My original idea was to do something like https://pid.codes/
#circuitpython-dev
slender iron
I'm happy to setup the domain to the github pages for the repo if you want to add a simple page for it
tulip sleet
i can put that on my todo list
slender iron
probably not too hard to render the markdown either
tulip sleet
@slender iron
I enabled github pages, and the README is now showing up as https://creationid.github.io/creators/. You can set up creationid.org as a custom domain:
https://docs.github.com/en/pages/configuring-a-custom-domain-for-your-github-pages-site/managing-a-custom-domain-for-your-github-pages-site
slender iron
Will do now
tulip sleet
I don't see a strong reason to do anything else right now. Any updates of the blurb could be done in the README, but it's quite fine as is
slender iron
ok, creationid.org DNS updated
fringe yarrow
@danh Just to be clear, there are far more than 2 sales of that board overall, just that seller is probably new to that item. 😄 There's LVGL graphics libraries for those boards, which support most of the variants, and platformio has definitions for all known variants.. You're right, however, that having a dozen or more incompatible clones makes it unattractive to add official support for.
slender iron
@danh Just to be clear, there are far more than 2 sales of that board overall, j...
I don't mind having one build/creationid for each as long as they are clearly documented so folks can know what they have.
tulip sleet
if people are going to clone boards it would be nice if they actually made them the same 🙂
fringe yarrow
I don't mind having one build/creationid for each as long as they are clearly do...
They do clearly indicate on the back with a part number.
slender iron
the 2432S028?
tulip sleet
afk for an hour or less
fringe yarrow
yes. That model has spi ili9341 display.. some others have different combinations of display and touch interfaces.
its super low priority as they're not exaclty an Adafruit or Lilygo device.. I'll play with mine as I have time and will revisit submitting a pr if it seems worthwhile
slender iron
I did the lilygo ones just because they are the only boards I have with parallel display connections
manic glacierBOT
fringe yarrow
which one has a parallel display? I was looking into doing support for a different version of that yellow board which had a parallel display and couldn't find an example to follow
fringe yarrow
oh yeah, look at that..
The "2432S022C" version has the ST7789 in a parrallel configuration, very similar to that.
manic glacierBOT
mental sigil
Howdy, I'm the dev who wrote Hancho. I'm looking at tannewt's branch out of curiosity.
@slender iron I wonder if "from pathlib import Path" should be part of the automatically-prepended-to-.hancho-files header
mental sigil
....clone all the submodules? 😄
slender iron
....clone *all* the submodules? 😄
ya, sorry about that! our build isn't simple
slender iron
Howdy, I'm the dev who wrote Hancho. I'm looking at tannewt's branch out of curi...
Talked about Hancho on my stream today too: https://www.youtube.com/watch?v=tgwDZFTaYDw&t=3300s
mental sigil
neat!
@slender iron have you tried doing the cd(file) in your function already?
I can't think of a clean way to do it in hancho yet
slender iron
I'm kinda doing that already by capturing the cwd on import and then globbing relative to that
I was wondering if I'm working backwards still but need to pass a board id into a top level hancho instead
it is weird though since the board dictates the mcu and therefore the compiler settings
using absolute paths everywhere is a bit annoying because the commands end up really long
right now we build from the port's directory
mental sigil
I'll poke at it again tomorrow
manic glacierBOT
short tendon
Would it be a correct thought that since the M5Stack ATOM Matrix ESP32 Development Kit has a CP port that adding the M5StickC Plus IoT Development Kit should be pretty easy, since it's the same chip?
manic glacierBOT
slender iron
Would it be a correct thought that since the `M5Stack ATOM Matrix
ESP32 Developm...
Yup, adding boards is easy
manic glacierBOT
austere acorn
Ohai 👋
I started work on porting CircuitPython for this board: https://electro-smith.com/products/daisy-seed
It's a STM32H750 mcu, wih 1MB RAM, but only ... 128KB of flash, so obviously I ran into some size issues with the build.
The board does have 8MB external QSPI flash, is it possible to load CircuitPython directly on that?
QSPI doesn't seem to be implemented for stm though, ports/stm/supervisor/qspi_flash.c only contains commented out code, do I need to figure out how to implement that first?
Any advice would be helpful, thank you! 
spare jacinth
that would be a duty of your bootloader writing (flashing) code there, not really something to implement on circuitpy (commented code would be to write/read from it at runtime, not too useful)
tinyuf2 has code for H7 but didnt work too good in my experience and i have the exact same board for which code is written
tulip sleet
Ohai 👋
I started work on porting CircuitPython for this board: https://elect...
You'll have to figure out how to do XIP with this chip. It is possible, but we haven't done this for STM at all. Is this an exercise for your own interest? Consider whether it's worth it.
austere acorn
You'll have to figure out how to do XIP with this chip. It is possible, but we h...
I already own this board, so I thought it would be an interesting challenge to add CP support, but it's proving to be an even bigger challenge than expected 😅
Still... the temptation of using the giant 65MB memory makes it worth it, for sure!
midnight ember
I already own this board, so I thought it would be an interesting challenge to a...
We don't do things because they are easy but because we thought they would be easy. 🧙
Having a microcontroller with 65MB of RAM would make me feel like Doc Brown. 21.1 Gigawatts!
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
tulip sleet
@lone axle do you have a script to find the libraries that need a release? I would imagine so. I would like to run it to check before I update frozen modules for 9.0.0. Thanks.
lone axle
<@382939733107408897> do you have a script to find the libraries that need a rel...
Yep, Adabot has that functionality for release sweep: https://github.com/adafruit/adabot?tab=readme-ov-file#making-releases-for-circuitpython-libraries it will compare latest commit with latest release on each library and prompt to release ones where they are out of sync.
theoretically the same list should end up here: https://circuitpython.org/contributing/library-infrastructure-issues but I've found it can be off by a day or two sometimes whereas the adabot report is run live. Also on this page they are seperated into categories of how far out of sync the release date from commit date is.
short tendon
@tulip sleet and @slender iron, question on the most recent merge into requests. There are other open issues on this, and IMO fast_close does not feel like the right approach. Would it not be better to fix it more correctly, then add a param?
I have been working on an approch with https://github.com/adafruit/Adafruit_CircuitPython_Requests/issues/153, just haven't finished
tulip sleet
I have been working on an approch with https://github.com/adafruit/Adafruit_Circ...
I did not see the alternative PR. I agree if it can be fixed without a param, that would be great. Can you ping the other author to discuss in your PR?
short tendon
I will, I probably won't have mine open until mid-week . My only concern is people using a param, that then gets dropped.
slender iron
I will, I probably won't have mine open until mid-week . My only concern is peop...
Drop a note on the PR too so that the PR author can participate in the alternative solution.
tulip sleet
<@&356864093652516868> The weekly meeting will be held in just under two hours. Daylight savings time has already started in the US, so this may be a different time than you are used to if you are not in the US. Here's the notes document if you have not already added your notes: https://docs.google.com/document/d/1wyHuyLhHVop15ZEMZg2WSXNp-kdvO98wNbPvHk5b6WA/edit?usp=sharing. See you soon!
lone axle
Is this fix going to get applied in 8.2.x branch? https://github.com/adafruit/circuitpython/issues/8986 I'm running into this exact same issue (web workflow reports that device is readonly even though I have storage.remount("/", False) in boot.py and confirmed that the REPL is able to write files). I was trying to work on https://github.com/adafruit/circup/issues/206 but I'm seeing this same issue on 8.2.10 and 8.1.0 tested on Feather S3 TFT and Feather S2 TFT.
If we don't get a fix for that in 8.2.x I'm not sure there will be much benefit in adding support in circup for < 9.0.x where this is already fixed. The change to the web workflow API had broken it, theoretically it could be updated to support the old API, but it appears this other issue results in it still being unusable anyway.
Presumably if I go back far enough I can find one that works because I know I used this successfully in the past. But that would leave gap in support between whatever versions end up being where it changed.
onyx hinge
<@&356864093652516868> Our weekly meeting is about 95 minutes from now -- the US (& Canada) switched to DST, so if you're in a different area this might mean the time of day is shifted by 1 hour for you. If you have notes, please take a moment to update the google doc: https://docs.google.com/document/d/1wyHuyLhHVop15ZEMZg2WSXNp-kdvO98wNbPvHk5b6WA/edit?usp=sharing
lone axle
@tulip sleet ive merged https://github.com/adafruit/Adafruit_CircuitPython_Typing/pull/40 now. I saw your message about the plan for release. Do you want me to make the new release 1.10.3 or would you rather handle everything you mentioned all at once?
tulip sleet
Is this fix going to get applied in 8.2.x branch? https://github.com/adafruit/ci...
I think it's not worth the time to add circup support for 8.x.x for web workflow, given the imminence of 9.0.0.
tulip sleet
<@329766224093249548> ive merged https://github.com/adafruit/Adafruit_CircuitPyt...
Go ahead and make a release if there are no other useful pending PR's
short tendon
I noticed playing with circup, that it doesn't take in account for frozen libraries. Would that be a good add?
tulip sleet
I noticed playing with `circup`, that it doesn't take in account for frozen libr...
in what sense? There are two reasons for freezing libraries: save RAM, and convenience. To override a frozen library, the updated library needs to be in /.
I think the use cases are complicated enough that it's probably not worth adding a feature right now.
short tendon
I just noticed when I did a install from my code.py file that it downloaded it. I guess you are right and it would be in lib, but not used. Although if you later did an upgrade, it would show that it upgraded it and in reality it wouldn't be used.
Just a thought
tulip sleet
I'm not sure it can detect which libraries are frozen and their versions, since they are not readable from the filesystem
yes, the pre-reqs often cause duplicate libraries to be downloaded but not used. That is also true of things like adafruit_bus_device.
what is frozen is kind of a "secret"
short tendon
If I were to do it, I would probably have circup pull down the json file that's in the docs
And see if the versions could be added to that
tulip sleet
it would be nice to reveal that, but it might require considerable code that is not going to fit on the boards where frozen libraries are important for space reasons
tulip sleet
If I were to do it, I would probably have circup pull down the json file that's ...
ah, I totally forgot about that. Though the version can vary per build, so it's a lot of data to keep
short tendon
That's true... Well If you ever want to go down this rabbit hole - let me know. Happy to tackle it
tulip sleet
it has not been that much of a support issue, so I think it is lower priority
lone axle
Are there any tricks or known issues with updating submodules on main atm?
