#circuitpython-dev
manic glacierBOT
ionic elk
trying to get back up to speed with circuitpython main and I'm running into about a billion of these errors:
error: submodule git dir '/Users/hierophect/Firmware_Projects/circuitpython/.git/modules/ports/stm/st_driver/CMSIS_5' is inside git dir '/Users/hierophect/Firmware_Projects/circuitpython/.git/modules/ports/stm/st_driver'
fatal: refusing to create/use '/Users/hierophect/Firmware_Projects/circuitpython/.git/modules/ports/stm/st_driver/CMSIS_5' in another submodule's git dir
Failed to clone 'ports/stm/st_driver/CMSIS_5'. Retry scheduled
is something borked with the ST submodule? I don't see any issues for it
viscid pine
St sub modules were moved around a bit
ionic elk
Do I just need to re-clone the whole project? It is wholly screwed up for me
viscid pine
That’s the quickest way
ionic elk
blegh
I wish there was an easier way to tell Git to just force something to match upstream
no matter what
viscid pine
if you start over, do make fetch-submodules instead of any other submodule commands the first time
ionic elk
instead of having to do a billion manual submodule changes
viscid pine
it fetches a lot less
tulip sleet
@onyx hinge I'm going to raise a support issue with github because we can no longer see the summary build pages: they always unicorn, e.g. https://github.com/adafruit/circuitpython/runs/5705450743?check_suite_focus=true
for 6196, and same for 6202
onyx hinge
I was noticing the same thing, and there's nothing on githubstatus. thanks.
manic glacierBOT
manic glacierBOT
manic glacierBOT
mental nexus
I'm working to add the RGB LCD display support to the ESP32-S3 builds, and I'm trying to figure out how to get the linker to realize I need some files from the esp-idf. In particular, I need the files in a component called esp_lcd. I found a way of creating the #include statement where it could find the header file, but when it goes to link it fails to find the files I need.
Any pointers on how to direct the linker to find the correct files over in esp-idf/components/esp_lcd/?
stuck elbow
add -l to cflags?
mental nexus
I assume that will be in the Makefile
mental nexus
I see this, which mentions other "components" but when I add esp_lcd it gives an error.
ESP_IDF_COMPONENTS_LINK = $(IDF_TARGET_ARCH) app_update bootloader_support driver efuse esp_adc_cal esp_common esp_event esp_hw_support esp_ipc esp_netif esp_pm esp_phy esp_ringbuf esp_rom esp_system esp_timer esp-tls esp_wifi freertos hal heap log lwip mbedtls mdns newlib nvs_flash pthread soc spi_flash vfs wpa_supplicant
Error: ninja: error: unknown target 'esp-idf/esp_lcd/libesp_lcd.a'
proven garnet
Does CircuitPython support f-strings yet? The MagTag example uses them but my understanding was that they weren't supported yet.
mental nexus
no idea about ninja, sorry
Thanks deshipu, I'll try asking again tomorrow when more folks are around.
jaunty juniper
Does CircuitPython support f-strings yet? The MagTag example uses them but my u...
they have been supported for a long time yes, not on some M0 builds though
proven garnet
Ah, got it, thanks!
manic glacierBOT
crimson ferry
@proven garnet there are some limitations https://github.com/adafruit/circuitpython/issues/4723
mental nexus
Thanks deshipu, I'll try asking again tomorrow when more folks are around.
Strange. I added the extra esp_lcd to the middle of the list of ESP_IDF_COMPONENTS_LINK and it failed. I added it to the final entry of the list and it worked. Odd, but I'm not complaining.
In the Makefile in the ports/espressif folder.
proven garnet
<@!135415127879974912> there are some limitations <https://github.com/adafruit/c...
Oh good to know, thanks!
manic glacierBOT
manic glacierBOT
jaunty juniper
I see some libraries where the docstring :params ...: are not properly displayed on RTD because there is a missing new line above it, and ..code-block:: python are also not properly displayed because they are missing a space after ..
is that because of a sphinx update or a configuration change ?
do they have to be updated by adding the missing space or is there a parameter that could fix it ?
solar whale
Does anyone use MakeCode on the CPX -- I have what I think is a simple quesion <#help-with-makecode message> but have not gotten any replies
manic glacierBOT
digital shoreBOT
manic glacierBOT
tulip sleet
do they have to be updated by adding the missing space or is there a parameter t...
Go ahead and fix them. They may never have worked. We should probably just keep up with sphinx idiosyncrasies, annoying as it might be
jaunty juniper
'k I'm at 12 and counting 😉
missing init parameters because they are documented in __init__ instead of at the class level too
idle owl
<@&356864093652516868> Greetings! This is your reminder about the CircuitPython Weekly meeting. It happens in just over an hour. Remember to add your hug reports and status updates to the notes doc, whether or not you're planning to attend and participate. Looking forward to talking with you soon! https://docs.google.com/document/d/1fJ0odbK7m4jWHTnKL1medCeZAGBTZiNIW0np6nOd4hc/edit?usp=sharing
jaunty juniper
any idea on a friendlier way to display module level constants on RTD ?
idle owl
any idea on a friendlier way to display module level constants on RTD ?
Good question. I'm trying to think of whether we have another example of that, and what it looks like. Can you link me to the applicable bit of code so I have some idea what you're dealing with on the backend?
jaunty juniper
Good question. I'm trying to think of whether we have another example of that, a...
tulip sleet
any idea on a friendlier way to display module level constants on RTD ?
I think this is OK, as long as you add a docstring under each one explaining it.
idle owl
<https://github.com/adafruit/Adafruit_CircuitPython_ADS1x15/blob/4ef9748a2c69c3c...
The example I thought we might have, the constants are not in the docs separately. They're in the lib, but they aren't individually noted on RtD.
I added docstrings to all of them.
Which I would suggest for your example.
jaunty juniper
ah it might just look different because of how it renders locally with sphinx, I get some differences with the online version
idle owl
ah it might just look different because of how it renders locally with sphinx, I...
Local and remote will have differences, but they shouldn't be massive. But if you're happy with how it looks remotely, then that's what matters.
proven garnet
And mazel tov @slender iron!
solar whale
@idle owl What do you do in your spare time? 😉
idle owl
<@!330227457296957440> What do you do in your spare time? 😉
Right? 😄
mental nexus
ESP32-S3 and display question: My example code shows no life (dot clock does not wiggle at all). How can I add print debug in a esp-idf "component" library to verify it's not falling into an error in the esp-idf code? Possibilities: 1) mp_printf doesn't compile if I add it into the esp-idf code 2) Does monitoring the UART port show the ESP-LOGD outputs?
crimson ferry
(2) yes, those messages get added to the debug console, along with the normal esp messages. compile with DEBUG=1, include static const char *TAG = "tagname"; in the module, then ESP_LOG*(TAG, "hello world"); as desired
mental nexus
Cool. "TAG" is already there. Where do I turn on DEBUG=1, somewhere in the CircuitPython toplevel Makefile?
crimson ferry
make BOARD=* DEBUG=1
mental nexus
Even easier. I'll give it a shot.
proven garnet
As a previously primarily windows user, I always had to use WSL and copy it over to windows afterwards, so exciting!
tulip sleet
@onyx hinge received this from GitHub support re unicorns when trying to display job status:
Our engineering team has developed a fix for this issue that we expect to be deployed this week. I will follow-up here once this fix has been released.
If you need to access logs in the meantime, you can edit the job link param check_suite_focus from true to false. This isn't a solve-all as it will need to be edited each visit to a job's URL for a run, but you should have access to the job's logs from making that change.
(i.e https://github.com/adafruit/circuitpython/runs/5705450743?check_suite_focus=false )
onyx hinge
@tulip sleet the pull request makes some of the names better but there are still {{}}s .. e.g., {{ cookiecutter.__dirname }}/{{ cookiecutter.__libprefix }}{{ cookiecutter.__libname }}.py vs "{{ cookiecutter and 'tmp_repo' }}/examples/{{ cookiecutter.library_name | lower | replace(\" \", \"_\") }}_simpletest.py"
candid sun
thanks folks!
mental nexus
See y’all!
onyx hinge
it's not that much shorter I guess
proven garnet
Thanks!!
manic glacierBOT
mental nexus
(2) yes, those messages get added to the debug console, along with the normal es...
Excellent. I confirmed this works. Now onward into the code dungeon to see what's going on.
jaunty juniper
got a couple of CI errors due to just released click 8.1.0 and black
a new black release fixed it a minute ago, but our version of black is pinned I believe ?
onyx hinge
@jaunty juniper link to the actions run? i'll check into it
ImportError: cannot import name 'get_terminal_size' from 'click.termui' (/opt/hostedtoolcache/Python/3.10.2/x64/lib/python3.10/site-packages/click/termui.py) https://github.com/adafruit/circuitpython/runs/5726044671?check_suite_focus=true
jaunty juniper
onyx hinge
pre-commit pins black, but black probably doesn't transitively pin click
jaunty juniper
yeah that's what I think
onyx hinge
click's release is 8.0 -> 8.1, so it's unfortunate that they introduced at least two different incompatibilities as well
proven garnet
Uh oh, happening in the libraries I merged too. 🥲
onyx hinge
This is a feature release, which includes new features and removes previously deprecated features. The 8.1.x branch is now the supported bugfix branch, the 8.0.x branch will become a tag marking the end of support for that branch. We encourage everyone to upgrade, and to use a tool such as pip-tools to pin all dependencies and control upgrades.
well they knew they are breaking changes, but just don't follow semver
OK the core is failing via transitive dependency through typer. They have a PR opened but who knows how long they'll take to act on it https://github.com/tiangolo/typer/pull/375
proven garnet
well they knew they are breaking changes, but just don't follow semver
Is the fix to patch pre-commit-config.yaml?
onyx hinge
@proven garnet for libraries? Probably. if there's a release of black that works.
proven garnet
got a couple of CI errors due to just released click 8.1.0 and black
a new black...
Which release? 22.3.0?
onyx hinge
black 22.3.0 does work with click 8.1.0
@idle owl hey not sure if you are still around but are you reading this?
proven garnet
@idle owl I figure this is something adabot can do but let me know if you want any help
Happy to script something up
onyx hinge
20.8b1 -> 22.3.0 is going to be .. interesting, at a minimum.
idle owl
<@!330227457296957440> hey not sure if you are still around but are you reading ...
@proven garnet That is one of the patches I wanted to run this week.
onyx hinge
I think it just got more urgent 😕
idle owl
Dan noticed it a couple of weeks ago.
Ok.
I'm on "misc" so, I can slot that in first.
onyx hinge
well, we can always merge with pre-commit failing, it just feels bad
idle owl
Need Eva to be around though. I think she is.
onyx hinge
https://github.com/psf/black/issues/2964#issuecomment-1080974737
- repo: https://github.com/psf/black rev: 22.1.0 hooks: - id: black additional_dependencies: ['click==8.0.4']```
proven garnet
There's a Yiddish saying that translates to "Man plans and God laughs" and it feels appropriate 
onyx hinge
there's a reason they don't merely call it "version heck"
proven garnet
For the doc patches by @jaunty juniper is it worth holding off until this is patched?
idle owl
@onyx hinge Wait, isn't click specific to CP?
proven garnet
They won't materialize in RTD in the meantime, I believe
idle owl
For the libraries, what's failing specifically?
onyx hinge
@idle owl https://github.com/adafruit/Adafruit_CircuitPython_HTU21D/runs/5726019182?check_suite_focus=true
idle owl
Oh.
onyx hinge
click is used indirectly by black
idle owl
Right.
onyx hinge
click is also used indirectly by cascadetoml, used by the circuitpython build
but each one will need a different fix
idle owl
Starting on the patch now.
jaunty juniper
For the doc patches by <@199811382793863168> is it worth holding off until this ...
