#circuitpython-dev
manic glacierBOT
orchid basinBOT
The weird bit is sometimes I see US english in one box and UK english in another, but when recreating in an incognito window both my language dropdowns were prepopulated with UK.
These are the recent versions where I've ended up with en-US (occasionally I may have left at en-US so take this list with a pinch of salt)
orchid basinBOT
Looks like the second language drop down doesn't get set if the web installer button isn't found in the first firmware box:
download.js:49 Uncaught TypeError: Cannot read properties of null (reading 'setAttribute')
at download.js:49:50
at Array.forEach (<anonymous>)
at updateFileLinks (download.js:46:9)
at findAndSetLocale (download.js:63:11)
at downloa...
tepid trail
I am trying to get the neopixel library running on a raspberry pi 5. Has anyone got it to work?
random junco
I am trying to get the neopixel library running on a raspberry pi 5. Has anyone ...
I would recommend asking in #help-with-linux-sbcs for Pi questions - this channel is for developing CircuitPython
tepid trail
Its a circuit python library. https://docs.circuitpython.org/projects/neopixel/en/latest/
onyx hinge
This is probably the best spot to discuss development of Blinka, which provides the low level hardware routines on SBCs. I don't know anything about this topic in particular, but there's an existing issue in which the primary Blinka dev acknowledges that the functionality doesn't exist right yet: https://github.com/adafruit/Adafruit_Blinka/issues/747
"RuntimeError: ws2811_init failed with code -3 (Hardware revision is not supported)"
manic glacierBOT
slender iron
Thanks for the example. That enumerates all of the combinations but doesn't really show me how it'd integrate into things. The nice thing about a bunch of classes is that it can be documented
midnight ember
It's done. I don't want to see another Pylint error for a couple months. https://github.com/adafruit/Adafruit_CircuitPython_IS31FL3731/pull/54 @lone axle I cannot begin to explain the level of frustration that caused.
fringe yarrow
I got a new ESP32 board in the mail.. It feels wrong that I didn't have to make a new Circuit Python board for it.
candid sun
<@&356864093652516868> We'll have our weekly meeting in about 90 minutes from now in this text channel and in the circuitpython voice channel. Please take the time to add your notes in advance to the document: https://docs.google.com/document/d/1qribXknsk4RKNPSjjIPpvTiuhSzboGrLf4m8cRLw9JI/edit -- I look forward to everyone's updates!
onyx hinge
@candid sun I'm happy to grab the "core" section if you like
candid sun
@onyx hinge thank you!
onyx hinge
dangit I know I have one of these but I just cannot find it anywhere 😦 https://www.adafruit.com/product/1667
midnight ember
bme280 is temp/humidity/baro sensor
it's my fav sensor
oh the 680 nvm. it's slightly different but in the same family of temp senosrs
mp3's with the s3? 👀 YES PLZ!
the bear and erins never ending story are awesome projects. will be cool to see it working with bookworm.
lone axle
I think it could be supported in adafruit_imageload library. But the specifics of the implementation are over my head too.
turbid radish
Thanks Liz!
onyx hinge
thanks liz
gilded cradle
Thanks
wraith crow
Thanks
lone axle
Thanks for hosting Liz. Hope everyone has a great week!
midnight ember
Thank you Liz for hosting. Looking forward to the CP version of the S3 BLE guide, someday. 🙂
gusty siren
Hmm, I can't comment on this Pull Request, but it looks like it is ready to be merged? https://github.com/adafruit/Adafruit_Learning_System_Guides/pull/2752
candid sun
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>
https://docs.google.com/document/d/15BPPqHPszkw6a2gGAYxj4CmFXQuhpvArvodMLLIJVJ0/edit?usp=sharing
cunning sundial
Made this with CircuitPython and the Circuit Playground Express. Nice having sensors/LEDs built-in: https://www.youtube.com/watch?v=00ZI6rT8ne8
orchid basinBOT
tulip sleet
https://blog.adafruit.com/2024/04/16/circuitpython-9-0-4-released/
- Fixed crash when writing to external flash when heap storage is low.