I'm getting this error:
fatal: transport 'file' not allowed
fatal: Fetched in submodule path 'lib/tlsf', but it did not contain 81d37795451578255c8d047404c4a179258dcec7. Direct fetching of that commit failed.
fatal:
::endgroup::
Duration: 1.7097632029999659
Traceback (most recent call last):
File "/home/timc/repos/circuitpython/circuitpython_copy/circuitpython/tools/ci_fetch_deps.py", line 237, in <module>
main(target)
File "/home/timc/repos/circuitpython/circuitpython_copy/circuitpython/tools/ci_fetch_deps.py", line 219, in main
fetch(submodules)
File "/home/timc/repos/circuitpython/circuitpython_copy/circuitpython/tools/ci_fetch_deps.py", line 133, in fetch
run(
File "/home/timc/repos/circuitpython/circuitpython_copy/circuitpython/tools/ci_fetch_deps.py", line 109, in run
subprocess.run(shlex.split(command), stderr=subprocess.STDOUT, check=True, cwd=cwd)
File "/usr/lib/python3.10/subprocess.py", line 526, in run
raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['git', 'submodule', 'update', '--init', '--depth', '1', '.', 'frozen']' returned non-zero exit status 128.
make: *** [Makefile:341: fetch-all-submodules] Error 1
I'd suggest doing remove-all and then fetch-all or fetch-port. Also are you using a newish git?
lone axle
❯ git --version
git version 2.34.1
the repo for tlsf was changed
lone axle
removed and re-fetching now.
tulip sleet
git >=2.36 uses a superior way of fetching submodules
i would suggest adding the git ppa if you are on ubuntu
lone axle
After remove-all and then fetch-all again it shows the same error but on a different submodule:
fatal: transport 'file' not allowed
fatal: Fetched in submodule path 'extmod/ulab', but it did not contain eacb0c9af47f85f5d4864b721c3b28661364e8e3. Direct fetching of that commit failed.
fatal:
::endgroup::
Duration: 146.4553917379999
Honestly I'm a bit superstitious around submodules. I'm going to try a fresh clone of the repo locally out of curiousity with my current version of git but then I'll add that ppa upgrade. I am on ubuntu.
lone axle
The freshly cloned copy was able to fetch-all and build successfully.
mortal kernel
After remove-all and then fetch-all again it shows the same error but on a diffe...
I've seen this sporadically for the last couple of months. I suspect there's an intermittent problem with github under load. When it happens, the download speed on the affected submodules drops to about 50K/sec before the failure.
When it does occur, I've found that a make remove-all-submodules followed by another attempt to fetch-all or fetch-port as Dan suggested does the trick.
tulip sleet
<@&356864093652516868> meeting starting in one minute
onyx hinge
turbid radish
☺️
midnight ember
devout jolt
What causes so many PyPi library downloads? Mostly for people doing Blinka or for stubs-type stuff for IDEs?
lone axle
What causes so many PyPi library downloads? Mostly for people doing Blinka or f...
Some of both I believe. anytime anyone uses pip install on a library. I don't think we get visibility to different sources though.
gilded cradle
I'm not even sure where the data comes from @devout jolt. I assume any time a blinka install or upgrade occurs, (even via being a dependency) it counts.
tulip sleet
What causes so many PyPi library downloads? Mostly for people doing Blinka or f...
I think it would be worth investigating this. I wondered if it was cumulative in some way, and not just weekly. The numbers do seem very high. Maybe many are just touches.
onyx hinge
344 623 4533
```slightly different number than I said but a *LOT* of people no matter how you sort it
onyx hinge
does someone have a link to the cardputer repl thing?
wraith crow
midnight ember
floppsy 🐰
onyx hinge
and now I have to run, thank you all and have a great week!
if 'compile_command' in dir():
# 9.0.0 alpha 7 or later supports multiple line statements
__line = input(",,, " if __cmd else "=>> " )
```nice, they found the "compile_command" code I added 🙂There are so many breaking changes from 8.x to 9.x go for it. Break all the things, might as well introduce it with 9.x
OAuth changes every year or so doesn't suprise me there are so many different iterations of it.
Every website seems to have their own quirks with 0Auth implementations.
A cheap way to emulate posterize is just drop the bit depth to 2 or 3. 😛
Thank you for hosting Dan. Wonderful updates this week. Thanks everyone!
devout jolt
Thanks @danh!
gilded cradle
Thanks
lone axle
Thanks for hosting Dan. Have a great week everyone!
tulip sleet
you're welcome!
midnight ember
200+ unique contributors is impressive. Do we have stats to compare to 6-7 and 7-8?
tulip sleet
Here is the notes document for next Monday’s CircuitPython Weekly Meeting. It is at the normal time of 11am Pacific / 2pm US Eastern here on Discord. Add your hug reports and status updates to the document before the meeting. If you are unable to attend but would still like to contribute, feel free to add your notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/1A5vXAYyCHsXmquGVkXvxFn0wosleHI8NHVKkD6Zr4S4/edit?usp=sharing
tulip sleet
@slender iron would it be ok if make the default branch of https://github.com/adafruit/tlsf be circuitpython instead of idf? Every time I go to look at it, I'm briefly confused about where your changes are.
slender iron
totally!
vital matrix
[low priority] just in case someone here has been playing with renode - could someone point me at a working configuration for any board, that runs CP? I was even unable to launch the micropython config that theoretically is available via this: https://renodepedia.renode.io/boards/rpi_pico/?view=software&demo=micropython and before diving into it, wanted to see if more experienced folks have tips.
slender iron
[low priority] just in case someone here has been playing with renode - could so...
I don't know of anyone who has
manic glacierBOT
midnight ember
Figured out after writing the ATECC eeprom it gets locked permanently. There is a warning in big red letters about this on the SparkFun site for this same crypto chip breakout. https://www.sparkfun.com/products/18077 No such bold warning on the Adafruit product page or learn guide or examples or code comment in the library.
I have figured out after being locked it goes into a permanently idle state and the only way to communicate with it is to send it a wake command. This command is malfunctioning in the adafruit_atecc library.
After being locked I can send it a wake command manually and it responds. The I2C device address becomes active and can print out the device serial number. That's as far as I've gotten so far.
The way it responds to wake frequency command prior to locking eeprom and after are different. There's something wrong with the wake function in the library after eeprom is locked.
The way it behaves suggests some type of watchdog timer. The wake command only wakes it up for about 10-20 seconds, then it goes back to idle. If continually fed print commands it stays awake indefinitely.
empty salmon
options for hardware revision checking in board.c?
We may end up with two versions of our hardware board. We would like to avoid forking our CircuitPython build for a small change.
Are there any popular / existing mechanisms which can be implemented in board.c? We do have the ability to make small changes to the second revision if there is something simple we can add and then detect.
Addendum: this is an RP2040 design
stuck elbow
**options for hardware revision checking in `board.c`?**
We may end up with two ...
the uf2 bootloader on the samd chips lets you store a bunch of info in the bootloader itself, and then you can use it in your program. I use that to read the screen offsets for the pewpew m4 console, because I have different revisions with different screens
empty salmon
Thanks. I should have added "RP2040".
stuck elbow
you could also have a small i2c eeprom memory on your board, like the raspberry pi hats do
or just encode some data with spare gpio pins
if you have a spare adc pin, you could encode the version with a resistor
empty salmon
I will look for an example of another board which accesses I2C. It would need to be accessible to both board.c and the end user.
empty salmon
I assume ADC would use a voltage divider?
stuck elbow
I guess so, but it could be switched on/off by another pin
empty salmon
I will look at some of the other boards with I2C and an EEPROM.
stuck elbow
I'm not sure there are any cp boards with eeprom
but there are some boards with i2c screens
empty salmon
Good point. Thanks for the starting points. I will do some digging.
stuck elbow
I'm thinking about adding an i2c eeprom to my robots to store the calibration data, so that I can swap the microcontroller board with the brain between the robots
willow totem
I'm thinking about adding an i2c eeprom to my robots to store the calibration da...
Good idea. Are there any kind of standards for this kind of thing, a bit like the pi-hat eeproms? I want to ensure I dont reinvent the wheel, but similarly wanted to store calibration data etc, but then it occurs to me I should have a standard mechanism for retrieving manufacturer / model / revision (of the "device" not eeprom chip)
stuck elbow
Good idea. Are there any kind of standards for this kind of thing, a bit like th...
the pi hats use an encoded linux kernel device tree overlay...
willow totem
I know linux kernel overlay device trees, but whats the encoded bit?
stuck elbow
well, you have to convert it into a single string
maybe I should have said serialized
willow totem
Aha I get you
stuck elbow
you could look at the uf2 format
willow totem
Yeah that seems like a win (for some device config stuff), hadn't seen it before some makecode thing pointed me to it recently
stuck elbow
it's still just a bunch of values packed in a struct
slender iron
eeprom has the challenge that it needs to be programmed
stuck elbow
https://github.com/adafruit/circuitpython/blob/main/ports/atmel-samd/boards/pewpew_m4/board.c#L38-L67this is how I access it in board.c
slender iron
if the pcb is different, you could use a shift register to shift in a number defined by the pcb
stuck elbow
that's three pins
you could also use those three pins in a similar fashion to how some i2c chips let you change the address, by connecting them to gnd, vcc, other pins or each other
empty salmon
I may end up using a voltage divider because it offers a "1 pin" detection mechanism that can be completely transparent to the end user.
stuck elbow
I may end up using a voltage divider because it offers a "1 pin" detection mecha...
if you are using a number of gpio pins that are only connected internally, you can have your detection pin connected to one of them depending on the version, then in board.c bring it low, set all those pins to pull-up and see which one is 0, then release them all and initialize normally, with the detection pin set to input, so it doesn't interfere with all the rest
if you can spare more such pins, you can encode more states, (internal pins)**(detection pins)
it doesn't work if the pins are connected to buttons or something else that might hold them low, because then it's unreliable
but things like a parallel display bus are perfect for this
wow, there are i2c eeproms in sot-23-5 packages
for 3 cents a piece in singles
unkempt dust
eeprom has the challenge that it needs to be programmed
It's pretty simple really. I made one class to store byte/bytearray objects. Then I used the struct class to turn regular variables into a byte object to store them. And retrieve them as a tuple
would have been easier in C/C++ even 😩
stuck elbow
the problem is that it is an extra step in your production process
empty salmon
@stuck elbow - using a voltage divider can be done with just 1 pin and varying the resistors allows for multiple possible values.
stuck elbow
<@351136009922019331> - using a voltage divider can be done with just 1 pin and ...
but then you have constant power drain
empty salmon
One issue I can see with an I2C EEPROM is the user could overwrite it (unless the board has multiple I2C pins combinations)
True, but using high Ohm resistors can mitigate the drain considerably.
tulip sleet
what are the differences between the board revs?
stuck elbow
One issue I can see with an I2C EEPROM is the user could overwrite it (unless th...
there is a write protection pin
empty salmon
@tulip sleet - We are trying to figure that out right now but it looks like it will be a slight difference in the display - requiring a small change to the init sequence.
vital matrix
I don't know of anyone who has
ack - thx!
lone sandalBOT
vagrant wind
In a pull request I have added a class for the keypad module. I'm trying to disable it for the atmel samd board because the flash is limited. I've tried to disable it in ports/atmel-samd/mpconfigport.mk but without success. Can anyone help me on this?
https://github.com/adafruit/circuitpython/pull/9035
empty salmon
<@329766224093249548> - We are trying to figure that out right now but it looks ...
After a discussion with a few hardware designers, we plan to take an unused analog pins and add a voltage divider. This will allow us to switch the resistor values to "hard code" a voltage value on the pin, thereby distinguishing the board revision.