I'm taking a break anyway, I'm running a search in the bundle for some of the common errors and there's a lot more to look at
mental nexus
Clearly I have failed to reach peak enlightenment:
Guru Meditation Error: Core 1 panic'ed (LoadProhibited). Exception was unhandled.
@onyx hinge Oof, 22.3.0 was released 39 minutes ago.
idle owl
https://github.com/psf/black/issues/2964#issuecomment-1080974737
> ```For those ...
Do I need to do this? Or is this only if you don't want to update click as well?
I'm reading it as only if you want to use an older version of Black.
jaunty juniper
and then to see if circup needs an update too
lone axle
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1em2Z3EF5T_eaz04e-XdesuIZrxCEB_3plFVzBiat_jc/edit?usp=sharing
mental nexus
ESP32-S3 display code: Whenever the code allocates the display framebuffer in PSRAM, I get a crash. I suspect that CircuitPython has already claimed all the available PSRAM space, so there isn't enough available. I updated the issue above. Two options I can think of: 1) I need to tell CP to leave some memory for the framebuffer, or 2) write a custom esp_lcd_new_rgb_panel function for the ESP-IDF with an input parameter of a framebuffer, where CP allocates the framebuffer RAM and passes it to the esp-idf.
idle owl
(╯°□°）╯︵ ┻━┻
idle owl
This PR for the patch is failing because of Black. So I updated Black on Adabot, and when it runs remotely, pre-commit Black is failing on a specific file. It passes locally. I tried running Black on that specific file locally, no change.
The remote version is different than my local. I viewed raw on GitHub, pasted into my local version, it showed modified on git status, ran Black, and it updated the file.
I did a pull before starting any of this... So I have no idea.
Then caused merge conflicts. Oof. Fixed, but oi.
Survey says..........
PASSED.
stuck elbow
did the computers go too far?
idle owl
Here goes something. Applying my first Adabot patch.
My fancy Git setup made it fail entirely. 😄 Trying again.
onyx hinge
ironically my change to cascadetoml failed because of black because of click
proven garnet
ironically my change to cascadetoml failed because of black because of click
It's turtles all the way down
idle owl
@onyx hinge @jaunty juniper @proven garnet Patch completed! (Finally!) You should be able to update your PRs, which should rerun them. I also think if you rerun the Actions on the releases, it should use the updated files.
proven garnet
Sounds great, thank you!
onyx hinge
<@!426200547024961539> <@!199811382793863168> <@!135415127879974912> Patch compl...
Thank you
tulip sleet
@onyx hinge I did not see the cascadetoml fix until just now. If you are not immediately working on a PR to circuitpython, I can do that now.
maybe there's nothing to be done? black and click are now compatible?
onyx hinge
<@!426200547024961539> I did not see the cascadetoml fix until just now. If you ...
Is it pinned in circuitpython or will a fresh build pick it up after the pypi release?
tulip sleet
I just restarted a small failed build: https://github.com/adafruit/circuitpython/actions/runs/2053908250
onyx hinge
Oh did click make a release too?
tulip sleet
I thought 22.3.0 is now compatible with click 8.1.0?
or black has to update?
I am not following that closely
onyx hinge
Oh sorry, names... It's typer that also would need a release
That's how click is brought into cascadetoml
onyx hinge
Okay that all may need updated too
I can look after dinner if it is still problematic
So many interrelated parts
tulip sleet
I'll see, and see if typer is on the ball about this or not
I see you already have followed this: https://github.com/tiangolo/typer/pull/375
ok, got it, typer has not had a release since 2021
onyx hinge
to be fair it's barely 2022 by months
but yeah it is better to plan for resolution that is not swift, imo
tulip sleet
there are two pages of PR's that have not even been released to run CI
onyx hinge
P(each networked step completes successfully) has to be pretty high when you have hundreds of networked steps per CI run. ugh.
E: Failed to fetch http://azure.archive.ubuntu.com/ubuntu/pool/main/libc/libcroco/libcroco3_0.6.13-1_amd64.deb 503 Service Unavailable [IP: 52.147.219.192 80]
tulip sleet
that particular IP is a very typical point of failure
onyx hinge
merge if ^^mostly^^ green 😕
tulip sleet
green with spots
mental nexus
I’m stuck and could use some guidance on allocating memory on the ESP32-S3 in CircuitPython. I think that CP grabs all the PSRAM for the heap, so when I go use heap_caps_aligned_calloc to create a big display framebuffer, it causes a crash.
Is there a “CP-friendly” way to malloc and get some SPI-PSRAM for a framebuffer?
crimson ferry
@mental nexus what size memory chunk do you need?
crimson ferry
oh, wow, yeah I don't know the psram side of things but my understanding is like you found that it's all cp heap
S3 has 250-300k free internal sram, but you need a lot
there must be a way to allocate from heap
mental nexus
In the RGB display code I think that the esp-idf uses low level code to try to allocate ram in the PSRAM, but CP allocated all of it.
Unfortunately that means I’ll need to make a separate version of the ESP-IDF. But if it works, I’ll do it.
Can I do just a normal malloc in CP for this large of a chunk?
crimson ferry
i don't know, jepler or danh or one of the other core devs would know if there's a way
pure malloc takes from esp-idf internal sram
keypad uses a structure called ring_buf, which uses gc_alloc, but I'm really reaching
mental nexus
Ok, and the code in the esp-idf uses some kind of “alignment” which I’m not exactly sure what that does.
Ok I’ll ping those two tomorrow during normal hours. Scott was giving me pointers but he’s going to be needing parenting pointers himself these days!
manic glacierBOT
solar whale
I finally had a need to use microcontroller.OnNextReset so I could reset to the Bootloader on a device that does not have easy access to the Reset Button -- worked perfectly!
tulip sleet
Ok, and the code in the esp-idf uses some kind of “alignment” which I’m not exac...
yes, you would allocate from the heap. The minimum size of a heap object might give you the alignment you want. If this is a fixed buffer, you could also allocate space before the heap is set up, using a storage_allocation
stuck elbow
I always use it, I have a 'bootloader.py' on my boards that I import from the repl, so that I don't have to type it all
crimson ferry
@mental nexus the Espressif example is (perhaps conveniently) 480 x 272 x 2 = 261,120 bytes, which fits in available internal SRAM on the S3 ...although it does look like they are allocating from PSRAM
proven garnet
@jaunty juniper I'm happy to rerun CI on the libraries I merged the other day to make sure the documentation updated, but don't know how to see that list, want to ping me on the ones that failed after merging?
gloomy shuttle
I always use it, I have a 'bootloader.py' on my boards that I import from the re...
I was looking to do this with a project where the board is not easily accessible. What code do you need to programmatically get back to the bootloader? Does this work on the rp2040 boards?
stuck elbow
import microcontroller
microcontroller.on_next_reset(microcontroller.RunMode.BOOTLOADER)
microcontroller.reset()
a bit of a mouthful
solar whale
I was looking to do this with a project where the board is not easily accessible...
it works fine on a feather rp2040 -- That is what I have been using. The example from @stuck elbow is what I am doing as well.
mental nexus
yes, you would allocate from the heap. The minimum size of a heap object might g...
Can you point me to an example of storage_allocation, I did a code search and found a couple of comments about it, but not exactly how to use it? To clarify, will that leave some memory space so that the ESP-IDF can access it?
tulip sleet
is the IDF doing the malloc, or do you hand it a buffer?
mental nexus
On a possibly-related note, I had to turn on a sdkconfig flag so that the ESP-IDF code will allocate from PSRAM CONFIG_SPIRAM_USE_MALLOC. But it raised the same kind of crash when I called the module espidf espidf.heap_caps_get_largest_free_block. I feel like I'm doing something wrong....
The IDF is doing the malloc.
tulip sleet
ah, ok, that's bad, I'm not sure how to get around that. Is there any other API for this functionality where you pass it the memory instead?
mental nexus
Not that I see. If I have to pass it some memory, I'll have to rewrite the ESP-IDF component code that I need, so that it will accept a buffer address.
I'm unsure if I can "fake" it by letting the IDF create a tiny dummy buffer, and then redefine it later in the CP core code.
tulip sleet
any malloc call is going to overlap with the heap. Or is there a way to specify the memory region for heap_caps_aligned_calloc() earlier, so that it respects some bounds?
we could pre-allocate a space for their heap, and for ours. I wonder if we are doing this already in some way. I would imagine there are other things that malloc stuff, and we figured out how to get around that, but I'm not familiar at all with storage allocation sharing with the IDF
mental nexus
Not sure I could call it earlier, since this would be happening whenever someone wants to instance one of these displays. If it grabbed memory before CP identified its heap, it would have to be "locked into" the CP build somehow.
tulip sleet
yes, I do mean that, you'd have to guess what size is reasonable to reserve
mental nexus
Doing it that way would mean the build would have to know it was going to have this specific kind of display attached. Doesn't seem very "circuitpython-y", if that's a word.
Any way to ask CP to shrink its heap to allocate some space for an IDF malloc call?
I don't really understand how CP manages memory and how it interact with the ESP-IDF. Any idea who I could ask about this?
tulip sleet
looking at supervisor/port.c for espressif:
// Heap sizes for when there is no external RAM for CircuitPython to use
// exclusively.
#ifdef CONFIG_IDF_TARGET_ESP32S2
#define HEAP_SIZE (48 * 1024)
#endif
#ifdef CONFIG_IDF_TARGET_ESP32S3
#define HEAP_SIZE (176 * 1024)
#endif
#ifdef CONFIG_IDF_TARGET_ESP32C3
#define HEAP_SIZE (88 * 1024)
#endif
Or would it be better to just go the route of letting CP allocate the space and pass it to the IDF?
tulip sleet
so maybe we use PSRAM all to ourselves
tulip sleet
I don't really understand how CP manages memory and how it interact with the ESP...
scott 🙂
mental nexus
Down lower there's a separate part for PSRAM.
tulip sleet
honestly, I would just be looking at the same code as you. Maybe @onyx hinge knows this off the bat.
I don't know why you are seeing crashes, because it looks to me that if there is PSRAM, and we use the whole thing, then you can ask their display code to allocate in regular RAM
onyx hinge
my impression is that we allocate the whole psram for circuitpython heap
mental nexus
I think there's something weird even in the baseline build, because gc.mem_free shows all 8.7MB the memory, but the espidf module memory size reader only shows 330kB.
onyx hinge
and leave the non-psram for the idf heap
tulip sleet
that's what I gleaned, thanks
mental nexus
Ok, that's what it seems like.
If I reduce the CP heap size, will that "automagically" leave it for the ESP-IDF to access?
tulip sleet
is it failing because there is not enough memory?