- Fixes to nRF AnalogIn.
midnight ember
Made this with CircuitPython and the Circuit Playground Express. Nice having sen...
Please post that to #show-and-tell
fringe yarrow
Is there something wrong with the PR process at the moment? My last PR failed on the docs runner with errors that have nothing to do with anything involved with my PR.
Processing ./circuitpython-stubs/dist/circuitpython-stubs-*.tar.gz
ERROR: Could not install packages due to an OSError: [Errno 2] No such file or directory: '/home/runner/work/circuitpython/circuitpython/circuitpython-stubs/dist/circuitpython-stubs-*.tar.gz'```I am just investigating this. It is due to a recent fix in setuptools, which no longer changes an _ to a -
anyway, don't worry about it right now
or maybe it now changes a - to an _ but it didn't before. It's a bit confusing
tulip sleet
@slender iron @onyx hinge I am fixing the doc build issue, per above. It broke the docs build for the 9.0.4 release, but not a big deal
onyx hinge
ah that makes some degree of sense. the "punctuation folding" where "-" and "_" are sometimes interchangeable in package names is 😕
thank you !
lone sandalBOT
tulip sleet
@short tendon I thought you had a written an Adafruit Playground note about ConnectionManager but I cannot find it
fringe yarrow
How goes the battle on that Doc runner, @tulip sleet ? Seems like you've been fighting it a while now 🙂
tulip sleet
How goes the battle on that Doc runner, <@329766224093249548> ? Seems like you'...
it iis fixed for 9.0.x, and I will make a PR to update main from 9.0.x in a few minutes
@lone axle when you make a library release, could you add the version number as a prefix to the release title, e.g.:
https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/releases/tag/5.0.9
This makes the release number really obvious
lone axle
Yes I'll go back to doing that. I was in that habbit for a while but have slipped out of it.
tulip sleet
I wish the release tag was in the right sidebar, but it is not
onyx hinge
@short tendon I was looking at the requests examples, specifically simpletest for wiznet. I think you were working in here recently due to connection manager. Anyway, is there a reason (besides inertia) to keep the style of response = requests....; ...; response.close() (which I would tend to view as error-prone, especially in the face of exceptions) vs the implicit close of: with requests...() as response: ... way? Not suggesting you need to change it, just wondered if you knew a reason it is like it is.
fringe yarrow
it iis fixed for 9.0.x, and I will make a PR to update main from 9.0.x in a few ...
Should this be working now?
tulip sleet
Should this be working now?
It isn't merged into main yet: https://github.com/adafruit/circuitpython/pull/9187
tiny peak
@slender iron I just committed a couple more prebuilt DFU files to my ocfpga repo that I built from orangecrab-examples/verilog/blink and orangecrab-examples/verilog/pwm_rainbow. I flashed them over JTAG, but I think they should work with dfu-util. There's also a Makefile I wrote to download and build yosys 0.40, nextpnr 0.7, and libtrellis 1.4 tools from source. I guess now I'm about to the point where it might make sense to try building a SoC.
short tendon
<@1152454697614069811> I was looking at the requests examples, specifically simp...
Most of these I just made small updates. I think these could all be updated to use with.
It actually brings up another question I keep forgetting to ask. Calling response.json() actually calls close and .text() does not. Any idea why? CPython requests doesn't do this
manic glacierBOT
lone axle
Most of these I just made small updates. I think these could all be updated to u...
I don't have a definite answer to why, but I did get curious and look back through the history of the commits. It looks to me like originally both text() and json() used content property internally and inside of content() it had the call to close(). So originally they both did get closed. But in pr #31 that call to close() was taken out of content() https://github.com/adafruit/Adafruit_CircuitPython_Requests/pull/31 when the library was updated to re-use sockets more. In that same change the close() was added inside of json() but not text()
As long as removing that close() in the current json() doesn't cause any trouble (and I don't know of any reason it would) I'm in favor of removing it to get closer to parity with CPython requests.
tulip sleet
I don't have a definite answer to why, but I did get curious and look back throu...