This will give us the ability to distinguish a few different revisions (if that should be needed in the future).
midnight ember
Interesting idea.
empty salmon
It feels a little bit like a kludge but at the same time, it has a big upside for the end user as we can make small changes to the hardware and preserve a simple experience on circuitpython.org for firmware. The end user need not be bothered with backend hardware revisions.
solar whale
Can anyone clarify this for me -- According to the README in the repo adafruit-circuitpython-asyncio is not inteded to be released to PyPi yet, here it is https://pypi.org/project/adafruit-circuitpython-asyncio/ If I am using CircuitPython code on a Raspberry Pi (via blinka) and I want to use asyncio, I had assumed I should just use the CPython asyncio. Is that correct? If I run pip install adafruit-circuitpython-asyncio it happily installs but that does not appear to be the intended use and it is not clear to me what I get if I then import asyncio....
from the README Installing from PyPI This library is meant to be a subset of the ` asyncio module in CPython <https://docs.python.org/3/library/asyncio.html>`_, and will not be made available on PyPI. Use the CPython version instead.
lone axle
Can anyone clarify this for me -- According to the README in the repo adafruit-c...
I believe the information in the readme (that it's not intended to be released to PyPi) is out of date. At one point me made the decision to release all CircuitPython libraries on PyPi even if they were not intended for use with CPython, because it can still serve a useful purpose with more fully featured IDEs that will use the locally installed instance to populate context hints and documentation directly in the editor. We probably never went back and updated any readme's that state otherwise.
With regards to using asyncio with Blinka and CPython, I don't know but my instincts match yours that it would be best to use the "real" CPython asyncio in that environment not the CircuitPython one. If you install both the CPython and CircuitPython ones, I am also unsure what happens when you use import asyncio I assume python has some precedence rules for name collisions but I don't know what they are.
solar whale
With regards to using asyncio with Blinka and CPython, I don't know but my insti...
Thank you! That is just what I needed to know.
short tendon
@tulip sleet or @onyx hinge do you have any knowledge of with adafruit_requests reads the rest of the response on .close(). I've played with it on a bunch of boards and have yet to have it cause any problems to not just close...
onyx hinge
@short tendon I think it has to do with avoiding the cost of re-establishing a https connection. If this cost is tolerable on relevant boards these days, it seems like it'd be better to just close and re-connect (at least if the response hasn't been consumed already)
short tendon
<@1152454697614069811> I _think_ it has to do with avoiding the cost of re-estab...
So, I tested that as well. On .close() the socket is flagged as free in ConnectionManager, but not actually closed. If you request the same host, it returns the same one and it's still connected. It had no issues grabbing the correct data.
tulip sleet
<@329766224093249548> or <@426200547024961539> do you have any knowledge of with...
I wonder if it is a leftover from ESP32SPI. adafruit_requests was originally written for ESP32SPI, way before we had native wifi ESP32 boards.
short tendon
I wonder if it is a leftover from ESP32SPI. adafruit_requests was originally wri...
I saw a comment from jepler on an issue about that. I've tested pretty hard on the ESP32SPI and can't find any issues
onyx hinge
I don't know who originally wrote the connection re-use code or whether they're still working on CircuitPython, I'd have to check out the git logs.
tulip sleet
I saw a comment from jepler on an issue about that. I've tested pretty hard on t...
if you look at the "blame" associated with that code, you could go back in time and see whether the read was added at some point
short tendon
Version 1.70 from Scott. First commit Switch to HTTP1.1 so that we can reuse sockets. https://github.com/adafruit/Adafruit_CircuitPython_Requests/commit/87d063329c97c00fb0988e2b255eaa827cccdd50 started the change (almost 4 years ago)
tulip sleet
so, in particular, the draining of the connection on close() was added, right? So we could ask @slender iron what he had in mind when he added that code
short tendon
Correct. It's also when request.Session was added
slender iron
right. I added the draining code
because you can't reuse a socket if a bunch of data is remaining
slender iron
In a pull request I have added a class for the keypad module. I'm trying to disa...
I'd suggest making a new module instead
tulip sleet
There are examples of disabling functionality in a module, e.g. CIRCUITPY_BUSIO_SPI. It raises NotImplementedEerror if turned off.
vagrant wind
I'd suggest making a new module instead
You were suggesting to make a complete new module and not adding DemuxKeyMatrix in the keypad module?
slender iron
I prefer being able to enable/disable on a module boundary
vagrant wind
Any suggestion for the module name?
tulip sleet
but a separate module might be easier. You need to check that CIRCUITPY_KEYPAD is on if CIRCUITPY_DEMUX_KEYMATRIX is on. Maybe call it keypad_demux_keymatrix. So keypad_demux_keymatrix.DemuxKeyMatrix. Since it's so releated to original keypad
slender iron
looks up what a demux is
tulip sleet
3-to-8, in this case
slender iron
ah, three binary bits select one of 8 inputs
tulip sleet
i wonder if it is a demux or a mux. I'm going to look at the cardputer schematic
uses this: https://www.ti.com/product/CD74HC138 ; a demultiplexer
tulip sleet
Any suggestion for the module name?
or maybe call it a MultiplexedKeyMatrix
short tendon
because you can't reuse a socket if a bunch of data is remaining
Do you have an example of where this would fail?
tulip sleet
maybe that's not such a great name, since the matrix is already "multiplexed"
slender iron
Do you have an example of where this would fail?
I'd have to look where I added the consume calls. It was done for cases where we were done with a request before all data was received
I don't have the context for it anymore so feel free to change it
short tendon
So how about we comment out the code, and just flag it as closed, and if it causes problems then use that to fix it?
I don't love it, but can't find a way to break it currently
crimson ferry
https://requests.readthedocs.io/en/latest/user/advanced/#keep-alive is there any performance data of draining vs. re-establishing TLS?
and can we set a request to [fully] close (vs. drain and re-use)?
vagrant wind
@tulip sleet Regardless the name, I can change it as you suggested, I have already implemented CIRCUITPY_KEYPAD_DEMUXKEYMATRIX to disable the module. Because the build of atmel-samd fails on flash size I added CIRCUITPY_KEYPAD_DEMUXKEYMATRIX = 0 but it keep fails. Can you help me to fix that?
short tendon
and can we set a request to [fully] close (vs. drain and re-use)?
Ahhh, so it might be when changing hosts, let me try that
crimson ferry
someone mentioned the other day about streaming responses that never end (can't be drained)
(we can set stream=False)
short tendon
So they could never close, becuse we read to the end.
I'm trying to get not reading to the end to cause an issue
crimson ferry
if I read right, using a session preserves thes socket for reuse, otherwise socket is actually closed (but I wonder if the latter case needs draining?)
slender iron
So how about we comment out the code, and just flag it as closed, and if it caus...
Just delete it. git has the code if we need it again
short tendon
So everything is now using a session. I can't find a case that not draining causes issues
tulip sleet
<@329766224093249548> Regardless the name, I can change it as you suggested, I ...
Are you seeing a clear size difference when CIRCUITPY_KEYPAD_DEMUXKEYMATRIX = 0? Make sure you do a make clean before recompiling. Also you could make including shared-module/keypad/DemuxKeyMatrix.c in the list of sources to compile be conditional.
short tendon
Just delete it. git has the code if we need it again
True
tulip sleet
I was looking to see if there was a lot of shared code between KeyMatrix.c and DemuxKeyMatrix.c, and there is, but maybe it's too complicated to merge the two and do run-time checks.
short tendon
It also fixes HEAD and allows for OPTIONS
tulip sleet
@vagrant wind but as a separate module it would be more straightfoward. It is a bit too bad there is a lot of duplicated code in shared-bindings, but that could be refactored at some point.
vagrant wind
<@658337305454444574> but as a separate module it would be more straightfoward. ...
What I've implemented here https://github.com/adafruit/circuitpython/pull/9035 is not already a new module?
tulip sleet
What I've implemented here https://github.com/adafruit/circuitpython/pull/9035 i...
It was, but now it's inside keypad. tannewt's comment was asking you to put it in shared-module and shared-bindings. I looked at your original commits and it was in espressif/boards. So if you go back to that commit, and rename the whole module something more generic than cardputer, and move the files, that would be a generic separately-enableable module
I didn't have chance to look at that commit before it was changed until just now
i.e. this commit was a separate module: https://github.com/adafruit/circuitpython/commit/9f9dcd013c854418efae9f1f20e35f4c2c089400
mortal kernel
@danh Noticed you using a leading '_' on the static function name _adapter_set_name in #9041. Is this a CP naming convention? Are there others I should be aware of?
tulip sleet
@danh Noticed you using a leading '_' on the static function name `_adapter_set_...
sometimes it's a convention 🙂 . It's not necessary, and we're not that consistent.
tulip sleet
So everything is now using a session. I can't find a case that not draining caus...
@short tendon @crimson ferry is someone making a PR? The conclusion is to remove the draining code, and you've now tested it? We can make a release and freeze this into the next release candidate
short tendon
<@1152454697614069811> <@360465312518504448> is someone making a PR? The conclus...
Yes, I'm working on it and should have it open today
tulip sleet
thanks!
manic glacierBOT
fringe yarrow
I'm going through the academic exercise of trying to get circuitpython to build and run on one of those as yet unsupported Sunton ESP32 OLED+Touch+SDCard devices. I know it works because I can upload basic arduino blink sketches via vscode and it's blinking away and printing serial output on the usb uart. Its a ESP-WROOM-32 at heart and every avenue I've tried so far has left me in a TG1WDT_SYS_RESET boot loop
plucky tulip
Not sure where to post such things but MEMENTO board is missing in adafruit.com "CircuitPython" products category
fringe yarrow
Not sure where to post such things but MEMENTO board is missing in adafruit.com ...
This one?
plucky tulip
This one?
Yes
If I was not precise, I mean it is not listed in a product list https://www.adafruit.com/category/956
fringe yarrow
its under sensors/cameras
plucky tulip
But it is also a CircuitPython board, isn't it? There are multiple products that are listed on multiple categories
fringe yarrow
you're probably right.. I have nothing to do with it, was just making sure your comment was clear..
wraith crow
I'm going through the academic exercise of trying to get circuitpython to build ...
I've had success using the "Adafruit HUZZAH32 Breakout with ESP32" firmware, although it sometimes takes a few power cycles/resets to start up properly
wraith crow
Also note, that you need to grab either CP 8.x or the absolute newest. The boot loop issue was resolved yesterday by https://github.com/adafruit/circuitpython/pull/9039
@fringe yarrow 👆
fringe yarrow
Further to my efforts to get circuit python to build and run on this Sunton "Cheap Yellow Display". If I flash it with a prebuilt 8.2.10 intended for the esp-eye device, it boots and runs hello world as you'd expect. If I flash it with the 9.0 prebuilt, it bootloops (TG1WDT_SYS_RESET). If I build from latest main for the esp-eye and flash it, its the same result, TG1WDT_SYS_RESET bootloop.. Curious..
tulip sleet
But it is also a CircuitPython board, isn't it? There are multiple products that...