ESP_GOTO_ON_FALSE(rgb_panel->fb, ESP_ERR_NO_MEM, err, TAG, "no mem for frame buffer");
tulip sleet
If I reduce the CP heap size, will that "automagically" leave it for the ESP-ID...
does this board have PSRAM?
mental nexus
Yes, it has 8MB of PSRAM.
onyx hinge
for the other framebuffer display types (e.g., rgbmatrix), it has to be able to both allocate the framebuffer on the gc heap, and allocate the framebuffer as a "supervisor allocation". but both assume the circuitpython is in charge of the allocation in one way or another.
mental nexus
It crashes with a Guru Meditation Error: Core 1 panic'ed (LoadProhibited). Exception was unhandled.
onyx hinge
it is a bit messy, but it was the best compromise I could come up with since displays need to (A) be created before the VM starts, on particular boards and (B) the display needs to persist after the VM is reset
tulip sleet
so we use that for heap, so you should set it to use regular RAM, and it should just work, because we don't use the regular rAM heap
onyx hinge
a DEBUG=1 build might show better information on the debug console
mental nexus
I need more space than will fit in RAM.
tulip sleet
@onyx hinge what is happening here is that the esp_lcd code in the ESP-IDF is doing its own malloc
tulip sleet
I need more space than will fit in RAM.
ok, I did not realize that. Then I would just say that their API is defective/inflexible
mental nexus
I got an error of where the memory access caused a crash, EXCVADDR: 0xe0191061 LBEG . I'm suspecting this is in lala-land.
tulip sleet
right, i would not expect this to work at all
there is no way to get this to work without modifying their code. Did you look to see if there was an issue about this? Is anyone else asking for a more flexible API?
mental nexus
Sounds like I should find a way to malloc inside CP and hand it to the IDF. That means I'll have to diverge from the ESP-IDF to grab 768kB for my framebuffer.
I haven't seen any issues specifically related to LCD framebuffer malloc's, but I'll go look again broader.
tulip sleet
IF you can tell its malloc to use a certain section of PSRAM only, and same for us, then it would work, but that's static partitioning of the PSRAM
are you saying you would not use esp_lcd_new_rgb_panel() and just do it yourself?
mental nexus
are you saying you would not use `esp_lcd_new_rgb_panel()` and just do it yourse...
That's the question, if that's the only way is to rewrite that part of the IDF.
tulip sleet
we do fork the IDF to fix things that we are waiting on from them, we have always been able to go back to their repo eventually
mental nexus
With CP, if I ask for such a big chunk, can I guarantee that it will be in PSRAM?
tulip sleet
if we are using PSRAM for the heap, then it will be in PSRAM
I don't think we allocate some in RAM and some in PSRAM.
mental nexus
In CP, can I hook up the DMA and also call the CacheWriteBack, or will that cause problems? Also, the esp_lcd_new_rgb_panel() sets up some interrupts, will that be expected to conflict any with CP?
tulip sleet
i don't know, if we use interrupt allocation mechanisms from the IDF, then there should not be a conflict
mental nexus
Ok, let me explore rewriting the esp_lcd_new_rgb_panel() and see if I can at least get a hack to run.
Thanks for all the suggestions and sanity checks!
tulip sleet
sure, np!
tidal kiln
is randomly getting RuntimeError: ESP32 not responding after some period of time something with a known issue / suggested work around?
https://forums.adafruit.com/viewtopic.php?f=62&t=189313
tulip sleet
is randomly getting `RuntimeError: ESP32 not responding` after some period of ti...
i had not heard of this before it started cropping up, but I don't know whether it's related to CircuitPython core or not. The person could try rolling back to 7.1.0, say, and see whether they see the same problem. That would be a helpful diagnostic.
tidal kiln
@tulip sleet thanks. has any issue been created anywhere for tracking? or is it just discussion in various places?
tulip sleet
I don't know of one, so it would be good to search, both in the lib and in core.
dusk mauve
does anyone else have an ESP-BOX device on hand? I ran into a bug (filed as issue 6206) where wifi.radio.connect() triggers a crash. Would be helpful if someone else could do a fresh build and flash on their own device to make sure its not my build environment, maybe
dusk mauve
alternatively, if anyone with a known-good esp32-s3 dev environment could build a fresh espressif_esp32s3_box binary from main that might help rule out my dev environment 🙂
proven garnet
There's an issue for the requests library about multiple cookies, and after looking into it, it seems like the two solutions are to either add multiple cookies to the Set-Cookie header delimited by , or to implement Response.cookies. I'm not super duper knowledgeable on requests but willing to take it on but not sure which is the better option.
Well, for full CPython compatibility my understanding is that if Response.cookies is done the other should also be, but adding the cookies probably requires more infrastructure and I'm not sure what the constraint is on the library for minimizing memory usage.
crimson ferry
@tidal kiln fwiw, I do routinely get Timed out waiting for SPI char and other wifi-related exceptions on PyPortal. I don't use wifimanager, so the tactic after a couple of retries is esp.reset(). There's always reload and reset for further escalation.
tidal kiln
<@!337989520962093057> fwiw, I do routinely get `Timed out waiting for SPI char`...
thanks. dan suggested CP firmware also. so not sure if there's a fixable issue, or just a suggested way to deal with the behavior. i couldn't find any open issues related to this, so suggested the forum poster start one.
tulip sleet
<@!337989520962093057> fwiw, I do routinely get `Timed out waiting for SPI char`...
have you gotten this since forever? Did it coincide with a NINA-FW upgrade or antyhing like that?
crimson ferry
yes, since forever, but I am running 7.2.0, latest libs, and NINA 1.7.4
tulip sleet
and it happened in, say, 7.1.0 and before?
tidal kiln
maybe another one:
https://forums.adafruit.com/viewtopic.php?f=60&t=189735
crimson ferry
I don't know, I've just built up a series of exceptions to catch over the years, and a system of retries and escalations, so I don't really track it anymore in detail. I just check the serial periodically.
tidal kiln
if it's a "feature", then having a guide cover all that ^^ would be helpful
tulip sleet
the MatrixPortal has been around for a while, but we're seeing this more often? Or maybe people are just doing longer-living projects? I don't know.
crimson ferry
as far as I recall, these exceptions have been around for a long time, but they are intermittent and can usually be recovered
I have not seen hard crashes
idle owl
@proven garnet How are lib things going? Did the update work?
manic glacierBOT
crimson ferry
if it's a "feature", then having a guide cover all that ^^ would be helpful
I wish there was some canonical way to do these things. Several other people have reported in Discord that they have built up similar code for resilience (maybe KeithTheEE was one of them). There are probably some universal suggestions, and maybe examples that could be instructive, but each environment and code priorities are different.
In some applications I just disconnect wifi on espressif and esp32spi (and reset esp on esp32spi) between each transaction, so it's always clean. But still there are exceptions.
proven garnet
<@!135415127879974912> How are lib things going? Did the update work?
Yup yup yup!
Managed to rerun most of the failed PRs in the libraries
tidal kiln
I wish there was some canonical way to do these things. Several other people hav...
yep. totally agree. but at least having something that could be linked to for anyone that runs into this for the first time would be useful.
proven garnet
New black is angry with some libraries now and I've been submitting PRs for those but the error we were getting beforehand is gone and it's back to being sensible, fixable things!
idle owl
New black is angry with some libraries now and I've been submitting PRs for thos...
That was expected. At least on my part. Thanks for dealing with the fixes!
proven garnet
That was expected. At least on my part. Thanks for dealing with the fixes!
If there's a way to get a list of libraries that need touchups I'm happy to burn that down 😁
idle owl
If there's a way to get a list of libraries that need touchups I'm happy to burn...
Um... hmm. I don't think Adabot has a check for Black failing. And either way, you'd have to run it manually because the cron job only runs once / day to update circuitpython.org/contributing/library_infrastructure or whatever the URL is.
Let me ask Eva.
She updated Adabot recently.
@onyx hinge Looking at cascadetoml, it needs a setup.py for PyPI to work.
idle owl
If there's a way to get a list of libraries that need touchups I'm happy to burn...
Getting you a list of repos that failed their last Actions run now. So it's a place to start.
proven garnet
Getting you a list of repos that failed their last Actions run now. So it's a pl...
Sweeeeeeet
onyx hinge
Weird, it's been released before .
idle owl
Weird, it's been released before .
To PyPI?
onyx hinge
Yes
onyx hinge
idle owl
Hmm.
onyx hinge
Yes I'm pretty confident that Scott ran some commands on his local computer in the past
idle owl
Well, the only PyPI release thing I know how to do is add a workflow that runs to release it to PyPI.
jaunty juniper
pypi has different methods that I'm not well versed in, but it looks like cascadetoml uses the pyproject.toml file
idle owl
So if he did some other magic, I don't know how to make the repo do that.
OK looking into the pyproject.toml file.
jaunty juniper
not sure if that's the magic file, I find all that very confusing
onyx hinge
Me too
idle owl
Um... I think this is for running pip within the directory.
But I could certainly be reading it wrong
though immediately below it it says setup.py (legacy)
So I guess it is the file. I have absolutely no idea how to build that into a workflow.
I am evidently not the person to help you with this @onyx hinge
idle owl
I mean now I want to know how to do it.
But...
I'll give my self 30 minutes to look into it.
onyx hinge
You can totally leave it be for now
idle owl
@proven garnet Eva's working on fixing the Black failure check in Adabot. In the meantime, if you're looking for one to fix - here is a new lib that I'm doing the initial release stuff for now. https://github.com/adafruit/Adafruit_CircuitPython_TSC2007/runs/5728358600?check_suite_focus=true
proven garnet
<@!135415127879974912> Eva's working on fixing the Black failure check in Adabot...
on it
idle owl
<:KittyRedEyes:814486759383629865> on it
Even better, here is the list! https://gist.github.com/evaherrada/1b0ea3eebbf78cf2299496a29a4070bf
proven garnet
Even better, here is the list! https://gist.github.com/evaherrada/1b0ea3eebbf78c...
Want me to PR or patch to main?
idle owl
Want me to PR or patch to `main`?
Hmm. I'm going to say for this one, if it's because of the Black update, patch to main. If it's including anything else, PR. I'm always a little concerned about pushing directly to main, but that's a lot of PRs.
proven garnet
Sounds good, I'll only push a pure black reformat to main!
idle owl
Thanks!
solar whale
I just cloned a fresh copy of CP and am trying to build on MACOS -- when I try to build mpy-cross I am getting this ```jerryneedell@Mac-mini circuitpython % make -C mpy-cross
Use make V=1, make V=2 or set BUILD_VERBOSE similarly in your environment to increase build verbosity.
mkdir -p build/genhdr
GEN build/genhdr/moduledefs.h
QSTR updated
../py/vm.c:742:25: error: array index -3 refers past the last possible element for an array in 64-bit address space containing 64-bit (8-byte) elements (max possible 2305843009213693952 elements) [-Werror,-Warray-bounds]
sp[-MP_OBJ_ITER_BUF_NSLOTS + 1] = MP_OBJ_NULL;
^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~
../py/vm.c:269:13: note: array 'sp' declared here
mp_obj_t *sp = code_state->sp;
^
similar errors removed
4 errors generated.
make: *** [build/py/vm.o] Error 1
jerryneedell@Mac-mini circuitpython %
It builds fine on Linux.
onyx hinge
@solar whale micropython has already selectively disabled -Warray-bounds diagnostics for the "javascript port" I found this comment .. ports/javascript/Makefile:# Disable errors for array-bounds warnings on "sp[-MP_OBJ_ITER_BUF_NSLOTS + 2]" access. ports/javascript/Makefile:$(BUILD)/py/vm.o: CFLAGS += -Wno-error=array-bounds so I imagine it'll turn out that macs need the same treatment for some reason.
idle owl
@proven garnet I have a mission for you, should you choose to accept it.
solar whale
<@!343570430042308609> micropython has already selectively disabled -Warray-boun...
