose of the line
onyx hinge
I agree, it's possible to figure it out when 'looking globally', and probably once you're proficient with the site. but for a person looking for the very first time, the entirety of jimmo's most recent post "looked like" something to skip over because it was (A) greyer than the rest and (B) had a vertical line, commonly used to indicate quotation. I'm just being a bit of a complainer, not really looking for a solution.
so yeah just wanted to complain
tulip sleet
I guess nested replies is a bit helpful, but I always wonder where to reply in these kinds of sturctures
onyx hinge
yeah I used to swear by threading in my messaging apps (trn, mutt) but now I'm more convinced that there's something good about the way forced-linearization of gmail threads & github issue discussions work actually. I wonder how you'd research such a topic
(because deeply nested and now very off-topic sub threads were a fact of life with the threaded way, and are broadly discouraged with the linearized way)
brazen hatch
If all goes well, pushing in 5, ready for review.
timid bolt
If I set the gcc flag -Og (optimizations for debug), the variables _ld_dtcm_data_size and _ld_dtcm_bss_size in ports/raspberrypi/supervisor/port.c are bogus. Does anyone know why?
jaunty juniper
what is the rule for using main versus the 8.0.x branch ?
tulip sleet
what is the rule for using main versus the 8.0.x branch ?
something that is a bug fix that should get in soon should be on 8.0.x. It is easier to merge from 8.0.x to main than to to cherry-pick from main to 8.0.x.
jaunty juniper
new boards ?
tulip sleet
If I set the gcc flag -Og (optimizations for debug), the variables _ld_dtcm_data...
new boards can be in main. I will make a 8.1.0-beta.0 soon enough and it will have the new boards.
I want to make one within a week or even this week. would have safemode.py and maybe the bangle.js
a lot of people are afraid to use something not marked as "stable"
jaunty juniper
ok
tulip sleet
If I set the gcc flag -Og (optimizations for debug), the variables _ld_dtcm_data...
I cannot see any obvious reason. If you compile -O2 vs -Os, or -O1 or -O0, do you see problems also? Is it optimization, or is it the addition of debugging info? How are they bogus?
timid bolt
-O1 seems to work and is debuggable. So I can go with that.
$2 = 16843009
(gdb) p _ld_dtcm_bss_size
$3 = 4286773247```I looked at the .ld files but there was nothing obvious to me
timid bolt
Yeah, it's definitely bizarre. π€·ββοΈ
proven garnet
Got rp2040js working with CircuitPython with a filesystem!
onyx hinge
@timid bolt I notice
>>> hex(16843009)
'0x1010101'
Is _ld_dtcm_bss_size in RAM? does something else write to it? maybe a watchpoint would be enlightening.
(and should it be in RAM? can it be made a constant that lives in flash instead?)
proven garnet
slender iron
interesting @proven garnet ! what do you plan to do with it?
proven garnet
interesting <@135415127879974912> ! what do you plan to do with it?
The original inspiration was using it in CI, specifically for seeing RAM usage changes when libraries are updated, but I'm sure there's a bunch of things that could be done with it.
slender iron
π
jaunty juniper
yeah that's a thing I would love to see, and getting memory dumps and learn more about the behavior of memory in general, make graphs (videos) like that π
https://youtu.be/baa5ILZTRkQ?t=856
still zephyr
WTG @proven garnet
proven garnet
I can create actions for setting it up in the CI and then have it be modular from there. I'm thinking that'll feed into those other things. I'm going to PR this upstream with instructions in the README, so I can post here for those interested in trying it out locally.
brazen hatch
Well with how I did it now, neither is needed
timid bolt
interesting <@135415127879974912> ! what do you plan to do with it?
Circuit Python in a browser!
brazen hatch
Browser in Circuit Python.
I will eventually make a tui browser.
I have rewritten nano, how much harder can it possibly be.
First I should make a proper html renderer though.
If I can render it to a file it should be as simple as wrapping less and giving it a bottombar
jaunty juniper
lynx seems like a natural progression from nano
jaunty juniper
Circuit Python in a browser!
note that rp2040js is the wokwi emulator, if you haven't seen it
https://wokwi.com/projects/new/circuitpython-pi-pico
brazen hatch
I could discard pystack_size huh
Imma do that too.
every byte counts
So. What should I do when the setting.toml value is invalid??
currently the code just defaults back to build default
Maybe print a message? Or trigger safemode?
And perhaps it's not wise running the resizable pystack code in safemode
For now making it a message.
It's not a fatal error really, we can just fallback to build default
However I should make checks as to if the alloc failed
if it failed it should redo the build default.
No idea how to check but fine
I will try it out, gdb will show which alloc failed.
slender iron
@brazen hatch I'd use the build default
brazen hatch
Oh it does use it in any and all errors, I'm just talking about the error message.
slender iron
I wouldn't add an error but it might be worth making it possible to read the value from python code
that way you could verify it worked
manic glacierBOT
slender iron
kk
brazen hatch
All done.
tulip sleet
Re https://github.com/adafruit/circuitpython/issues/7333
It is because of code m...