I will bring this up internally.
tulip sleet
@short tendon check your DM's
solar whale
@tulip sleet Is this setting for the matrixportal s3 an error https://github.com/adafruit/circuitpython/blob/main/ports/espressif/boards/adafruit_matrixportal_s3/mpconfigboard.mk#L16 Should it be disabling CIRCUITPY_ESPCAMERA instead of CIRCUITPY_ESP32_CAMERA?
tulip sleet
<@329766224093249548> Is this setting for the matrixportal s3 an error <https://...
yes, that is obsolete, thanks for catching that. a quick PR would be great, or I'll do it.
solar whale
I have to run out, but I can do it this afternoon.
tulip sleet
i can just do it now, but I'll credit you
tulip sleet
But it is also a CircuitPython board, isn't it? There are multiple products that...
now added to category
wraith crow
Further to my efforts to get circuit python to build and run on this Sunton "Che...
It looks like you saw the fix for the TG1WDT_SYS_RESET bootloop, but trying to use the 9.x esp-eye firmware is probably still going to cause a problem because starting in 9.x that device specifies a PSRAM size of 8M and I suspect your device only has 520K (I got mine from a different seller but it has the same ID on the back). The "Adafruit Huzzah32 Breakout" firmware is working on both 8.x and the "Absolute Newest" 9.x build (Adafruit doesn't have a web installer for the absolute newest so you have to use esptool or I suspect the esp webflasher would work).
fringe yarrow
It looks like you saw the fix for the TG1WDT_SYS_RESET bootloop, but trying to u...
That's likely the issue I'm having, yes, the psram mismatch.. Thanks for the tip.
tulip sleet
@lone axle I was using your instructions above and in the adabot repo to do a release sweep to see if anything in circuitpython frozen/ needs updating, but adabot installs locally as UNKNOWN instead of as adabot. Do you understand what is wrong here:
halbert@cod:~/repos/adafruit/adabot$ pip3 install .
Defaulting to user installation because normal site-packages is not writeable
Processing /home/halbert/repos/adafruit/adabot
Installing build dependencies ... done
Getting requirements to build wheel ... done
Preparing metadata (pyproject.toml) ... done
Building wheels for collected packages: UNKNOWN
Building wheel for UNKNOWN (pyproject.toml) ... done
Created wheel for UNKNOWN: filename=UNKNOWN-0.0.0-py3-none-any.whl size=1815 sha256=52cd05fa477ed99dca782887492296719671e88e5426410415a79f5b4fbd7a8d
Stored in directory: /tmp/pip-ephem-wheel-cache-6877435f/wheels/a7/8a/43/75aa8bf31ec17cf2d0324d08ecf1126c009bdefb067aa2960d
Successfully built UNKNOWN
Installing collected packages: UNKNOWN
Successfully installed UNKNOWN-0.0.0
the repo clone is up to date
python is 3.10.12
lone axle
<@382939733107408897> I was using your instructions above and in the adabot repo...
Hmm, I don't really. when I run the install with pip install . those UNKOWNs end up being adafruit-adabot
Successfully built adafruit-adabot
Installing collected packages: GitPython, adafruit-adabot
Attempting uninstall: GitPython
Found existing installation: GitPython 3.1.35
Uninstalling GitPython-3.1.35:
Successfully uninstalled GitPython-3.1.35
Attempting uninstall: adafruit-adabot
Found existing installation: adafruit-adabot 0.0.0+auto.0
Uninstalling adafruit-adabot-0.0.0+auto.0:
Successfully uninstalled adafruit-adabot-0.0.0+auto.0
Successfully installed GitPython-3.1.41 adafruit-adabot-0.0.0+auto.0
Building wheels for collected packages: adafruit-adabot
Building wheel for adafruit-adabot (pyproject.toml) ... done
Created wheel for adafruit-adabot: filename=adafruit_adabot-0.0.0+auto.0-py3-none-any.whl size=57548 sha256=18e95f327e61ea84c42ef7999dd568e0857720272b88c3baef180e18f2f3ccc4
what is your python version, and are you in an updated adafruit/adabot clone?
maybe it's some setuptools thing
lone axle
Python 3.10.12 but my pip is a bit out of date: 23.2.1
tulip sleet
mine is older 22.0.2
tulip sleet
i am at 69.1.1
lone axle
@tulip sleet are you in a virtualenvironment or directly in main python install?
tulip sleet
main python install
i don't have trouble with any other pip3 installs
ubuntu 22.04
lone axle
This comment chain is discussing the same symtom https://github.com/pypa/setuptools/issues/3269
tulip sleet
maybe I will do a websearch for "UNKNOWN" ... aha
i will take a look at that, thanks, in a meeting right now
lone axle
I'm able to reproduce the UNKOWN thing if I install it in the base python installation. It does work successfully in a venv with a fresh clone for me though. I'm not sure what could cause that difference. Presumably there is something we could do inside pyproject.toml or one of the other config files to have some effect on it, but I'm not sure what.
tulip sleet
I'm able to reproduce the UNKOWN thing if I install it in the base python instal...
I installed an upgraded pip3 in .local/.... That works but it started complaining about other existing installations. I'll undo that if necessary. Awaiting Ubuntu 24.04 soon anyway
manic glacierBOT
manic glacierBOT
manic glacierBOT
fringe yarrow
I got a good one for someone with more knowledge of the undergubbins of the circuitpython boot procedure. I have this board that looks just like an 01space ESP32-C3-0.42LCD device. I flash it with 8.2 or 9 from the web installer or from my own build, and after the post-flash reset, it boots and runs hello world and displays the IP on the tiny LCD. If I power cycle it, it never boots cleanly again. It reports CircuitPython core code crashed hard. Whoops! Heap allocation when VM not running. and gives me safe mode, though it still displays the IP info on the LCD. I've flashed it with the QT-Py ESP32-C3 image, and it boots fine. I'd like to have the proper image with the fully accurate pin and lcd definitions on it.. Any ideas? It's possibly a clone, but if so, it seems fully pin and peripheral compatible. I'd blame bad flash or something like that, but the fact that the QT Py ESP32-C3 image flashes and runs indicates the hardware is ok.
fringe yarrow
I got a good one for someone with more knowledge of the undergubbins of the circ...
on further investigation, it seems like power cycling it is actually fine, its when i use the onboard reset button, after that it crashes and won't recover from a soft-reboot
tulip sleet
I got a good one for someone with more knowledge of the undergubbins of the circ...
I think this may have something to do with the on-board display not resetting properly on a hard pushbutton reset. In board.c for the 01space board, there is no reset pin specified for the display in that code (NULL argument in common_hal_i2cdisplaybus_i2cdisplaybus_construct) The code may be having trouble talking I2C to the display because there is no way to reset the display to get it back to a known state without power-cycling.
I looked at the 01space schematic, and there is no pin shown for the OLED display reset. Maybe one could be wired up with a spare pin -- I don't know.
I would say this is poor hw design.
the arduino examples in https://github.com/01Space/ESP32-C3-0.42LCD/ also indicates no reset pin is connected
getting an I2C OLED reset in the proper sequence is tricky; we have extra circuitry on various Adafruit boards to do this right
fringe yarrow
I looked at the 01space schematic, and there is no pin shown for the OLED displa...
Sounds reasonable.. Is there something special in how CircuitPython does this vs a basic arduino sketch? I can upload a basic LCD hello world sort of sketch from vscode and it will run and reset cleanly when i use the button..
tulip sleet
Sounds reasonable.. Is there something special in how CircuitPython does this v...
maybe just adding a delay somewhere. CircuitPython sets up the I2C bus and display really early. Though I would think that would be true of the arduino sketch too. COuld you point to the Arduino sketch?
you could init the display in Python code instead of relying on the builtin one. Also I would want to see where the non-VM heap allocation is happening, but that requires some debugging and I don't have the hardware
fringe yarrow
maybe just adding a delay somewhere. CircuitPython sets up the I2C bus and displ...
#include <U8x8lib.h>
U8X8_SSD1306_72X40_ER_HW_I2C u8x8(/* reset=*/ U8X8_PIN_NONE); // EastRising 0.42"
void setup(void)
{
u8x8.begin();
Serial.begin(115200);
}
void loop(void)
{
u8x8.setFont(u8x8_font_chroma48medium8_r);
u8x8.drawString(0,0,"123456789");
delay(2000);
Serial.print("Loop End");
}```which u8x8 library are you using?
/* reset=*/ U8X8_PIN_NONE
so they aren't resetting either
fringe yarrow
u8g2 v2.35.10
if its pertinent, this is what I get on boot after a rst button press: ```ESP-ROM:esp32c3-api1-20210207
Build:Feb 7 2021
rst:0x3 (RTC_SW_SYS_RST),boot:0xd (SPI_FAST_FLASH_BOOT)
Saved PC:0x40048b82
SPIWP:0xee
mode:DIO, clock div:1
load:0x3fcd5820,len:0xd0
load:0x403cc710,len:0x7c8
load:0x403ce710,len:0x2518
entry 0x403cc710
Auto-reload is off.
Running in safe mode! Not running saved code.
You are in safe mode because:
CircuitPython core code crashed hard. Whoops!
Heap allocation when VM not running.
Please file an issue with your program at github.com/adafruit/circuitpython/issues.
Press reset to exit safe mode.```
tulip sleet
if its pertinent, this is what I get on boot after a rst button press: ```ESP-RO...
you get the same safe mode message on 8.2.x?
fringe yarrow
yes
tulip sleet
I looked at the arduino repos and don't see a discussion of the reset problem. ... It's suspicious it's showing useful stuff on the display but you are getting that error. It -may- be unrelated to the display after all, or only partly related. You tried with rc.0? Could you also try with the absolute latest build from here? https://adafruit-circuit-python.s3.amazonaws.com/index.html?prefix=bin/01space_lcd042_esp32c3/en_US/
fringe yarrow
I've built from the latest commit for these tests..
tulip sleet
In that case, please open an issue with the safe mode error message above, and say what you see on the screen after a hard reset (or take a picture). Maybe we can simulate this with some other hardware even though we (well, I, maybe someone else does) don't have the board.
is your code.py empty or just hello world?
fringe yarrow
hello world
tulip sleet
we need to figure out how to reproduce this and then do some logging to see what allocation is causing the safe mode
@slender iron rc.1 status:
I have brought the draft release notes up to date, so could release any time.
- ESP32-H2 issue https://github.com/adafruit/circuitpython/issues/9049: I don't have any H2 board. If you have an idea of what the problem is, could work on that. I tentatively marked as 9.0.0 but it's a rare chip. 4 way handshake
- https://github.com/adafruit/circuitpython/pull/9025 needs a re-review, could go in, but also not vital for 9.0.0.
manic glacierBOT
tulip sleet
hello world
Thanks for the issue.
fringe yarrow
thanks for being responsive 😄
slender iron
<@252717193496756235> rc.1 status:
I have brought the draft release notes up to ...
my guess is that the h2 issue is the same as esp32
slender iron
thanks for being responsive 😄
can you do a debug build?
midnight ember
Think this is the first time I remembered to run black before submitting a PR. It still failed. https://github.com/adafruit/Adafruit_CircuitPython_Requests/actions/runs/8284863120/job/22671343869?pr=160 Something about isort?
lone axle
Think this is the first time I remembered to run black before submitting a PR. I...
try running pre-commit locally instead of black directly if you can.