Ah -- Thanks -- is that new? I'll look into it more. or give up on building on MAc!
onyx hinge
@solar whale I assume that mac changed
likely the c compiler was updated or something
solar whale
Thanks for the pointer to the issue...
onyx hinge
you should feel free to file an issue, it'll affect us sooner or later in ci builds
diff --git a/mpy-cross/mpy-cross.mk b/mpy-cross/mpy-cross.mk
index 642fe9b5a..c7bd91ad1 100644
--- a/mpy-cross/mpy-cross.mk
+++ b/mpy-cross/mpy-cross.mk
@@ -29,6 +29,8 @@ INC += -I$(BUILD)
# compiler settings
CWARN = -Wall -Werror
CWARN += -Wpointer-arith -Wuninitialized
+# Disable errors for array-bounds warnings on "sp[-MP_OBJ_ITER_BUF_NSLOTS + 2]" access
+CWARN += -Wno-array-bounds
CFLAGS = $(INC) $(CWARN) -std=gnu99 $(CFLAGS_MOD) $(COPT) $(CFLAGS_EXTRA)
CFLAGS += -fdata-sections -ffunction-sections -fno-asynchronous-unwind-tables
``` you could also try with this change manually appliedThank you -- I was also curious if anyone else had run into this ....
I though I had messed up my system -- hence the re-install of brew so I thought it was just something I did.
I have to go offline now, but I will try your suggested fix in the morning and file an issue if it works or I guess if not as well...
jaunty juniper
I haven't run into this, but I'm still on Catalina
so like, not relevant at all (sorry)
warm flame
Anyone knows what the beta status of the ESP32-c3 really mean for dev ?
idle owl
@warm flame It means that likely not all the features are fully implemented, and there may be instability.
But we wanted to get it out there for folks to use, so we call it beta to be clear that it's not 100% ready.
warm flame
Great
Just found the lilygo t-oi plus that is made of a ESP32-C3 and has all the features I want
I know it's not the rp2040 I wanted but it got the size I need
idle owl
I'm uncertain of the specifics for the C3 build, i.e. I don't know specifically what features aren't completely ready or what makes it unstable. But it's definitely worth trying. I'm also not sure if we have it specifically for that board. Meaning anything you load on it may not have the proper pins available, if it's not specific to that board.
warm flame
Then I will have to test all the pins
idle owl
I believe we have a guide on adding a board to CircuitPython. If you're up for it, you're welcome to put together a board definition for it and submit a PR to CircuitPython. (I have no idea what your Git/GitHub skill level is.)
warm flame
I will have to wait this supper
summer
Industrial Automation advanced programming is taking most of my time
idle owl
No rush! We'll be here when you're ready. 🙂
proven garnet
@idle owl I was doing these manually before I remember that computers are a thing. Automating the bulk of checking and commiting the pre-commit reformatting errors, but I'll go through and manually check them and push them tonight for repos that don't require anything else
idle owl
Great!
tulip sleet
@onyx hinge re unicorn-ing CI status pages, I just got this:
Our engineering team deployed a fix for this behavior today (March 29th), so we expect no further page timeouts to occur when accessing Actions runs.
idle owl
@proven garnet I have a different mission for you, if you're up for it.
proven garnet
<@!135415127879974912> I have a different mission for you, if you're up for it.
I'm all ears
idle owl
<a:eyesshaking:662645652668547083>
This is a creepy emoji 😄
idle owl
I'm all ears
Ok, we have made a number of updates to the libraries over the last however long, and we've been really bad about updating cookiecutter. Your mission, should you choose to accept it, is to run a diff on the CI files in the generated cookiecutter content (i.e. not the CI for cookiecutter itself) against an updated library CI file set, and see if there's anything we missed.
proven garnet
Ooo sounds fun, yeah I can give that a shot
idle owl
Ok great! It's mostly CI things, but I'm about to update a badge, which I can submit a PR for.
But we kind of got behind with updating cookiecutter in general.
proven garnet
I'll try to generate a list of repos and maybe some key differences ASAP if that helps
idle owl
The repos should, in theory, all be the same. But I guess some of the most recent ones might not be. 😕
A list would be good too.
proven garnet
Wait, are you asking to see if the generated cookie cutter CI files match what were actually using?
idle owl
Wait, are you asking to see if the generated cookie cutter CI files match what w...
Yeah. So the CI files aren't specific to the library.
idle owl
So you should be able to diff them against an older library without generating anything with cookiecutter to check.
They're in the generating directories, but they should be standard.
proven garnet
Got it, will report back when I get a chance to start it
idle owl
Thanks!
jaunty juniper
speaking of cookiecutter I remember updating an outdated gitignore in adafruit_debouncer
tulip sleet
why are you freaking out?
proven garnet
Thanks for the reply @idle owl, I submitted an alternative PR but no worries about getting to it tonight. I'll continue patching everything else in the meantime. If that PR is the correct solution, it's probably what's wrong with the ones that need more than black rerun, is my guess.
Most of these new black reformatting needs are either changing:
x ** y
to
x**y
or stripping spaces on either side of docstrings like " Leading space would be removed"
tulip sleet
@idle owl @jaunty juniper @onyx hinge Re pyproject.toml: it is more modern. This is a good explanation: https://stackoverflow.com/questions/62983756/what-is-pyproject-toml-file-for. When I looked into it a while ago, its one deficiency seems to be what is mentioned several places there: that it didn't (originally) support "editable installs" (pip install -e .), which make it easy to develop a project that is locally pip-installed. However, see a new comment that says:
Update 2022: PEP 660 is implemented in pip (from 21.3.0) and some backends, e.g. flit. However no progress in setuptools. –
kap
Jan 8 at 13:50
PEP-660 is "editable install support for pyproject.toml". I think when I last looked, that was not true. So maybe pyproject.toml is ready for prime time now, and might be easier to maintain
lone sandalBOT
warm flame
why are you freaking out?
The unknown 😜
manic glacierBOT
solar whale
```patch
diff --git a/mpy-cross/mpy-cross.mk b/mpy-cross/mpy-cross.mk
index 642f...
This patch does fix the problem for me. I'm still a bit puzzled why no one else is running into this. I'll open an issue.
manic glacierBOT
slim portal
Is there also a Discord for MicroPython? What is used the most CircuitPython or MicroPython and what are the main differences?
solar whale
What version of black should we be using -- I am running into this conflict with the required version of "click" for CP ```Collecting click==7.1.2
Downloading click-7.1.2-py2.py3-none-any.whl (82 kB)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 82.8/82.8 KB 1.1 MB/s eta 0:00:00
Installing collected packages: click
Attempting uninstall: click
Found existing installation: click 8.0.4
Uninstalling click-8.0.4:
Successfully uninstalled click-8.0.4
DEPRECATION: Configuring installation scheme with distutils config files is deprecated and will no longer work in the near future. If you are using a Homebrew or Linuxbrew Python, please see discussion at https://github.com/Homebrew/homebrew-core/issues/76621
DEPRECATION: Configuring installation scheme with distutils config files is deprecated and will no longer work in the near future. If you are using a Homebrew or Linuxbrew Python, please see discussion at https://github.com/Homebrew/homebrew-core/issues/76621
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
black 22.3.0 requires click>=8.0.0, but you have click 7.1.2 which is incompatible.
Successfully installed click-7.1.2
I can build CP, but I can't run black...
jaunty juniper
What version of black should we be using -- I am running into this conflict with...
when do you get that ?
what could be the source of click==7.1.2
solar whale
It is part of the requirements for building CP
sudo apt install python3-pip
# Install needed Python packages from pypi.org.
pip3 install -r requirements-dev.txt
# Force a particular version of click. It does not get updated to the
# proper version sometimes
pip3 install --upgrade click==7.1.2```yes -- I see that on Linux black is version 20.8b1
It won't work with the most recent click....at least, I don't think so.
jaunty juniper
@tulip sleet we need to remove click==7.1.2 in the "Building Circuitpython" guide
yeah the requirements-dev.txt pins click to 8.0.4 now
solar whale
ah -- I'll try that
I see - it was 8.0.4 then it went back to 7.1.2
now black is happy
manic glacierBOT
solar whale
and I can still build CP
jaunty juniper
Is there also a Discord for MicroPython? What is used the most CircuitPython or ...
there's a page about the differences:
https://docs.circuitpython.org/en/latest/README.html#differences-from-micropython
micropython has a forum (and a... Slack I think ?)
https://micropython.org/ https://forum.micropython.org/
solar whale
@jaunty juniper Thanks -- there are so many things to keep track of in the build process....
slim portal
there's a page about the differences:
<https://docs.circuitpython.org/en/latest/...
Ah thanks, so looking at this it doesn't really matter for a beginning user who wants to try MicroPython whether he uses the Adafruit fork or the original one
Do you run out of RAM quickly when creating a larger MicroPython file? Or can you do quite some larger projects too
stuck elbow
we don't really have much MicroPython experience here
you can compile files to .mpy to avoid memory overhead of parsing them
and of course there is a huge difference in memory available between samd21 and, say, rp2040
jaunty juniper
personally I really prefer the CP workflow to MP, having to upload files through the same serial as the REPL is a constant annoyance in MP, and I don't find the dedicated tooling like Thonny to be very good, but other than that it's perfectly fine and it might have features that CP doesn't, or match your workflow better
slim portal
Thanks for the answers! (I notice I'm in the wrong channel so I will move to #help-with-circuitpython for further questions 🙂 )
tulip sleet
<@!329766224093249548> we need to remove click==7.1.2 in the "Building Circuitp...
Thanks, I have removed that specific line and am now asking the user to do pip3 install --upgrade -r requirements-dev.txt, which should do the right thing.
solar whale
On Linux I ran iton an issue with an old version of gitlint -- still wanting click 7.1.2 but I just did pip install --upgrade gitlint and it is now happy -- does the requirements-dev.txt also update gitlint?
onyx hinge
@solar whale mpy-cross uses "whatever the system compiler for the local CPU is", so there's a great variety. you just got an upgrade or something, that nobody else had reported yet. doesn't mean nobody else had a problem yet, either.
solar whale
Life was simple when everyone used MS-DOS and Borland C....
Not really -- it was a nightmare then as well 😉
especially when you had to reload the complier from 31 floppy disks...
stuck elbow
let's face it, computers are a mistake
solar whale
or at least -- using computers isa mistake...
tulip sleet
On Linux I ran iton an issue with an old version of gitlint -- still wanting cli...
We don't use gitlint, so it's not in any of the requirements lists.
stuck elbow
or at least -- using computers isa mistake...
we are using them wrong
solar whale
We don't use gitlint, so it's not in any of the requirements lists.
ah -- OK -- I ran into a few pip packages that complained about the upgraded requirements, but I'll wait to see if/when I ever use them to deal with them -- Probably from some old project....
tulip sleet
it would be nice if there was a simple --upgrade-all for pip, but they don't have that. Same for pip list; I have no idea why you can't say pip list click and find out just about click
solar whale
pyocd and awscli complained
But I don't know why I have them installed
they were well out of date -- upgraded and everything seems happy -- or happier
proven garnet
@idle owl, @trim elm libraries are all reformatted! I'll do that few that need other touchups next.
proven garnet
Also, if anyone knows MIDI, feel free to chime in on this PR, I know only what I've Googled so a second set of eyes is appreciated! https://github.com/adafruit/Adafruit_CircuitPython_MIDI/pull/49
winter mortar
Also, if anyone knows MIDI, feel free to chime in on this PR, I know only what I...