@timid bolt Hi - are you preparing a PR to add MICROPY_PY_LWIP_ENTER/EXIT, or are you working on something else? Fixing 7333 would be great. This could be an 8.0.x PR
if you are busy someone else can do this
timid bolt
<@1065334137243836517> Hi - are you preparing a PR to add MICROPY_PY_LWIP_ENTER/...
I am working on it. It turned out MICROPY_PY_LWIP_ENTER wasn't the solution to the issue. It was something else, which I'll send a PR for, and an explanation in the PR. I will send a PR for MICROPY_PY_LWIP_ENTER too since it is still technically needed.
tulip sleet
great, thanks!
jaunty juniper
@wraith crow I don't want to add to the thread, but note that changing the LED in code does nothing, it's off by default, and the core takes control of it anyway, I believe the OP was talking about the REPL (hence the one-liner they posted)
maybe someone using Thonny (which sits in the REPL until you press "run", not the CP normal workflow, but one used by some people)
wraith crow
I was just working with the neopixel_write module so it was fresh in my mind π€· I was thinking it would be an option to type/paste in the REPL
jaunty juniper
yeah a little long though
wraith crow
Clearly Scott's/Dan's suggestion is the way to go π
jaunty juniper
if I remember correctly, the default for dotstar was influenced by the FunHouse having 5 LEDs on top of it, which was not fun when in the REPL π‘ π‘ π‘ π‘ π‘
jaunty juniper
if I remember correctly, the default for dotstar was influenced by the FunHouse ...
or maybe I just got used to it π π§βπ¦―
tulip sleet
we have had sample variability in the status LED's. I can't remember if it was just dotstars -- it was some tiny ones. It's hard to pick a standard brightness
jaunty juniper
I was thinking it could be a define that can be overridden per board, especially for boards that have multiple LEDs
tulip sleet
that would be a good idea; in the case i mentioned, it was the same board that had wildly varying brightnesses based on manufacturer or batch
jaunty juniper
hmm
tulip sleet
manic glacierBOT
I don't think I'll add X channel support because it'll limit frequencies to the prescaler capable ones. VAL1 is used for top value usually but then it leads to a low pulse every cycle on X. This would prevent 100% duty cycle.
timid bolt
What do I have to do to make the 8.0.x branch appear in my fork on github?
tulip sleet
What do I have to do to make the 8.0.x branch appear in my fork on github?
git fetch adafruit 8.0.x:8.0.x assuming adafruit is your name for the upstream remote. I think it used to be there automatically. I was puzzled why it was so hard.
i had to do a lot of websearching to find this for myself yesterday
tulip sleet
What do I have to do to make the 8.0.x branch appear in my fork on github?
but your q is on github, oops, that I haven't researched yet
maybe if you just do it locally and then push
timid bolt
well, I have the 8.0.x branch appearing in my github repo, however if I rebase my commit to be on top of 8.0.x, I cannot push it back to my github repo
slender iron
what branch are you trying to push to? whats the error? @timid bolt
timid bolt
I want to rebase gneverov/issue_7333, the branch of this PR (https://github.com/adafruit/circuitpython/pull/7589) to be off gneverov/8.0.x instead of gneverov/main, where gneverov=https://github.com/gneverov/circuitpython.git
slender iron
and you have it rebased locally? you can likely do --force-with-lease when pushing to github
it allows you to force but only if your local copy of a branch matches the remote
timid bolt
Great, that worked! The PR is ready for approve/merge now.
timid bolt
would you like the lwip enter/exit changes in main or 8.0.x?
tulip sleet
would you like the lwip enter/exit changes in main or 8.0.x?
If you think it is clearly a bug fix, and will do some testing, I think 8.0.x is good.
timid bolt
Sounds good. I tested it as part of my #7333 changes. I think it is a low-risk change as adding unnecessary enter/exits should be harmless.
analog bridge
This aws warning makes no sense:
warning: Skipping file /home/runner/work/circuitpython/circuitpython/mpy-cross/mpy-cross.static/. File does not exist.
Error: Process completed with exit code 2.
https://github.com/MicroDev1/circuitpython/actions/runs/4190046131/jobs/7263056567#step:10:461
It says file does not exist but the file is right there as shown by the logs from tree mpy-cross above.
manic glacierBOT
Steps: (See: https://docs.github.com/en/actions/managing-workflow-runs/manually-running-a-workflow#running-a-workflow)
- Fork the
adafruit/circuitpythonrepository. - Activate GitHub Actions in your fork.
- Select "Custom board build" workflow.
- Click on "Run workflow".
- You will be presented with the following menu.
lavish saffron
I'm trying to track how many sockets have been assigned to socket objects in the Winzet5k driver. I have added a list of bools as a class variable and it is easy enough to track the assignment. The problem is tracking when the socket.socket instances are released. I've added a __del__ method to socket.socket but is not called.
e.g.
a = socket.socket()
a = True
The instance of socket.socket should be destroyed when a is set to True but __del__() is not called. Am I barking up the wrong tree?