That repo has had an extra check added to it's pre-commit tasks I think if you run it locally it should make the changes for you.
midnight ember
I don't think pre-commit works with Github Desktop. Seems it's looking for git executable but it doesn't exist.
pre-commit version: 3.6.2
git --version: Executable `git` not found
sys.version:
3.10.5 (tags/v3.10.5:f377153, Jun 6 2022, 16:14:13) [MSC v.1929 64 bit (AMD64)]
sys.executable: c:\Program Files\Python310\python.exe
os.name: nt
sys.platform: win32
error information
An error has occurred: FatalError: git failed. Is it installed, and are you in a Git repository directory?
Traceback (most recent call last):
File "C:\Users\Devon\AppData\Roaming\Python\Python310\site-packages\pre_commit\git.py", line 58, in get_root
cmd_output('git', 'rev-parse', '--show-cdup')[1].strip(),
File "C:\Users\Devon\AppData\Roaming\Python\Python310\site-packages\pre_commit\util.py", line 117, in cmd_output
returncode, stdout_b, stderr_b = cmd_output_b(*cmd, **kwargs)
File "C:\Users\Devon\AppData\Roaming\Python\Python310\site-packages\pre_commit\util.py", line 111, in cmd_output_b
raise CalledProcessError(returncode, cmd, stdout_b, stderr_b)
pre_commit.util.CalledProcessError: command: ('git', 'rev-parse', '--show-cdup')
return code: 1
I ran it from the directory I thought it needed. C:\Users\Devon\Documents\GitHub\Adafruit_CircuitPython_Requests\examples\wifi\expanded>
I can get black to work with ```py
C:\Users\Devon\Documents\GitHub\Adafruit_CircuitPython_Requests\examples\wifi\expanded>black requests_wifi_api_rocketlaunch_live.py
reformatted requests_wifi_api_rocketlaunch_live.py
All done! ✨ 🍰 ✨
1 file reformatted.
I updated pip beforehand ```py
Successfully uninstalled pip-23.3.1
Successfully installed pip-24.0
and pre-commit
```py
Successfully uninstalled pre-commit-3.1.1
Successfully installed pre-commit-3.6.2
I wouldn't wait for H2 for 9.0.0 either. Though I can look today
lone axle
I don't think pre-commit works with Github Desktop. Seems it's looking for git ...
theoretically the Github Desktop app must have a copy of git command line tool somewhere internally in order to work. It might be possible to find it and add it's location to your system path so that other programs will be able to find and use it, but I am not sure.
If not that I think you could install Git seperately with installer from https://git-scm.com/download/win (or probably use the portable one if you add it's location to the path). Then pre-commit should be able to find it and continue on with it's process.
It's been a long time since I used Git on windows though so I don't know much of the specifics.
It should also be possible to run isort directly similarly to how you're doing black but I don't know the syntax for it, maybe something as simple as isort <filename> Justin may have an idea how to do it.
midnight ember
yeah i'm trying to track down how to redirect pre-commit to the github desktop git. It's gotta be in there somewhere.
It's installed at C:\Users\UserName\AppData\Local\GitHubDesktop\app-3.3.11\resources\app\git\cmd\git.exe
lone axle
if you add that to your system path it should make it so that it can be run from any CMD instance including the pre-commit one I think. I don't recall the exact way for windows though. Control Panel / System / Environment vars or something maybe.
the directory specifically would get added to PATH Just:
C:\Users\UserName\AppData\Local\GitHubDesktop\app-3.3.11\resources\app\git\cmd\
not the exe.
midnight ember
@lone axle Thank you for all your help. That worked! For some reason even though Github Desktop was using 3.3.11 the path was still using C:\Users\Devon\AppData\Local\GitHubDesktop\app-3.0.3\resources\app\git\cmd simply manually updated the app version path and voila! 🤗
Why Github Desktop doesn't do this automatically is a mystery.
I'd like to archive this here just in case I need to revert for some reason. I feel like this could also be added to the pre-commit learn guide as a special section for Github Desktop users. It will require occasionally keeping an eye on the path env manually. https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/check-your-code
tulip sleet
@slender iron how does the H2 clock PR fix it not coming up? The changes are all in neopixel_write. So i thought it would just be that the neopixel wouldn't work
midnight ember
Ugh because I chose --all-files it ran on every file in the repo and not just in the folder I wanted. :/
tulip sleet
Why Github Desktop doesn't do this automatically is a mystery.
They say install Git for Windows also: https://github.com/desktop/desktop/issues/3801
slender iron
<@252717193496756235> how does the H2 clock PR fix it not coming up? The changes...
the status neopixels was triggering an assert
slender iron
because the clock is 32mhz, not 40
pulse in and pulse out use 1mhz or less so they should be ok
eats lunch and then runs an errand
midnight ember
I keep failing isort. In the past I think it would tell me the sort it wanted. ```py
C:\Users\Devon\Documents\GitHub\Adafruit_CircuitPython_Requests\examples\wifi\expanded>isort requests_wifi_api_rocketlaunch_live.py
Fixing C:\Users\Devon\Documents\GitHub\Adafruit_CircuitPython_Requests\examples\wifi\expanded\requests_wifi_api_rocketlaunch_live.py
next time i run pre-commit it still fails on isort
crimson ferry
this is why I don't do code reviews o_O
midnight ember
went through pre-commit-config.yaml and looked at the version numbers. my isort was out of date. ```
Successfully uninstalled isort-5.10.1
Successfully installed isort-5.13.2
@slender iron I think we are all ready for rc.1 !
midnight ember
I'm using pylint 3.1.0 but the yaml is looking for 2.17.4 is this a problem?
because i'm failing pylint
nvm pylint was just being cranky locally i suppose. PR checks passed. https://github.com/adafruit/Adafruit_CircuitPython_Requests/pull/160
Can someone take a look at this PR from 2023? Apparently it was approved but never merged. https://github.com/adafruit/Adafruit_CircuitPython_PCA9685/pull/55
candid sun
Can someone take a look at this PR from 2023? Apparently it was approved but ne...
merged 💯
midnight ember
merged 💯
Thank you. 🤗 I didn't realize it until someone posted in there today and it showed up in my github notifications.
candid sun
no prob!
midnight ember
After successful merge it offers to delete my working branch. Is this something that is recommended to do or can I just keep it syncd manually in the future?
tulip sleet
After successful merge it offers to delete my working branch. Is this something...
I delete my PR branches, because I don't reuse them. I don't do a "Working Branch". I create a new branch for each PR. That way I can have multiple PR's underway, and the branch names mean something.
a branch is cheap and expendable
orchid basinBOT
midnight ember
I swear I've tried that and it Github Desktop will merge all branch commits under main into main.
midnight ember
Will keep an eye on it and experiment with it. Will delete since I don't plan on working with that library again anytime soon. Can always make a new clone.
midnight ember
Let's say I have working branch 1 & 2. When submitting PR's it wants to merge to my clone main and then submit all the PR's together. Maybe just something with github desktop.
tulip sleet
that writeup is submitting a PR to "development", not to "main", but that would be typical for a larger organization that keeps a separate "main development" branch. We don't do that.
there's nothing unusual about our workflow. One PR per feature or bug fix, not combined
midnight ember
ahh I've been doing it wrong. Must have missed the publish branch when it shows up. I've been doing pull origin/push origin which I suppose just merges all the commits.
tulip sleet
you would push the PR branch, it would ask you if you want to create a new branch. That's what it does in the CLI
midnight ember
Thank you! Slowly figuring things out. This git stuff can be a maze.
tulip sleet
I use the CLI; I don't use GitHub Desktop at all
The Adafruit Learn Guides on git explain the workflow to use. You should be able to do that workflow in GitHub Desktop no problem
midnight ember
I learned a LOT from that article. Recommend a link to that in the Github Desktop section of a learn guide. A lot of great info in there. If it is there then I completely whiffed on seeing it. 🤦
tulip sleet
I learned a LOT from that article. Recommend a link to that in the Github Deskt...
do we have any discussion of github desktop in a learn guides?
we do ... but just for managing credentials
crimson ferry
I only used git CLI once, Github Desktop every time since... I don't know how to do it there, but I use [deprecated] Atom and that makes it easy to select commits to stage
obsidian panther
@slender iron Your esp32h2 fix gets us through a good flash and into REPL, however I'm not seeing a CIRCUITPY drive on either target system. I'm available for device testing if you need anything checked. (do we have a debug build)
slender iron
<@252717193496756235> Your esp32h2 fix gets us through a good flash and into RE...
The H2 doesn't have native USB. so you won't see the drive
Only the S2 and S3 do
obsidian panther
Interesting. So what is the procedure with the H2 for transferring a python program to the device?
obsidian panther
That's clear. Thanks....
tulip sleet
obsidian panther
>>>
>>> import wifi
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: no module named 'wifi'
Are we including 'wifi' in the esp32h2 build?
stuck elbow
the list of build in modules is listed on circuitpython.org for each board
obsidian panther
Doesn't look like the esp32h2 has been added to that list yet.
slender iron
```Adafruit CircuitPython 9.0.0-rc.1 on 2024-03-14; ESP32-H2-DevKitM-1 with ESP3...
I don't think the H2 has wifi support
its the first esp chip without it
manic glacierBOT
fringe yarrow
I see in the u8g2 code, they use "U8X8_START_TRANSFER()" at the beginning of the lcd init sequence, and that seems to mention a built in delay. Perhaps thats a point where their methodology differs from what Circuitpython is doing to init the screen, and why an arduino sketch using that library survives restarts?
tulip sleet
I see in the u8g2 code, they use "U8X8_START_TRANSFER()" at the beginning of the...
could you give me a link in github? I was looking for a delay, but missed that
fringe yarrow
in https://github.com/olikraus/u8g2/blob/74393ef48b860f4a9a18df63c34bcb60a203ba0f/csrc/u8x8_d_ssd1306_72x40.c#L91 it is referenced in the comment, but i haven't dug deeper than that yet.
tulip sleet
in https://github.com/olikraus/u8g2/blob/74393ef48b860f4a9a18df63c34bcb60a203ba0...
I think the delay is part of the init sequence. But the safe mode error is more like some initial I2C problem.
that is not a call but just a sequence of init bytes
fringe yarrow
I've changed the init sequence and it seems to be functioning now
tulip sleet
oooh
fringe yarrow
let me keep restarting it for a while to see if it's just a fluke
fringe yarrow
I changed the ordering, which i'm not sure matters, < 0x2e, 0, //deactivate scroll < 0x40, 0, // start line is 0 < 0xa4, 0, // output ram to display < 0xad, 1, 0x30, // set internal ref to 240uA < 0xd3, 1, 0x00, // display offset added these lines, and changed 0xd9, 1, 0x22 vs the original 0xd9, 1, 0xf1. Basically I copied the u8g2 init sequence for this board..
midnight ember
which one? <https://docs.github.com/en/desktop/working-with-your-remote-reposito...