I threw in my 2 cents (or CCs).
dusty spade
Anybody use the circuitpython extsnion for VSCode? I can't get it to run any of the commands from the command palette. Is there some trick to setting it up other than just installing the extension?
proven garnet
I threw in my 2 cents (or CCs).
Thanks!
proven garnet
Anybody use the circuitpython extsnion for VSCode? I can't get it to run any of ...
I do, works fantastic for me on Windows, hurts a lot when on Linux
Was plug and play for the most part on Windows. Sometimes I have to close the window and reopen it to get it to work though.
dusty spade
Well, I am on Ubuntu and the command palette items for the extension don;t work
dusk gull
I am trying to gather up information on using external clocks, and I am trying to understand why the focus has been on SAM OSC32 source and not the OSC0 and OSC1 sources. In my experience these are the main ones used and the 32K is only when you need higher precision RTC. I am worried I am missing something and I want to make sure I understand. A good example is the Feather M4 CAN, which higher speed CAN (which is how it is mostly used now) want/requires something better than the typical 2% internal clock source. But it does not have an external crystal or oscillator for the OSC0 or OSC1, but it does have a 32K.
idle owl
PEP-660 is "editable install support for pyproject.toml". I think when I last lo...
It straight up says setup.py is legacy now. So... Ease of maintenance doesn't really enter into it. We should probably look into updating.
idle owl
Thanks for the reply <@!330227457296957440>, I submitted an alternative PR but n...
No worries. Thanks for thinking to add it!
idle owl
<@330227457296957440>, <@362023826105827340> libraries are all reformatted! I'l...
Thank you!
tulip sleet
It straight up says setup.py is legacy now. So... Ease of maintenance doesn't re...
sounds good. it's really hard to find out what is current practice right now. There's so much legacy stuff. There should be a page python.org/best-practices-NOW
idle owl
sounds good. it's really hard to find out what is current practice right now. Th...
Agreed. Otherwise it's all over the place to find anything useful. I need to figure out how it works first, but It's probably worth doing.
proven garnet
That's how I feel about packaging as a whole. I'm trying to learn more and get into it, but sometimes it feels like a hodgepodge of different solutions 
idle owl
Packaging is its own special creature.
misty garnet
@dusty spade like Tekktrik, I often have to close and reopen the program for the CircuitPython commands to work properly in Windows
idle owl
@tulip sleet Is it possible to successfully demonstrate asyncio with two NeoPixel rings and one button? Instead of multiple buttons and one ring.
Feels like maybe not.
tulip sleet
sure, you can make two neopixel tasks. They just have to async.wait(0) frequently
idle owl
That would simplify the wiring setup, because a bunch of these boards have one built in button.
OK, thanks. I will almost certainly need help with the code, but I'm not quite there yet.
Not sure if I've ever soldered headers onto a QT Py RP2040.
idle owl
@tulip sleet I still don't have headers or rings hooked up (waiting on a meeting that apparently might not be happening... the other person is now 25 minutes late). But I have a question - if I'm using only one button, do I still want to use keypad for its other features? Or do I not bother at that point? The pressed/released feature might come in handy.
tulip sleet
<@!329766224093249548> I still don't have headers or rings hooked up (waiting on...
keypad even with just one button is a convenient way of doing debouncing, and sorting out up/down with no work. That's how I think of it.
idle owl
keypad even with just one button is a convenient way of doing debouncing, and so...
Ok, that's kind of how I was thinking of it, but wasn't sure I was right in that. Thanks!
idle owl
Oof. Forgot to tone down the brightness, because I was initially working with the onboard NeoPixel... just got two full-brightness 16-pixel rings to the face.
goes back to code to back it down
blissful pollen
Ouch, funny enough I just had 1 full brightness 16 pixel ring to my face and that's bad enough
warm flame
I just ordered a dev board with a esp32-c3 on it. And 5 t-oi lilygo (20 to 40 days shipping)
I'll get the dev board by tomorroe
Tomorrow
idle owl
@tulip sleet I made asyncio do different things on two rings that changes while the button is held!
No idea if I did it properly.
But I made it do Things.
Rainbow and blink: button press rainbow reverses and blink speeds up, button release rainbow goes back to forward and blink slows back down
blissful pollen
One thing I think could be helpful in a guide is showing people how to stop one or more tasks and move to new tasks. Like a button to move from phase 1 to phase 2. It's not too hard but took me a bit to find
idle owl
One thing I think could be helpful in a guide is showing people how to stop one ...
This is going to be an asyncio template to go in all guides. (Which I just realised means I can't use the built-in button, or I need to make two templates.) I don't quite understand what you mean, but I am super not familiar with asyncio, which is to say, I basically know nothing. If it makes sense to put in this template, I'm happy to add it. But I think this is supposed to be simple, so it might make more sense in Dan's guide.
Or one template without embedded code, and you have to choose the right example to use an external button versus a built-in button. Anyway, this is template logistics, not specific to asyncio.
blissful pollen
This is going to be an asyncio template to go in all guides. (Which I just reali...
Yeah an advanced guide may make more sense. Basically in a couple projects I have had a "setup" phase and then a "running" phase. Where a button press canceled the setup tasks and started the main running tasks
idle owl
Yeah an advanced guide may make more sense. Basically in a couple projects I hav...
Fancy. Dan's guide has a lot of asyncio stuff in a guide already. Probably it would go there. https://learn.adafruit.com/cooperative-multitasking-in-circuitpython-with-asyncio/overview
blissful pollen
if you ever need some examples or such just let me know. I'm still loving asyncio
jaunty juniper
I did a test case of running a series of sounds, and pressing a button cancels them. I don't know if task cancelation is supposed to be used/work like that, I was trying to learn about it.
Well it's actually a mock-up, it prints instead of playing sounds
idle owl
if you ever need some examples or such just let me know. I'm still loving asynci...
Definitely going to want a set of eyes on this code. Dan will, I'm sure, take a look, but I'll take you up on it too, if you're up for it.
idle owl
I did a test case of running a series of sounds, and pressing a button cancels t...
Oooh. Unpublished page in Dan's guide on Canceling Tasks.
Because I bastardised this code pretty hard to make it do things. There might be WAY better ways to do this.
blissful pollen
Definitely going to want a set of eyes on this code. Dan will, I'm sure, take a ...
Sure thing. Never mind reviewing code / tutorials
idle owl
Thanks!
blissful pollen
I did a test case of running a series of sounds, and pressing a button cancels t...
Looks similar to what I did too!
idle owl
In a meeting, or I'd send the code now. It's currently working, and I should probably show it off.
proven garnet
@idle owl I'm prepping for the cookiecutter activity. Do you want diffs on just the workflows in the .github folder, or are you looking for more extensive diffs on all the library infrastructure like .pre-commit-config.yaml and things like that?
idle owl
<@!330227457296957440> I'm prepping for the cookiecutter activity. Do you want ...
All the lib infrastructure, please! Hopefully the diffs aren't major, but we'll find out.
@blissful pollen Limor likes the demo. So I should plan to stick with the practical functionality. But that obviously doesn't mean the code can't be changed.
Need to dinner soon though.
idle owl
Also I realise I could have done that with the LED Animation library, but Limor really wanted a raw NeoPixel asyncio example with two NeoPixel things happening on two separate rings or whatever.
Ok, afk for a bit at least.
proven garnet
Got all the git diffs on these libraries, now to just parse them
Hopefully faster. Maybe. 
timber mango
hi I am working with the ST25DV16K, this Adafruit board is wonderful, but it turns out that I have problems to read with android cell phones, it does not recognize the message and says it is empty. But with the Iphone it works, could someone help me to see what the problem is?
timber mango
i found 🙂
stuck elbow
what was it?
marble hornet
@mental nexus , saw your stemma qt socket on the hackaday news letter, very cool!! (And congrats on getting featured😊)
blissful pollen
<@!228322543507865601> Limor likes the demo. So I should plan to stick with the ...
Looks good to me. Matches what I have done in the past.
I wonder if a note or something about "what if my program ends when you said it runs forever". I have a feeling a mistake that will get made a lot is where a coroutine ends (say the while True look is forgotten) and then things stop. I just foresee that question coming up a lot.
proven garnet
@idle owl I wrapped up the diff comparison! I have the results, but maybe you want to meet briefly either in a voice or thread channel today to go over it so I can communicate it better?
Edited but also happy to do a a dog comparison, honestly 
onyx hinge
@tulip sleet good morning! could we put an alternative to 6214 in 7.2, with just the change to ringbuf_free and keep the rest for main as cleanups? That's if https://github.com/adafruit/circuitpython/commit/9ee6ec66fb50c2ddd8562273a0b73fa013df1bd8 on its own resolves the crash.
tulip sleet
<@!329766224093249548> good morning! could we put an alternative to 6214 in 7.2,...
That is OK. Some of the other changes do reflect actual bugs, but they are less important. I will make a new PR, and repoint the current PR to main (I think I can do that).
I think the main other bug is in the broadcom port, which is not crucial to fix at all
maybe also stm
onyx hinge
oh the non-use of "ringbuf_init"?
tulip sleet
no, missing checks of allocated_buffer
onyx hinge
I wonder why ringbuf_free has to do anything, it could just NULL the pointer and let GC take care of it.
tulip sleet
it's not re-running when I re-point the base. I don't trust this to work right. I will put it back, close it, and do cherry picks
it could, I was thinking of that, but I thought not to do it. Let me remember why, which may take a few minutes
onyx hinge
OK, and maybe you can try again to tell me why allocated_ringbuf is needed, when there's r->heap, because I have still not understood it I guess.
tulip sleet
@onyx hinge let's talk about this in a few minutes; I need to eat breakfast
mental nexus
<@!366059101203202052> , saw your stemma qt socket on the hackaday news letter, ...
Thanks, just a simple design but maybe folks can find some use for it. Hope all is well with you and your projects, including the TG-Watch.
tulip sleet
@onyx hinge ok ready to talk, audio channel?
onyx hinge
sure
timber mango
hi I am working with the ST25DV16K, this Adafruit board is wonderful, but it tur...
Could anyone help me with ST25DV16K NFC tag? in the beginning it all works fine as I tried to write information like url and text with an NFC app on my iphone, but then at some point the NFC stops working and the app says the tag is not supported, So I guess i made it corrupted somehow, is there anyway to fix it?
timber mango
what was it?
I am sending data to the NFC st25dv16k , to then read it with the cell phone, all the time I read it with the iphone 10, and I had no problem, but then I tried it with the android and it didn't work. I tried to find the error but it was not easy. In the end I added a space to the url and then it works for both iphone and android.
timber mango
Could anyone help me with ST25DV16K NFC tag? in the beginning it all works fine ...