This one sorry was busy with a project. https://docs.github.com/en/desktop/making-changes-in-a-branch/managing-branches-in-github-desktop I learned a lot from that one. Didn't know you could use a commit to create a branch either. It was very informative. The other one I already knew how to create and issue & PR.
fringe yarrow
i've been continuously restarting it for a few minutes and letting it complete hello, world then restarting it.. It has crashed once in a few dozen boots.. but before it was every reset without fail.
tulip sleet
i've been continuously restarting it for a few minutes and letting it complete h...
I am not that familiar with the init sequences. I wonder if it is missing some part of the initial sequence, but it is less critical when you reordered it.
try going back to the original sequence, but add mp_hal_delay_ms(1000); to the top of board_init(). You may need to add an include for py/mphal.h. If that works try reducing the delay, say to 100msecs or less
fringe yarrow
I'm not sure where the original author got the init sequence, but that 0xd9 is the "precharge period" and the 0.42" OLED's datasheet specifically says it should be 0x22.. and i just reverted to original code, changed that one thing, and it seems stable..
tulip sleet
and if that is not 0x22, then it fails?
fringe yarrow
i dont know about other values, but if its 0xf1 it fails every reset.
acutally.. bizarre.. i just reverted to original code to confirm that statement.. and now it seems stable..
tulip sleet
are you doing a debug vs a non-debug build?
fringe yarrow
no, just a normal make BOARD=
tulip sleet
was your clone up to date when you were building previously? Did you update from upstream?
fringe yarrow
it was up to day as of earlier today, but before the rc.1 branch.. i guess its possible something was introduced there?
but i did just pull about an hour ago..
after rc.1
tulip sleet
I would have expected something from a day or two ago to make a difference, possibly
are your submodules up to date? git status looks good?
fringe yarrow
yes looks good..
tulip sleet
try the circuitpython.org rc.1 if you have not already. Also could be a warm-up thing. Maybe leave it unplugged for a big
fringe yarrow
disregard all that.. for some reason without me having changed anything.. after the 5th or 6th reset, it is now crashing every time again.. (with the unmodified latest build)
tulip sleet
so, try adding the delay I talked about
fringe yarrow
now it seems to wait approximately 1000ms after reset before crashing: ```Terminal ready
ESP-ROM:esp32c3-api1-20210207
Build:Feb 7 2021
rst:0x3 (RTC_SW_SYS_RST),boot:0xd (SPI_FAST_FLASH_BOOT)
Saved PC:0x40048b82
SPIWP:0xee
mode:DIO, clock div:1
load:0x3fcd5820,len:0xd0
load:0x403cc710,len:0x7c8
load:0x403ce710,len:0x2518
entry 0x403cc710
😆
I would like to reproduce this locally to instrument what leads up to the crash. I'm not going to do that right now but the issue is open. I hope I can connect up an external SSD1306 to some board and emulate this board.
fringe yarrow
I dont THINK i'm going crazy.. but.. hear me out.. if i have the board laying flat on the table and tap the reset button, it seems to crash every time.. HOWEVER.. if i'm holding the board such that the pins GPIO3 through GPIO10 are in contact with my finger, it boots every time..
short tendon
LOL
fringe yarrow
this seems more and more like a hardware fault and perhaps i've wasted everyone's time
crimson ferry
is the table conductive?
fringe yarrow
no.. if there is nothing in contact with the board, it crashes.. if i hold it, it boots.. it's very repeatable..
this is either straight up faulty hardware or, possibly, some pin being pulled low through a very high resistor makes it stable on boot..
but the issue only comes up when initializing the display on i2c..
GHOSTS
short tendon
I really want to ask what happens if you boot it while clapping your hands...
Only because I had an SSL issue this week that would work roughly on every other boot, so took FOREVER to find out the problem
manic glacierBOT
fringe yarrow
I'm going order another one because otherwise i'll always wonder.
wraith crow
Well except, it works with arduino code....
wraith crow
Maybe something to do with how CP resets pins on startup, I believe CP lets some pins float that maybe the arduino environment pull one way or the other?
wraith crow
I've been playing around with an esp32/ili9341 display board and decided to build a custom CP image because the python display and touch libraries were using up most of the PSRAM. I've got the firmware built and initalizing the display. The display seems to work fine until code.py exits at which point the display stops working. I can run the python screen initalization and the screen responds but the board.DISPLAY object doesn't work once code.py exits. It's like exiting code.py causes a displayio.release_displays() but I don't understand why that would be happening.
midnight ember
Thats been happening to my st7796s display. Pretty sure its caused by timing issues in the init sequence that I haven’t quite gotten right yet.
wraith crow
Thanks, gives me a place to look 😁
midnight ember
Integrated Connection Manager into my first API. Piece of cake! Thank you for making the transition so easy! https://github.com/DJDevon3/My_Circuit_Python_Projects/tree/main/Boards/espressif/Unexpected Maker Feather S3/4.0 ST7796S/RocketLaunch_Live API Display
wraith crow
Doh! Had a left over LED_STATUS pin definition stomping on the SPI bus of my display. No wonder I was losing the display when the board stopped running code.py.... I'm sort of surprised it worked as well as it did 😁
spare jacinth
I've been playing around with an esp32/ili9341 display board and decided to buil...
out of curiosity, did it save much RAM?
sure, having the "bare" C implementation instead of pure-Py or C+Py-bindings equivalent would take less space, but i want to get an idea how much that is
and i guess it is also helpful as it would now be using another "piece" of RAM, instead of the one reserved for Python's VM (?) 🤔
midnight ember
welp figured out what i was doing wrong with pre-commit. i kept doing pip install pre-commit instead of pre-commit install on the repo. isort started working after that.😅
wraith crow
@spare jacinth I don't think so, my program still didn't run so I didn't look closely at the available space, I just reduced the size of the bitmaps I was trying to load which made a big enough difference that the generic firmware I was using worked. It's getting late or I'd load up the custom firmware again to check the free memory at boot for you but it's going to have to wait. I'll ping you tomorrow with the available memory at boot for both firmware builds.
midnight ember
I think Justin will be happy to hear I updated the new rocketlauch_live requests API example with connection manager too. Added a new commit. Might be the first api example in there with Connection Manager. I plan on switching and updating all of the api examples to use it.
It was a breeze to work with and cut out a couple imports making scripts look a little bit cleaner. I like it!
wraith crow
@spare jacinth The custom firmware resulted in exactly the same gc.mem_free() after a clean (no code.py) boot. The custom firmware initalized the display so moving the display function to the firmware didn't help memory at all. I also froze a few libraries though and doing so increased mem_free after my application loaded by 14208 bytes. The libraries I froze were all adafruit mpy libraries so freezing some of my own py libraries (or even compiling them to mpy) might have increased the available memory even more.
digital shoreBOT
adafruit
vital matrix
(Sorry if I missed a faq answered elsewhere) if someone picks up https://github.com/adafruit/circuitpython/issues/8971 - would you ideally see it land as a community library, or as an adafruit one? (not that I'm signing up 🙂 just checking into adafruit org preferences on drivers for boards also in the store)
tulip sleet
(Sorry if I missed a faq answered elsewhere) if someone picks up https://github...
It would be an adafruit library, because it's our product. Someone internal might be working on a library -- I can check.
There are a couple of other INAnnn libraries that might be starting points, and there is already an Arduino library for the INA228
I might not hear back before Monday
vital matrix
Ack, and thanks for the pointers. [no rush] - just an excuse for me to see how these APIs are structured
tulip sleet
ok - it is in process
midnight ember
pylint doesn't seem to understand some imports ```py
C:\Users\Devon\Documents\GitHub\Adafruit_CircuitPython_Requests\examples\wifi\expanded>pylint requests_wifi_api_mastodon.py
************* Module requests_wifi_api_mastodon
requests_wifi_api_mastodon.py:9:0: E0401: Unable to import 'adafruit_connection_manager' (import-error)
requests_wifi_api_mastodon.py:10:0: E0401: Unable to import 'wifi' (import-error)
any advice on how to make pylint happy here?
I used # pylint: disable=import-error if that's a bad idea please let me know.
short tendon
Do you have pre-commit installed?
midnight ember
yes
short tendon
What happens if you run pre-commit run --all-files?
midnight ember
it will run on everything in the expanded directory
short tendon
It may be because you are in the examples directory
And that's okay, it will take a little more time, but everything else will pass
midnight ember
black failed immediately
short tendon
On the new file?
midnight ember
on --all-files
short tendon
But it should list what files it failed on
midnight ember
the list is about 1000 lines of errors
most of them are the unexpected line ending error. tons of them.
midnight ember
well isort works so that's progress
short tendon
And it looks like black fixed it's issues
I wouldn't worry about the line ending issues
midnight ember
oh is it fixing them?
short tendon
Pylint doesn't fix, but that's okay. By default git will commit with the right line endings
This is an area that Windows and the rest of the world fight every day...
Are there any other errors that aren't line endings?
midnight ember
This runs in Github Desktop which it never did before so that's pretty neat. I feel like I'm really close.
short tendon
If it's just line endings, I would commit and push. You should be okay
midnight ember
it won't let me. this is the window that pops up when i attempt to commit.
midnight ember
When I use cmd line it gives me a score of 10/10.
C:\Users\Devon\Documents\GitHub\Adafruit_CircuitPython_Requests\examples\wifi\expanded>pylint requests_wifi_api_mastodon.py
---------------------------------------------------------------------
Your code has been rated at 10.00/10 (previous run: -0.29/10, +10.29)
Then when pushing commit pylint runs inside Github Desktop and fails me but sees the previous 10/10 from the CLI.
It's like Github Desktop is running a different pylint than I am from cmd line.
short tendon
git commit --no-verify -m "message here
short tendon
Add the --no-verify
And reading for the weekend: https://docs.github.com/en/get-started/getting-started-with-git/configuring-git-to-handle-line-endings
midnight ember
C:\Users\Devon\Documents\GitHub\Adafruit_CircuitPython_Requests\examples\wifi\expanded>git commit --no-verify -m "messge here"
[MastodonBranch f88e6b4] messge here
1 file changed, 69 insertions(+), 73 deletions(-)
And now you can git push
midnight ember
with cmd line or github desktop?
i've never used cli for git. such a rabbit hole. 😬
midnight ember
I went to Github Desktop and it was just sitting there ready to be pushed. Hit push and it worked. https://github.com/DJDevon3/Adafruit_CircuitPython_Requests/tree/MastodonBranch
Am I going to have to do this every time? 🥹
short tendon
No, you need to update your git checkout settings
midnight ember
I'm trying to create the 2nd separate PR but it seems like Github wants to automatically merge them. I usually do the actual PR from the Github website.
short tendon
Are you still on the other branch?
midnight ember
yes
short tendon
Are you trying to commit to that branch or a different one?
midnight ember
I have a PR for the rocketlaunch.live api example
but i don't want to merge this one with that and overload a reviewer with multiple api examples
so i'm trying to do 2 separate PR's from 2 different branches
Just want to make sure that what this looks like isn't going to merge with my other PR, because I've had that happen before.
the first PR is from "WorkingBranch" and this one is from "MastodonBranch"
short tendon
That image shows from your to Adafruit's. So should be good. Do you just see the one file changed?
midnight ember
yes i'm just hesitant because of what happened last time.
hey it worked 🎉 Danh will be proud. have 2 separate PR's from 2 separate branches.
Thank you 🙏 There's no way I would have figured that out on my own.
and it passed on the first attempt whaaaaat. 😯 https://github.com/adafruit/Adafruit_CircuitPython_Requests/pull/161
short tendon
Niiiiice!!!!
midnight ember
I don't understand why isort is doing this ```py
import os
import time
import adafruit_connection_manager
import wifi
import adafruit_requests
usually imports for circuit python have no spaces between imports.
short tendon
That's what isort does:
- Python imports
- pip imports
- local imports
midnight ember
ohhh is this going to be a new thing?
well if it's all circuit python then shouldn't they all be local imports?
or does isort recognize that os & time happen to also be the names of python imports?
short tendon
So isort knows what's included in python like os what's not but isn't local like wifi and what's local like requests
thorny jay
Thank you @vagrant wind for the M5Stack matrix keyboard.
I am trying to create the map from key_number to USB HID keycode with code like this from Jepler: https://learn.adafruit.com/ibm-pc-keyboard-to-usb-hid-with-circuitpython/coding-the-keyboard
I just hope that was not done in a stream by @lone axle or by you.
vagrant wind
Thank you <@658337305454444574> for the M5Stack matrix keyboard.
I am trying to ...
No, Ihaven't done it but it was the next thing that was needed. I'm glad if you can do it
thorny jay
No, Ihaven't done it but it was the next thing that was needed. I'm glad if you ...
I plan to do the "first layer" just to identify the various "location" and the matching keyid. And I plan to do it with your pattern GPIO order:
km = keypad.DemuxKeyMatrix(
row_addr_pins = (board.KB_A_0, board.KB_A_1, board.KB_A_2),
column_pins = (board.KB_COL_0, board.KB_COL_1, board.KB_COL_2, board.KB_COL_3, board.KB_COL_4, board.KB_COL_5, board.KB_COL_6)
)
So in my ongoing quixotic quest to add support for unpopular devices, I've started working on a "ESP32-2432S022C" which has a SD slot, touch screen and one amplified GPIO for speaker. I have gotten all those functioning, though I had to use a modified adafruit cst8xx library, as the cst8xx touch in the device reports a chip id nowhere close to what the library expects, but works if the library is set to ignore that. I have run up against nothing but trouble trying to get the display working. It's a 240x320 lcd driven by an ST7789 in parallel mode (not spi). I believe I have all the pin configurations configured appropriately, I have an init sequence that works, gleaned from an arduino library that did make the display work correctly, but no matter what I try, when booting circuitpython and trying to run display tests, the display always looks like this:
midnight ember
pretty sure there's a circuit python st7789 driver. it's what i based the st7796s driver on. they're almost identical drivers except the st7789 has inverted colors and required a specific madctl sequence in the init sequence in comparison.
ah but you're running it in parallel mode, nvm.
just woke up, haven't had my coffee yet, ignore me
fringe yarrow
the lilygo-tdisplay is the only board with built in parallel display that has board definition files.. I've copied mostly from that, but since it's a different lcd, i got the screen working with the Arduino_GFX library first, then dug the init sequence out of that and put it in my board.c.. before the screen never even initialized, so it's progress.. but why it's lines and garbage, i haven't yet figured out.
thorny jay
No, Ihaven't done it but it was the next thing that was needed. I'm glad if you ...
Dear @vagrant wind here is what I could do (I hope it help):
https://gist.github.com/dglaude/6f2a3f9c3c1e2f1bdf055245f1a2edb7
You can reuse code.py, or check the table to fix some character mapping.
I am 90% satisfied, I took ESC for the upper left, I don't know what to do with the "fn" key, some special character might be wrong but this is a good start.
I did not compare with the demo code provided with the card computer (the USB and BT keyboard part of the app).
wraith crow
Dear <@658337305454444574> here is what I could do (I hope it help):
https://gis...
Thanks for this 😁 , it made updating my virtual repl in preperation for EDario's merge a snap. While I was at it, I threw in CAP/FN locks and implemented held key repeats
thorny jay
Thanks for this 😁 , it made updating my virtual repl in preperation for EDario'...
Happy to help... PyDos and @lone axle recent streams are the reason for getting one and testing the keyboard support in CP.
manic glacierBOT
manic glacierBOT
manic glacierBOT
aren't these pull up resistors on the relevant lines? (it is weird that there are 3 R4s on the schematic though, there's another on on IO8)
manic glacierBOT
wraith crow
Happy to help... PyDos and <@382939733107408897> recent streams are the reason f...
Cool, if you're using the artifact with the DemuxMatrix and PyDOS you should grab the version of m5stackcardputerkbd.py from the cardputer_repl github repo. It improves the Shift/FN key behavior, adds key repeat and performs a CAP lock/FN lock with the OPT-Shift/OPT-FN key combination.
wraith crow
@fringe yarrow At one point did you mention that you had a chart comparing the differences between various "Cheap Yellow Display" versions? If so, is it somewhere I can grab a copy?
manic glacierBOT
fringe yarrow
<@239506299044233216> At one point did you mention that you had a chart comparin...
https://github.com/rzeldent/platformio-espressif32-sunton/blob/main/esp32-2432S022C.json has a bunch of different cyd playformio definitions meant for use with the lvgl graphics library.
fringe yarrow
<@239506299044233216> At one point did you mention that you had a chart comparin...
I've got the 2432S028R en route. Should be here this week. I think that one has spi screen and should be simple to make work.
wraith crow
@fringe yarrow Mine is the 2432S028, it has the ili9341 spi screen/XPT2046 touch screen and works well with CP. I've used pre-built esp32 firmware from other devices and I also built a custom CP firmware for it. It has an SD card slot, speaker amplifier (pwm) and I've gotten I2C working on one of the JST connectors.
fringe yarrow
<@239506299044233216> Mine is the 2432S028, it has the ili9341 spi screen/XPT204...
Yeah the 22c is similar in all respects except it has a parallel wired screen which im continuing to work at, but at this point I think it will take some reworking of the commonhal parallelbus code which is currently only used by the lilygo tdisplay and has some magic numbers and assumptions that I don't think hold true for this screen.
fringe yarrow
<@239506299044233216> Mine is the 2432S028, it has the ili9341 spi screen/XPT204...
You going to submit a pull request for your custom build? 😄
wraith crow
I don't think so, it sounded like there were too many similar boards for Adafruit to try and support. I may spend a little time looking at the variants and see if there's some sort of generalized build that could work but at first glance it doesn't look promising. I did push my board files to https://github.com/RetiredWizard/micropython/tree/cheapyellowdisplay/ports/espressif/boards/cheap_yellow_display if you're interested for when your board comes in.
midnight ember
just to document this for myself. when working with isort it wants things in a very specific order (and does not re-order them for me so I don't see the point).
Windows cmd line
First do: black requests_wifi_api_fitbit.py # to clean up formatting
Next do: pylint requests_wifi_api_fitbit.py # clean up any errors pylint has until `Your code has been rated at 10.00/10`
Next do: git add .
Next do: git commit --no-verify -m "Update Fitbit API Example with Connection Manager" # add commit description
Go to Github Desktop and the push should be ready. Click push.
Go to Github Website for the repo, create PR, edit description and details, submit PR.
sit back and usually watch it fail on isort
What do you mean it doesn't order them for you?
isort --profile black .
Should update everything for you.
Or before you commit:
pre-commit run
Which will run pylint, black and on this repo isort
orchid basinBOT
midnight ember
```
isort --profile black .
```
Should update everything for you.
Or before you ...
Sometimes it does, sometimes it doesn't. Haven't quite figured it out yet. What I "should" have to do and what I have to do are 2 diferent things unfortunately. :/ It probably had to do with using Github Desktop. I'd love to figure a way for Github Desktop to do all of this for me before a commit.
manic glacierBOT
short tendon
Sometimes it does, sometimes it doesn't. Haven't quite figured it out yet. Wha...
You can set it up that on commit that pre-commit runs. Did you fix your line endings?
Running:
pre-commit install
Will make it run before you commit (and fail if there are errors, so you need to fix and try again).
Or
pre-commit install --hook-type pre-push
Will make it run before you push (I prefer this personally).
Either one will work with git desktop or CLI
midnight ember
I've been doing them individually C:\Users\Devon\Documents\GitHub\Adafruit_CircuitPython_Requests\examples\wifi\expanded>pre-commit run --files requests_wifi_api_premiereleague.py is this a bad way to do it?
short tendon
No, other than you need to remember what's been changed. Typical work flow would be:
- Change filed
git add file_a,git add file_bpre-commit run- Fix issues
- Repeat from #2 until no issues
If you are doing one thing at a time git add .
midnight ember
I've been wanting to learn how to make separate PR's since October of 2022. Because all of my PR's merged into 1 commit Hacktoberfest only counted them as 1 instead of 5 and I got screwed out of getting a hacktoberfest t-shirt for not making enough PR's. 😤
short tendon
What IDE do you use?
midnight ember
Mu
oh and Notepad ++ to actually save the file into the local github repo folder. Maybe that's where a lot of line ending errors are coming from...
I copy from Mu to Notepad++ and hit save.
short tendon
Ahhh, I might recommend pycharm. The free version is still quite powerful and allows selecting exactly what you want to commit (like just these x changes from a file)
I think you can do some of that in git desktop too
midnight ember
So I have a working file in Mu and a file to save directly to the local github folder.
I've been thinking maybe it might even be easier just to copy/paste the file to github website PR somehow. Let adafruit's github actions do it all?
short tendon
Tomorrow I'll send you a screenshot of how I use pyCharm. The pro version (sadly, not sure why it isn't in the free one), allows you to pick a folder and sync it to your device...
midnight ember
I'll try pycharm. I do have it set as the default editor for .py files 🤦 I have a decade+ of habit of using Notepad++ for quick edits.
short tendon
Which for you and your monster repo would work really nicely
midnight ember
I do have pycharm syncd to my circuit python device it's just that Mu repl I find much easier and more enjoyable to work with.
short tendon
I use sooo many things. Each has their purpose. Old enough to still use vim for a lot of things, text editors for quick and dirty, vs code for multi project and pycharm for serious work (and that's just my high use ones)
Ahhh, yeah, I use tio which works great for me
midnight ember
This is true. I see FoamyGuy using nano sometimes. I mean if it works it works. I did forget that Notepad++ has a really bad habit of not using UTF-8 line endings. JAVA used to complain a lot about that.
Might just be as simple as reconfiguring Notepad++ to be more python friendly.
midnight ember
Yeah I have no problem with anyone using anything as long as it works and in this case I think I might be shooting myself in the foot by forgetting about Notepad++ issues.
a simple issue in my workflow I completely forgot to mention... 😬 sorry
FoamyGuy's been knocking out my PR's but as he merges 1 I submit another haha!
short tendon
You could get all yours open in draft, and then mark them all ready when you are done...
midnight ember
That's what isort does:
- Python imports
- pip imports
- local imports
oh i think i see how isort wants things. each import section is then alphabetized?
fringe yarrow
I don't think so, it sounded like there were too many similar boards for Adafrui...