I think you must write in NDEF format, I have not taken the apps to write data to the board via RF, I have done it with the I2C. With the I2C I must write in ISO 15693 form.
tulip sleet
@onyx hinge nothing is ever easy. It turns out the static board. objects were not being gc'd.
not in the root pointers list or part of a _gc_collect() routine
onyx hinge
uh oh, so the buffers could get freed anyway? ouch
tulip sleet
yes, and I managed to hit that while testing
onyx hinge
hooray for testing
it would have been easy to write a test program that didn't generate enough garbage, or that you wouldn't notice 😕
onyx hinge
https://blog.adafruit.com/2022/03/31/compose-with-markdown-in-google-docs-on-web-google/ hum I wonder if this'll complicate our deliberate insertion of markdown in the notes document
crimson ferry
When we freeze in libraries, I wonder if there are ways we can advertise this to- new and existing users... update the guides, include a statement on the libraries page on .org , readthedocs, other?
tulip sleet
When we freeze in libraries, I wonder if there are ways we can advertise this to...
you mean have a list of frozen libraries for each board?
like we have a list of included native modules?
crimson ferry
yes, there was some confusion helping someone with a PyPortal (and I didn't even know my PyPortal had frozen libs)
and it has changed over time, so that's going to be a point of confusion
tulip sleet
I think there may even be an issue for this in circuitpython-org or circuitpython
crimson ferry
OK, thanks, I'll look
we need to train people about putting libraries in root for library features newer than the circuitpython release
idle owl
Looks good to me. Matches what I have done in the past.
I wonder if a note or so...
Thank you! I'll make a note to make a note. 🙂
idle owl
<@330227457296957440> I wrapped up the diff comparison! I have the results, but...
Hah! I saw dog, but knew what you meant. Sure, we can chat today.
@tulip sleet I guess it's probably easiest if I make a PR to Learn with my asyncio code? Then you can literally review it instead of me simply pushing it to my fork and you telling me what you think (which was my first plan).
tulip sleet
<@!329766224093249548> I guess it's probably easiest if I make a PR to Learn wit...
sure. in the code you did above, the only thing I saw was that buttons was plural even though there was only one button, but you may have fixed that.
idle owl
sure. in the code you did above, the only thing I saw was that buttons was plura...
Oh. I had it button but then added controls back into it, and made it plural again. But I guess controls isn't a button. so I can put it back.
Thanks!
proven garnet
Hah! I saw dog, but knew what you meant. Sure, we can chat today.
Are you free now or is later a better time?
idle owl
Are you free now or is later a better time?
Now works. Do you need to share your screen? I think we disabled that in the CircuitPython voice channel.
proven garnet
No, I uploaded everything needed to a public folder in my Google drive
I'll get setup
idle owl
Nice ok
idle owl
@proven garnet So we're not sure what level of access is needed to run patches. Good chance write access is enough. But easy enough to find out by having you try. So we'll teach you how to do it, and have you give it a try. Might need admin access though, in which case it will fail for you. 🙂 We'll set up a time soon, maybe once you have the list of patches ready.
@tulip sleet I'm documenting this code. How do I describe Controls? py class Controls: def __init__(self): self.reverse = False self.wait = 0.0 self.delay = 0.5
tulip sleet
@onyx hinge reason for PR delay: the problem I am seeing is not GC, it's that the default UART tx pin is pin 0 😕 😬
idle owl
Also, am I right in thinking that even an external button doesn't need digitalio if using keypad?
tulip sleet
right, you give keypad pins, not digitalio's
idle owl
perfect. Makes this template code a lot easier to sort out.
Thoughts on documenting Controls?
tulip sleet
it is the "knobs" you can vary to change the animation. You could call it AnimationControls, if that makes it clearer
idle owl
That would be good. Thank you.
proven garnet
<@!135415127879974912> So we're not sure what level of access is needed to run p...
Sounds like a plan!
idle owl
@tulip sleet I tagged you on this PR since Limor wanted you to review it. https://github.com/adafruit/Adafruit_Learning_System_Guides/pull/2129 It's now documented and some variable names are updated, all tested.
tulip sleet
<@!329766224093249548> I tagged you on this PR since Limor wanted you to review ...
? REMOVE THIS LINE AND THE REST OF THIS DOCSTRING BEFORE SUBMITTING TO LEARN
idle owl
? `REMOVE THIS LINE AND THE REST OF THIS DOCSTRING BEFORE SUBMITTING TO LEARN`
It's how I document the docstrings in template code where I explain how to update a piece of the code to match the template. That applies to everyone who uses it moving forward, not to the initial submission.
tulip sleet
ah got it
idle owl
So, when the template code is submitted for the template usage, it doesn't have instructions in it.
tulip sleet
@idle owl on CPB, the builtin buttons are active high, is that right (I remember that from CPX, maybe I'm wrong)
idle owl
<@!330227457296957440> on CPB, the builtin buttons are active high, is that righ...
Oh right. Hmm.
Is that the only board for which that is true?
I can make two examples, and refer to both in one template, as a "Choose the one that matches your buttons"
I feel like FunHouse is like that or something.
tulip sleet
I dunno, maybe check magtag? Not sure what else
idle owl
Maybe I'm wrong.
tulip sleet
right maybe funhouse
idle owl
Ok. I'll look into it and create a second example.
tulip sleet
so it changes only while the button is held down
for that, you don't even need debouncing, cause there's no real state
idle owl
I'm confused then
tulip sleet
but maybe leave in because it's more transferable to other examples
idle owl
keypad?
tulip sleet
yes
onyx hinge
<@!426200547024961539> reason for PR delay: the problem I am seeing is not GC, i...
is that an additional problem, or were you mistaken about the gc problem?
idle owl
I mean, I'm probably the only person using this, so I can probably add an instruction in the docstring to change it for certain boards.
And not have whole second example.
tulip sleet
right, that's for CPB. but I'm talking about the example in general. It does slow reverse only while the button is held down
lines 65-70
tulip sleet
is that an additional problem, or were you mistaken about the gc problem?
The GC was probably not wrong, but it wasn't very general
idle owl
I totally agree leaving it in makes more sense so folks can add more buttons more easily. Or whatever.
tulip sleet
oh you mean this in general didn't need keypad?
in this particular case, yes, though maybe without deboucning it would stutter a bit. ok to leave in. Or the example could have a toggle boolean
press once to reverse. Press again to go back to original state
tulip sleet
I think so too
less complicated, I just wanted to make sure the intent is that you have to hold the button down
idle owl
Yeah that's what I was doing.
(I'm adding more instructions to the example though for active high buttons)
tulip sleet
is that an additional problem, or were you mistaken about the gc problem?
The GC now assumes that only UART needs GC, and doesn't try to GC the I2C or SPI objects
I will not necessarily fix that now. I'm still trying to figure out why GPIO0 for TX is making it crash hard
I don't see any obvious problems. I did find a bug in .in_waiting that didn't re-enable the IRQ when trying to guard the ringbuf. Looks like a copy pasta bug
oh, wait, yes, there is a bug... if there is more than one UART it will not GC the second one
idle owl
OK so wait, if setting up the button is a Pull.DOWN, is it active low? The FunHouse lib is Pull.UP, magtag and CPB are Pull.DOWN. So I am a bit confused.
tulip sleet
OK so wait, if setting up the button is a `Pull.DOWN`, is it active low? The Fun...
if it's pull.down, then it's default low, like CPX. So magtag, CPB, and CPX have active high buttons
if you ground the button when you it's pressed, then you would use pull.up
tulip sleet
i do a mini-rederive in my head every time. If the pin is not pushed, then it is unconnected, so the pull should be opposite of what the other end of the button is connected to
so if the button is connected to ground (low), then without being pushed, it should be default high
basically I am thinking voltages
idle owl
hmm ok
tulip sleet
the pull gets overriden by the hard "pull" of closing the button when it's pushed. Pushing the button is like connecting a zero-ohm resistor to either ground or +v
the pin pull is a non-zero resistance
more current goes through the zero-ohm resistor, so it wins
gloomy shuttle
https://blog.adafruit.com/2022/03/31/compose-with-markdown-in-google-docs-on-web...
I think you would have to turn it on as its opt-in
tulip sleet
@onyx hinge see slack if you have not already
onyx hinge
thanks, answered
idle owl
@tulip sleet Instructions updated. https://github.com/adafruit/Adafruit_Learning_System_Guides/pull/2129
tulip sleet
@idle owl maybe we need board.BUILTIN_BUTTONS_ACTIVE_HIGH True or False or something like that. Or the active low buttons should be called BUTTON_INVERTED
idle owl
Hmm.
stuck elbow
board.NOT_BUTTON
idle owl
That could apply to literally every other pin.
tulip sleet
If the former, then you can query the board and set your pulls appopriately.
stuck elbow
board.MAYBE_BUTTON
tulip sleet
board.UPSIDE_DOWN_BUTTON
stuck elbow
UNBUTTON
tulip sleet
boards.BUTTONS_ARE_OPPOSITE_OF_WHAT_YOU_THINK
jaunty juniper
OK so wait, if setting up the button is a `Pull.DOWN`, is it active low? The Fun...
(I think you swapped them here)
idle owl
(I think you swapped them here)
That's what I got out of the applicable libraries for each board. But my understanding of what's active vs default was backwards yes.
jaunty juniper
the MagTag uses pull ups, but in the mean time we discovered that deep sleep wake-on-pin works better with pull down, which is why the FunHouse has pull downs
the pull is the rubber band that holds it in the default position until you activate it
idle owl
I also didn't know the term "default" was in the mix until today.
I'm hoping now having a term for both statuses will make it stick in my head.
Not holding my breath though.
jaunty juniper
but wait for it, if it has an external pull, it will overcome the internal one, and then the value of button.pull becomes irrelevant
idle owl
@tulip sleet Do you want to do an official review on that PR? Or not concerned about it?
crimson ferry
(assuming the external pull is lower resistance than the internal... likely)
tulip sleet
I was just finishing, multi-tasking too much
idle owl
Ok thanks, Anne was asking as she would merge it if needed.
onyx hinge
thanks Dan!
idle owl
@tulip sleet So..... I had a variation of this line of code (ring_two[:] = [(0, 0, abs(ring_two[0][2] - 255))] * num_pixels) from part of a guide explaining NeoPixel blink on a CPX to show the difference usually shown using standard red LED blink code and led.value = not led.value red LED blink code. But because the code was using NeoPixels, I felt the need to show it using NeoPixels. I, um... clearly had help writing that line of code, and had help updating it to blink a whole ring instead of a single LED. In the guide I pulled it from, I explained abs, but totally glossed over the code that cycles the blink animation. Which figures because I don't really understand it now, meaning I definitely didn't understand it then. So, I'd like your thoughts: Do I gloss again since the point of the example is to use asyncio, not explaining blink animations? Or should I explain it?
tulip sleet
You could do blink by filling with a color and then another color (or black). That would be a lot easier to comprehend
idle owl
You could do blink by filling with a color and then another color (or black). Th...
I tried that, and it didn't work. That was the first thing I did. Then used that line, and it worked.
I don't know why.
tulip sleet
it should work, there is some other issue, then
jaunty juniper
the issue is that you have to save the state to know what color to switch to
making it more complicated than not
idle owl
I must be doing it wrong.
tulip sleet
i have to be afk but will be back
idle owl
It's not blinking this time either.
Ok.
Added another show() and now it is mostly blue, and flickers off for a moment intermittently.
So it's not blinking really.
But at least it turned on....
jaunty juniper
I can take a look
jaunty juniper
👍
idle owl
Normally one show() in a single function works. I didn't expect to need two.
Or know they needed to be inside the asyncio.sleep() bits.
Pressing button successfully blinks faster.
Ok, I'll do this instead.
I like my fancy line of code, but don't relish explaining it.
jaunty juniper
I wouldn't expect to need two
idle owl
It doesn't work without both.
This works. py async def blink(animation_controls): """Blink animation on ring two.""" while True: ring_two.fill((0, 0, 255)) ring_two.show() await asyncio.sleep(animation_controls.delay) ring_two.fill((0, 0, 0)) ring_two.show() await asyncio.sleep(animation_controls.delay) await asyncio.sleep(animation_controls.wait)
Otherwise, it seems to ignore one part or the other.
jaunty juniper
oh ok yeah that's normal, i thought you were doing something similar to the previous version by "switching the color" in a single place
idle owl
No.
That's what I took out 😄
Because it's not as understandable.
This is "typical" blink code.
And therefore will not need a specific explanation.
jaunty juniper
yeah and that's the thing with asyncio, since it's its own loop on the side, you can write it as a linear code, without having to bother with having something else happen in-between
idle owl
hmm
I will almost certainly need help explaining the asyncio part of this example. I don't understand it at all.
Getting close to that point.
blissful pollen
Is there a specific part you're looking for help explaining?
idle owl
Well... I explained everything up to the async def rainbow_cycle(animation_controls:) I realised I was about to gloss over the async part of it, and say something like "here's the animation code" but that kind of defeats the purpose of the template.
Let me get the updated code onto GitHub so I can send you the latest version of it.
blissful pollen
Sure, reading back a bit I just got home
idle owl
Sure, reading back a bit I just got home
Weird I thought it would take me to my fork, but it does a weird thing within the destination repo.
idle owl
Is there a specific part you're looking for help explaining?
So... basically everything involving async. 😖 Dan has the guide he wrote, and I'm trying to read through it in hopes of explaining this example.
(Guide I'm referring to: https://learn.adafruit.com/cooperative-multitasking-in-circuitpython-with-asyncio/overview)
I might have to bail soon for dinner, I'm not sure what the plan is there.
In fact, let me afk for a moment to check to see if we came up with anything. BRB.
No plans. And everyone in the house is busy with something or other. So maybe it's snack dinner. 🙄
blissful pollen
No problem, I can put some thoughts down. I really liked Dan's guide (I learned from it to start). So it has lots of good info.
I think to break it into parts is good. You create tasks, you start co-routines and you pass control information around. Kinda backward to the layout of the code, as the tasks are created last in the code (but first when running if that makes sense).
idle owl
It doesn't make sense, but I can see it in the code.
I mean, it kinda makes sense.
But I can see it practically speaking in the code.
blissful pollen
The way I think of it, is when I create a task I start that function running. So I create a task for every function I want to run concurrently (glossing over the fact they're not actually running simultaneously). And then the asyncio.gather(...) function will run until all the tasks have finished (beginner guide so ignore that you could cancel tasks for now)
idle owl
Run and repeat, right?
blissful pollen
So the main function creates everything I want to run at the same time and then waits until each function finishes (or in this case never does finish)
The await keyword to me really just says "Hey I'm taking a break let someone else run for now". In most of the example it is calling sleep (sleep(0) when it just wants to give up control politely like giving someone else a turn vs actually needing to sleep)
blissful pollen
And probably worth mentioning putting async in front of the functions is just a way of saying "this function uses asyncio" (It probably means more technically - I don't know 🙂 )
idle owl
I like that explanation
@proven garnet Wow. That's massive.
@proven garnet Do you know what all of that is referring to or is a bunch of it out of your wheelhouse?
proven garnet
Haha yeah, a bunch of it probably comes from one-offs, I could do some more digging if need be
idle owl
Because, it would be worth dividing it up with comments as headers to sort of make it more readable. But only if you kind of know what everything does so you even know what headers to make.
Some of this can go, as well, imo. I'll review it tomorrow.
proven garnet
To be honest? Not much of it. Some things I know, others I could guess, and most is beyond me haha
proven garnet
Sounds good. Is the best way to PSA this via In The Weeds? Didn't think pinging was a good idea haha
idle owl
Sounds good. Is the best way to PSA this via In The Weeds? Didn't think pingin...
Yep, In the Weeds will work. I'll add it to the notes doc on Monday and we can discuss it then.
blissful pollen
brb
idle owl
@tulip sleet For when you return - Limor tagged you specifically on this PR after I tagged CPLibians. So I'm linking you here in case it's lost in your email because of the double review request. https://github.com/adafruit/Adafruit_Learning_System_Guides/pull/2130
idle owl
And probably worth mentioning putting `async` in front of the functions is just ...
Dan's guide says "every function or method that contains an await must be defined as async def, to indicate that it's a coroutine." So I guess that explains that.
Apparently I should read this more thoroughly. 😕
blissful pollen
Dan's guide says "every function or method that contains an `await` must be defi...
Yeah that makes sense.
It took me a while to really get it myself
idle owl
Also need to find out if I should be copying info from this guide into the template page, or if I'm simply glossing over most of it and linking to the guide at the end. Or something in between.
Will find out tomorrow.
@blissful pollen I pasted your notes into a notes section for myself to have for the guide template if it makes sense. Thank you for the explanations.
blissful pollen
<@!228322543507865601> I pasted your notes into a notes section for myself to ha...
Welcome! I am not quite sure if copying or glossing over it is best myself. There is a lot of info to cover either way so I'd almost lean to a quick explanation and the link.
idle owl
Agreed. The in betweeny method.
@blissful pollen Are the animation code blocks functions or methods? I'm.... fuzzy on the distinction.
blissful pollen
I believe they are functions as they do not belong with an object (class)
tulip sleet
<@!329766224093249548> For when you return - Limor tagged you specifically on th...
it works fine, right? I think it's a good example, showing how you can have multiple asyncio.sleep's
and the blinking is really obvious
idle owl
it works fine, right? I think it's a good example, showing how you can have mult...
Yes it works fine. Blinking away happily, and speeds up on button press.
Needed more show() in the right places, was why it didn't work before.
So it's a little different than normal when async is involved, apparently.
But I figured it out.
tulip sleet
if you replaced all those with time.sleep()'s, it would still blink
you just couldn't control it as easily
idle owl
exactly.
Which is almost what I did the very first try, but realised that was blocking and couldn't be the right way to do it.
So I pulled more from your code.
Feel free to merge the PR.
Thank you!
And I can forever be annoyed that I misspelled "simplify" in my commit message until I have to update the example again.
Should have amended it when I noticed, but eh. I let it ride.
Fixed the PR title anyway.
idle owl
@blissful pollen You said this the other day: I wonder if a note or something about "what if my program ends when you said it runs forever". I have a feeling a mistake that will get made a lot is where a coroutine ends (say the while True loop is forgotten) and then things stop. I just foresee that question coming up a lot. I'm wondering exactly how to word this note in the guide?
blissful pollen
Give me a minute I'll think about it
idle owl
@tulip sleet Thanks for the feedback on the .gitignore PR. I'm bringing it up In the Weeds on Monday. It is intended to be a group decision.
manic glacierBOT
idle owl
Give me a minute I'll think about it
Thanks. I have a notary coming in 5 minutes, then dinner was worked out, so I'll be AFK, probably for the night at that point anyway. So no rush. Tag me please. 🙂
blissful pollen
Thanks. I have a notary coming in 5 minutes, then dinner was worked out, so I'll...
My program ended? What happened?
await.gather(...) runs until all the listed tasks have finished. If gather returns that means all the tasks listed have finished.
The most common causes of a task ending are: an exception occurred causing the task to end, or the task function finished. If you want to ensure the function executes forever have a loop that never ends in the function (e.g. a while True loop)
Still may need some massaging in the wording. Or even a short over simplified example of a async function with a while True loop.
idle owl
Thank you!!
idle owl
My program ended? What happened?
`await.gather(...)` runs until all the listed t...
something like py async def some_name_here(): while True: await asyncio.sleep() ?
or is that too simple.
and not actually valid.
blissful pollen
something like ```py
async def some_name_here():
while True:
await a...
Yeah exactly. If you want to make it extra clear just put in a print statement like print("I'm looping").
Looks valid, minus sleep needs a value
blissful pollen
I think most people will get that example
idle owl
Right on. I really appreciate your help!
blissful pollen
No problem. I just went through all this myself so that's helping me explain it
idle owl
I'm happy to leverage that! 😄
blissful pollen
I never want to write a main loop with all sorts of time.monotonic() - last time > 0.1 to try to coordinate timing of several tasks again 🙂
idle owl
Hah that's fair. I might have to break out of that too.
jaunty juniper
your code here
idle owl
I can't say I grasp this more than superficially with what I explained.
But this template is getting pretty solid, I think
Might include Dan's summary from his guide.
Covers the basics in a bullet list.
tulip sleet
https://blog.adafruit.com/2022/03/31/circuitpython-7-2-4-released/ (of particular interest to plain Adafruit Feather ESP32-S2 users)
idle owl
Thanks for doing the release, @tulip sleet!
blissful pollen
I think you can really see the power if you try to write non-asyncio code to do the same as your demo, rotate a ring and blink another at different rates. The timing becomes a nightmare
idle owl
I think you can really see the power if you try to write non-asyncio code to do ...
Dan does that in his concurrent tasks page. With bare LEDs.
I'm not sure it's worth including in this template. I'll link folks to his guide for that in depth stuff.
Need to get a GIF, add a few more notes, and link to Dan's guide before this template is ready for primetime.
blissful pollen
Yup it's a good thought exercise if nothing else. At least helped me when I started writing it before going "let's see what this asyncio does"
idle owl
I think I'll get two GIFs, one for a built in button, one for external button, and upload them to GitHub, and then embed them directly from there, if that's even possible. I'll find out.
Otherwise the GIF won't match. I'm not supposed to do one for every board guide, but I feel like it should at least look like the wiring setup in the guide.
I'm grateful to only do it once (or twice).
idle owl
Yup it's a good thought exercise if nothing else. At least helped me when I star...
For sure! I have no idea how much we want to start pushing use of asyncio. We already have folks trained on time.monotonic() so it's easier to use with less explanation.
Alright I think dinner is home and in need of reheating, so I am headed out.
Thanks again for all the help!
blissful pollen
Enjoy! You are welcome
onyx hinge
@tulip sleet the (pico-specific?) rp2040 uart gpio00 bug, did you file an issue yet? I wasn't sure how to reproduce, but wanted to look at it
[about to step away for an hour, but I'll read back]
tulip sleet
<@!329766224093249548> the (pico-specific?) rp2040 uart gpio00 bug, did you file...