I think if anyone was going to push one of these CYDs for approval into CircuitPython, it should be the 2432S028. It seems to be, by far, the most popular one when I'm trying to find pinouts. Any search for the 22 brings up results for the 28.. I've taken your code and incorporated some things I hadn't done yet into my local branch. Hopefully it shows up this week.
empty salmon
@short tendon - only to fill out the "IDE" discussion, I use VSCode for all my work but recently evaluated IDEs for a workshop series I will be teaching.
I settled on Thony.
Thony has one quirk and once you explain it, things make sense ... and even make for a better workshop experience.
That "one thing" is that Thony, by default, lets you edit files on your computer and "run" them on CircuitPython without actually saving them to the microcontroller. Thony will just stream the code into REPL.
The quirk is to remember that the code won't persist on the microcontroller unless you save it to the microcontroller as code.py.
manic glacierBOT
short tendon
<@1152454697614069811> - only to fill out the "IDE" discussion, I use VSCode for...
I need to play with thony. I've heard good things. Since Python is my primary language these days, I like the simplicity of using one IDE, although of course that's across multiple computers since I need to keep work and personal separate...
midnight ember
Used Pycharm this time and no difference. pre-commit run still complained about a hundred unexpected line endings. So it's not Notepad++ fault. If I copy/paste everything from Mu to PyCharm then perhaps the improper line endings are getting copied with it. Maybe it's Mu?
This time I copied the file from Mu to PyCharm. Saved the file with PyCharm. ```py
pre-commit run
pylint requests_wifi_api_github.py
git add .
git commit --no-verify -m "Update Github API Example with Connection Manager"
I think some of the issues with isort failing I can work around if I manually setup the imports how it likes it anyway.
or maybe that's the point? I mean Pylint doesn't fix all errors it just tells you where they are. isort only tells me there's a sorting issue it doesn't show me the order it wanted. now that i know the order for the most part i can just work around it.
short tendon
So git and windows checks out files with crlf by default, but that doesn't work when pylint is expecting just lf. You will need to switch the default for this repo (and any Adafruit repo) so it leaves them as-is
midnight ember
the reason i want to document this here is even though i'm committing a lot this weekend i might go months without making a PR. i guarantee you i'll need to look this back up in the future.
short tendon
The git link I sent the other day should help you switch. Might be worth getting some documentation updated for that if it doesn't exist yes
midnight ember
yes yes i thought about that just before you typed it. i didn't do my homework this weekend. 😅
too busy making PR's!
short tendon
There's also tomorrow... 😜
midnight ember
I wanted to try to get them all done on the weekend in a blitz. the fitbit api is complicated due to the use of nvm and took a whole day. :/
short tendon
I hope you are having fun
midnight ember
I definitely am now that I can submit PR's more easily... and as separate PR's from branches. 🏗️
fine i'll go read that line ending stuff before working on more api examples. ugh homework. 😛
ok i did this yesterday git config --global core.autocrlf true but everything i'm working with is still producing those errors with pylint. do i need to checkout a fresh clone to make it work or something?
tulip sleet
ok i did this yesterday `git config --global core.autocrlf true` but everything ...
When I installed git for Windows I took its defaults and it did the right thing. But I rarely use git on Windows
midnight ember
I did look at the other program you suggested but I do prefer how easy github desktop is to use in most circumstances. it's only problematic when i attempt to make a PR for adafruit. on my personal repos i don't have to run any actions so normally it never complains.
tulip sleet
I did look at the other program you suggested but I do prefer how easy github de...
if you are going to use git in a terminal window install git for Windows. Don't use the git that's internal to github desktop. I posted a link that noted that.
midnight ember
yes but i'm finding some work arounds 😅
and it does actually integrate with adafruit actions. i just need to find a way around these line ending errors and should be good.
tulip sleet
I'll repeat the link: https://github.com/desktop/desktop/issues/3801
this is official advice from github
midnight ember
the only reason i need the terminal right now is to add . and commit no verify but that does also work in github desktop. i'm sure when i fix the line ending errors it'll work fine.
to be clear i'm not ignoring your advice. i'm just trying to find workarounds to keep my workflow somewhat the same as it was.
tulip sleet
i was just wondering whether you got the wrong line endings because of not using a version of git that was not installed with the git for windows defaults
midnight ember
only other git i've installed was WSL git but I think that should be separate and only exists in WSL Ubuntu.
I'm running the git.bat recommended at the end of that github issue you linked to. It should keep up with github desktop versions now if it works as advertised. Closed and reopened github desktop so it seems to work so far.
short tendon
@tulip sleet the .pylintrc specifically calls for:
expected-line-ending-format=LF
Meaning the default GIT windows behavior auto of checkout crlf, commit lf doesn't work.
I am often the only windows dev in the Mac/unix world and default to everything as lf
tulip sleet
i guess we might consider changing that to nothing. It's either nothing (meaning don't care), LF, or CRLF
but the issue has been CRLF's infiltrating into the repos.
midnight ember
i've been using circuitpython_base64 as base64 a community library to do base64 conversions. should i be using binascii instead for examples? the opensky api examples are using them. wondering it would be better to switch to something that's built in?
tulip sleet
i've been using `circuitpython_base64 as base64` a community library to do base6...
binascii will be faster
short tendon
but the issue has been CRLF's infiltrating into the repos.
I saw it in the doc too, but adding a .gitattributes might help. I've seen some success in groups. As long as you find the right settings...
tulip sleet
but we would need to add it to all repos?
or is it something you wouldn't check in, or could be global?
I have to sleep, sorry
short tendon
Sadly to all repos, so could go into cookiecutter, and then applied as a patch to all existing repos...
midnight ember
binascii worked like a treat.
vagrant wind
Dear <@658337305454444574> here is what I could do (I hope it help):
https://gis...
Excellent work!
orchid basinBOT
slender iron
@fringe yarrow the display can look like that if the memory hasn't been written
that can happen if the bounds of the display are incorrect
fringe yarrow
<@239506299044233216> the display can look like that if the memory hasn't been w...
Could be.. It's a 240x320 screen, which I believe is the max resolution of the st7789.. I've tried defining it as 240x320 and 320x240 with no joy
slender iron
Could be.. It's a 240x320 screen, which I believe is the max resolution of the s...
that doesn't seem likely then. does it flicker like data is being written?
fringe yarrow
that doesn't seem likely then. does it flicker like data is being written?
No, unchanging.. I've delved into common-hal/paralleldisplaybus/ParallelBus.c for the last couple hours.. Found a few things that I thought were ah-hah moments, like how there's several config parameters not being set by the code, and the defaults in the esp-idf struct definitions are not the same as this board's configuration.. also, there was no code for performing a software reset vs a hardware reset depending on the absence or presence of a reset pin, etc.. nothing has made a spec of change 🙂
slender iron
have you used a logic analyzer to confirm it is outputting what you expect?
fringe yarrow
no, i don't have such things handy..
slender iron
hrm
onyx hinge
Looking forward to seeing my <@&356864093652516868> pals in about 1 hour. Please take a moment to add items to the notes doc if you can! https://docs.google.com/document/d/1A5vXAYyCHsXmquGVkXvxFn0wosleHI8NHVKkD6Zr4S4/edit?usp=sharing
slender iron
No, unchanging.. I've delved into common-hal/paralleldisplaybus/ParallelBus.c f...
what speed are you running it?
try lowering it
fringe yarrow
running which? the frequency passed to common_hal_paralleldisplaybus_parallelbus_construct_nonsequential?
The board specs I found say 12mhz, which is 3 mhz less than the st7789 datasheet says is minimum 🙂
or, no,.. f 1/t, so 15 is the maximum.. disregard..
i was running at 12mhz.. will downclock to 5mhz to see if it makes any difference
nope, same..
vagrant wind
I'm trying to build the circuitpython documentation with 'make html' but it fails. Is there a guide how to setup the enviroment to build circuitpython documentation?
tulip sleet
I'm trying to build the circuitpython documentation with 'make html' but it fail...
pip3 install -r requirements-doc.txt did you do that
you may need to update some of those pip3 installs if they are already installed but old
also do a make clean at the top to start
slender iron
running which? the frequency passed to `common_hal_paralleldisplaybus_parallelbu...
yup, that's what I'm thinking
solar whale
I noticed that in the sdconfig files for the esp32s3 boards there is a line like ```# LWIP
CONFIG_LWIP_LOCAL_HOSTNAME="espressif-esp32s3"
end of LWIP
This is just a hostname change. There is some default hostname. It's not going to affect size in any significant way.
solar whale
Thanks -- it ssemed too easy 😉
fringe yarrow
yup, that's what I'm thinking
Running it any higher than 12mhz, it doesn't init, just bright white screen.. I can bring it down to 3.25mhz and nothing changes.. any lower and it doesn't init,..
solar whale
This is just a hostname change. There is some default hostname. It's not going t...
What about either of these OPTIMIZATION_FLAGS = -Os CIRCUITPY_PARALLELDISPLAYBUS = 0 I would expect the optimization will make it use more RAM, not less.
tulip sleet
that is a flash size optimization for smaller flash. i dont' think it will make much of a RAM difference either way
slender iron
I noticed that in the sdconfig files for the esp32s3 boards there is a line like...
I want to remove this in the long run. There's probably a function we can call to do the same
solar whale
I just wanted to make sure it did not invoke some completely different part of lwip.
onyx hinge
@lone axle do you still make back-up recordings? I upgraded my Debian since the last time I ran a meeting so I'm a little extra worried I might run into trouble (pre-meeting checks all look OK though)
lone axle
<@382939733107408897> do you still make back-up recordings? I upgraded my Debian...
I do still try to run backup recordings each week.
onyx hinge
OK I'll breathe a little easier then 🙂
turbid radish
👏
midnight ember
I missed a hug for @candid sun for some quick PR reviews and feedback. 🤗
@lone axle I like how approachable your cardboard projects are. No special enclosure needed.
@solar whale I'm sure it will help. I really want to get around to using your library. Congrats on your first PCB! 🎉
link in notes doc doesn't work @solar whale
" This project is owned by another user. " even when I'm logged in.
solar whale
hmm -- not sure what to do to fix that .
short tendon
I think there is an option for a permalink...
solar whale
I just "shared" it
midnight ember
Was nice to see the Hancho dev stop by here too.
solar whale
midnight ember
That one works. The one in the notes doc doesn't for some reason.
solar whale
ah -- here it is https://oshpark.com/shared_projects/M1wdG7Iq
midnight ember
that works too 👍
solar whale
your mileage may vary 😉
midnight ember
isort... my nemesis.
solar whale
That one works. The one in the notes doc doesn't for some reason.
fixed in notes.
short tendon
That might make the next PR kinda complicated...
@lone axle let me know if I can help in anyway with looking at ruff
solar whale
Will this be automatic via "pre-commit" or will we all have to install/learn to use the new tool?
midnight ember
CRLF vs LF
slender iron
Will this be automatic via "pre-commit" or will we all have to install/learn to ...
I think we'd make it work through pre-commit
midnight ember
It's likely just my environment if it's not affecting anyone else's commits (from Windows users).