Thanks - I will file an issue if I don't make progress on it right away.
jaunty juniper
I'm not clear on the "never reset" thing. In status_led.c we do this for each color component:
if (PWMOUT_OK == red_result) {
common_hal_pwmio_pwmout_never_reset(&rgb_status_r);
}
this means don't reset the pin to a known state (usually GPIO input, often floating, sometimes not) when restarting the VM
just leave it alone. That is, its state is persistent across VM instantiations
jaunty juniper
it only does that for the PWM pins it seems
and I get that on the C3 if I enable the status LED (which is PWM)
Traceback (most recent call last):
File "code.py", line 17, in <module>
File "adafruit_rgbled.py", line 107, in __init__
RuntimeError: All channels in use
after a reload (it works on the first run)
tulip sleet
in this case the PWM objects should be preserved and reused so they can be used for status. It sounds like something is not noticing that and is trying to re-create them
so maybe same underlying cause as your https://github.com/adafruit/circuitpython/issues/6061
jaunty juniper
shouldn't the PWM objects be deinited like the pins are, with the current way of doing the status LED ?
tulip sleet
maybe so. I haven't looked at that bug yet
jaunty juniper
keeping the PWM object when the status LED is not used by CP on the C3 would terribly starve the user code of PWM peripherals, wouldn't it ?
tulip sleet
i agree, it should probably just shut down the status LED. Or it could steal the peripherals away when needed.
this is not the qt py, right, it is one of espressif's boards?
jaunty juniper
the AI Thinker board, with PWM status LED
tulip sleet
so how would we report status issues that are normally reported via neopixel colors, etc.? How many PWM periphs are there?
jaunty juniper
I assume like the others, setup the peripheral when we need it, deinit it when the user code runs ?
tulip sleet
and grab it back when there's an exception, etc. to indicate
jaunty juniper
we don't "never_reset" the neopixel data pin for example
tulip sleet
try it the other way, and see if it works out, is what I would do
proven garnet
@idle owl depend on adabot's capabilities, I've checked out this library and it's pretty good at editing YAML files. Would need some massaging afterward, but whatever adabot can't path for those I'm extremely confident should be doable that way.
jaunty juniper
try it the other way, and see if it works out, is what I would do
it works without it, as far as I can tell, I don't know what the downsides might be
tulip sleet
try writing a program that raises an exception
or otherwise has an error (syntax error, etc.)
jaunty juniper
ctrl-C gets the LED to blink red
jaunty juniper
I get the green one, it's not white in the REPL
orchid basinBOT
solar whale
Is the QTPY ESP32C3 build still having known issues with the REPL? I am having trouble accessing it via rshell. Via screen or tio, I get to the REPL, but seem to be dropping some characters... I was curious if you wanted issues files on this yet or if it just needs more time. It is not a problem, I was just trying it out. The ESP32C3 devkit works well with rshell (modified to us binascii) but the QTPY ESP32C3 hangs when connecting to the REPL.
jaunty juniper
The LED PWM controller can generate independent digital waveform on six channels
solar whale
note the messy formating on teh REPL ```tio 10:59:29] tio v1.32
[tio 10:59:29] Press ctrl-t q to quit
[tio 10:59:29] Connected
Adafruit CircuitPython 7.3.0-beta.0 on 2022-04-01; Adafruit QT P
Adafruit CircuitPython 7.3.0-beta.0 on 2022-04-01; Adafruit QT Py ESP32C3 with ESP32-C3FN4
help('modules')
future collections neopixel_write terminalio
main digitalio onewireio time
_asyncio displayio os touchio
_bleio errno pulseio
adafruit_bus_device espidf ulab
adafruit_bus_device.i2c_device fontio rainbowio ulab
adafruit_bus_device.spi_device framebufferio random ulab.fft
adafruit_pixelbuf gc re ulab.linalg
analogio getpass rgbmatrix ulab.numpy
array gifio rtc ulab.scipy
atexit i2cperipheral sdcardio ulab.scipy.linalg
audiocore io select ulab.scipy.optimize
audiomixer ipaddress sharpdisplay ulab.scipy.signal
json socketpool ulab.scipy.special
bitbangio keypad ssl ulab.utils
bitmaptools math storage uselect
board mdns struct vectorio
builtins microcontroller supervisor watchdog
busio micropython synthio wifi
canio msgpack sys
Plus any modules on the filesystem
hmmm I haven't noticed something like that on the AI thinker C3 (yet)
solar whale
No isses on the esp32c3 devkit wither
REPL is clearly funky. Tryin to cut/paste into it is losing characters.
onyx hinge
I can't imagine why any c3's usb/repl would behave differently than another. it's a good puzzle.
solar whale
there is some stuff about Serial/JTAG that seems to be unique to the QTPY
jaunty juniper
we have pasting issues with tio on NRF52
solar whale
it seems better in screen -- import wifi seems to kill it.... just hangs
jaunty juniper
I integrated that change to transmit more than 1 byte at a time https://github.com/tio/tio/issues/110 thanks to @onyx hinge (yes, still open) in my fork of tio
solar whale
still not right in screen either. but import wifi is clearly broken.
I also had problems with wifi on the QTPY ESP32C3 with Micropython.
jaunty juniper
are the wifi issues specific to the QTPY too ?
solar whale
yes -- works ok on the ESP2C3 devkit.
That's why I just was curious if the QTPY was not yet ready for users.
I don't want to create issues before it is appropriate.
jaunty juniper
ok, so far I tested the version without never_reset on the AI thinker C3, Pimoroni tiny2040 and Xiao BLE sense and it seems to do the color blinking as expected, or at least not worse on the NRF52 than it did in issue 6061 (in which I did test removing the never_reset too at the time, with a minor effect of making "All timers in use" happen a few pins later)
onyx hinge
https://blog.adafruit.com/2022/04/01/circuitouspython-0-1-0-ahpla-0-released/
@tulip sleet I have some concerns about the syntax. Is it too late to make improvements? maybe we can discuss it in the weeds at today's discord meeting
tulip sleet
i was also working on spirals
jaunty juniper
.si ti yaw eht tcefrep s'ti kniht I
lone axle
I'm kind of tempted to make something that will parse and execute CircuitousPython.
onyx hinge
https://en.wikipedia.org/wiki/Boustrophedon This literally was a thing in the past
tulip sleet
latest Segger J-Link release (V7.64) says it fixes:
- [DLL]: Raspberry RP2040: Reset did not work properly. Fixed.
onyx hinge
#!/usr/bin/python3
# a legal boustrophedon program # # margorp nohtpy lagel a #
(lambda f=open(__import__("sys").argv.pop(1)):exec("\n".join(line[::(-1)**i]
for i, line in enumerate(f.read().split('\n')))))()
and also a version that can go in a #!-line on linux systems (and maybe others) so that you can invoke your .by files directly:
#!/usr/bin/python3 -c(lambda f=open(__import__('sys').argv.pop(1)):exec('\n'.join(line[::(-1)**i]for i, line in enumerate(f.read().split('\n')))))()#!```Speaking of obvious improvements on April 1st, I had the misfortune of visiting Stack Overflow today and ow my eyeballs.
dusk gull
I think I have hunted down most of what I need to know and where to get started related to suggesting change to clock setup to allow more options for XOSC0 and XOSC1 and whether or not they are crystals.
I do feel like having somebody that worked more in CircuitPython give more direct feedback would help with context and making sure am headed down the right path.
If somebody that had more knowledge of the clock setup sections of the SAM chips had even 10 minutes I feel like that would help me a great deal.
I also understand everybody's time is valuable and I am happy to be invoiced for effort.
stuck elbow
I think I have hunted down most of what I need to know and where to get started ...
sounds like a good subject to bring up in the "in the weeds" section of the weekly meeting
tulip sleet
I think I have hunted down most of what I need to know and where to get started ...
I suggest you just go ahead and make a PR. We have certainly worked on that code, but it's been a year or two, and we don't have the details at the top of our brain, so we'd be doing the same research that you would.
dusk gull
sounds like a good subject to bring up in the "in the weeds" section of the week...
Just keep in mind I am not an Adafruit employee. I have worked with Micropython but has moved mostly to SAM chips now (specifically E51) and have clients wanting Circuitpython. I am new to CircuitPython but have been working with Micropython for years. I am not a super strong C developer, so I am always nervous to do pull request, however I have a pretty good idea on what I want and need.
I am also assuming that most people just use adafruit boards, but I am looking for Circuitpython to be flexible enough custom boards is easy. I have worked with Micropython and Damien to help move things that way but still see limitations on the Circuitpython side.
Short answer I am not sure I could be in that meeting, but want to help and willing to contribute even if it is just money to help more advanced designers/developers have features they need.
tulip sleet
it is not money at all, it is time and resources (people)
stuck elbow
the most precious resource: human attention
shockingly, most people here are not Adafruit employees
dusk gull
I suggest you just go ahead and make a PR. We have certainly worked on that code...
Totally understand, not that I have a lot of money, but I also have more money than time. Freelancer working on multiple development projects. I just know it can be easier to focus when it is not a favor and paid for.
I am mostly trying to figure out design intent. There is some things that happened that is limiting, so I worry that doing it well would be a much bigger project, over just getting it to work.
Definitely some of this is me being hesitant to break something and not know it.
tulip sleet
for me at least, it is a lot easier to offer suggestions about how to modify a proposed change than to do it from whole cloth
dusk gull
for me at least, it is a lot easier to offer suggestions about how to modify a p...
Very true, and I would agree. I think I am still wrapping my head around what is there and why, so I know how best to modify it. That said sometimes you just need to start tearing into it.
I guess that is really more of what I am asking, if somebody had a little time to go over what is there and how it works now, so I can make my suggested edits.
Again totally understand people are busy and this is just not something anybody here can/wants to worry about.
tulip sleet
I am not sure they are still correct, but see the top few lines in circuitpython/ports/atmel-samd/asf4_conf/samd51/hpl_gclk_config.h, and see ports/atmel-samd/peripherals/samd/clocks.c and peripherals/samd/sam_d5x_e5x/clocks.c. ALso see this PR: https://github.com/adafruit/circuitpython/pull/888
dusk gull
I am not sure they are still correct, but see the top few lines in `circuitpytho...
Thanks, the asf4_conf related stuff was an area I was not looking at. I was mostly focusing on the other locations that you mentioned.
If I remember correctly, defines in the mpconfigboard.h file should overwrite these values, right?
dusk gull
I am not sure they are still correct, but see the top few lines in `circuitpytho...
I guess maybe not, but at least I have a good place to look at at worst edit on a fork and test. I think I can chase down using the "BOARD_HAS_CRYSTAL" and a good reference on when clock settings can be changed when config origniates from mpconfigboard.h
dusk gull
I am not sure they are still correct, but see the top few lines in `circuitpytho...
I see something about a wizard for the clock configs? Are you using something to generate these, like atmel studio or atmel start?
tulip sleet
I see something about a wizard for the clock configs? Are you using something to...
we only did that to get the original generated asf4_conf files. The wizard is not that helpful; it just exposed all the possible settings, and we edited the files by hand. Atmel START is painful to use.
dusk gull
we only did that to get the original generated asf4_conf files. The wizard is no...
Thanks, and that makes sense. I am on Mac and didn't start using SAM chips until recently so I am using MPLABs and Harmony, which can be annoying but clock config is not so bad. That said I do not think it translates well to this format.
So far the only thing I really see configurable is if there is an internal vs external 32k clock
tulip sleet
There is this, but I don't know if anyone uses it now: https://docs.circuitpython.org/en/latest/shared-bindings/samd/index.html. submitted by a user:
see
https://github.com/adafruit/circuitpython/pull/785 for original PR samd module (not present by default)
dusk gull
There is this, but I don't know if anyone uses it now: <https://docs.circuitpyth...
Thanks, I saw that and was using it as one of my references.
One thing I have done in the past is use an external clock at 48MHz to make the minimum amount of changes to dividers and other settings and just focus on the source.
Thanks again, I will try to not bug you for awhile.