A PR for this would great.
tulip sleet
@onyx hinge @slender iron got time to approve ? thanks
https://github.com/adafruit/circuitpython/pull/9187 (sphinx fixes and other stuff from 9.0.x; will fix current CI rtd and doc failrues) https://github.com/adafruit/circuitpython/pull/9188 (update frozen modules)
the build failures on these PR's are due to the sphinx issue and will be fixed by #9187.
onyx hinge
looking @tulip sleet
tulip sleet
I will start on a 9.1.0-beta.1 after these are merged
tulip sleet
@onyx hinge tnx!
crimson ferry
there are no UF2s for pre-CP8 anywhere, right?
could we consider retaining the last version of each major version number? (good for regression testing)
stuck elbow
there are no UF2s for pre-CP8 anywhere, right?
there are some https://github.com/adafruit/circuitpython/releases?page=18
crimson ferry
3.0 lol
that was before my time
I have to go see if I still have 7.x.x installed on some board
short tendon
I don't have a definite answer to why, but I did get curious and look back throu...
I wonder if it would. I can imagine a bunch of non-ideal code out there where close isn't directly called and the exit handler never runs, and so you update libraries and you start running out of sockets...
manic glacierBOT
slender iron
Nice @tiny peak ! Will have to look next week
@vital matrix I started a port for renode was talking to the renode folks this week
tulip sleet
there are no UF2s for pre-CP8 anywhere, right?
We save every release, including alphas and betas: https://adafruit-circuit-python.s3.amazonaws.com/index.html?prefix=bin. We just stopped making GitHub artifact releases. Note there are OLD/ directories in some board directories.
@stuck elbow ^^
orchid basinBOT
manic glacierBOT
lone axle
I wonder if it would. I can imagine a bunch of non-ideal code out there where cl...
If we go through the libraries and update any that use json() to either call close explicitly or use the context manager would that mitigate most of the possible risk?
Anything remaining would be specifically user code outside of libraries that uses json() without the manager or call to close(). The solution for it there is the same, to either use the context manager or call close() which will make the code more closely compatible with CPython anyhow so that is a good change to make if I am understanding everything correctly.
orchid basinBOT
short tendon
If we go through the libraries and update any that use `json()` to either call c...
It would, and as long as enough people helping know about the change, it should be okay
short tendon
@lone axle should I add that change with the use of with in the examples?
Or keep it separate?
lone axle
<@382939733107408897> should I add that change with the use of `with` in the exa...
The change to remove close() from json()? if so I think it's okay to do that at the same time as updating the examples.
short tendon
Perfect
midnight ember
The change to remove close() from json()? if so I think it's okay to do that at ...
I'm cleaning up a few other examples this weekend and then will start the process of changing all uses of .close() for with statements.
short tendon
I'm cleaning up a few other examples this weekend and then will start the proces...
I've already started on this one, making some other changes with it to fix some other issues.
midnight ember
I've already started on this one, making some other changes with it to fix some ...
Sounds good to me. I'm honestly a bit tired of updating every example over and over and fearing the wrath of pylint.
short tendon
Hopefully we'll switch to ruff soon...
midnight ember
rust or ruff?
short tendon
fixed
manic glacierBOT
wraith crow
@alpine rapids 👆
manic glacierBOT
short tendon
@midnight ember done - https://github.com/adafruit/Adafruit_CircuitPython_Requests/pull/188. I see you have a couple other PRs with new examples (like the Queue-Times one) that should switch over to with
manic glacierBOT
midnight ember
<@126221640559362048> done - https://github.com/adafruit/Adafruit_CircuitPython_...
Yep was in the process of doing that one and 2 other examples. If you want to plow through the rest of them go for it, please. That'll be less pylint anxiety for me.
short tendon
Added this to in the weeds for tomorrow, but have had this idea for a while to make wiznet5k and esp32spi socketpool more like the built-in one: https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/pull/158
Sharing so people have the opportunity to think about it
manic glacierBOT
orchid basinBOT
crimson ferry
Espressif spends >25% of revenue on R&D!
stuck elbow
I only do D&R
lone axle
<@&356864093652516868> The weekly meeting is coming up in just under 1 hour here on Discord. Feel free to fill in hug reports or status updates in the doc beforehand if you'd like: https://docs.google.com/document/d/15BPPqHPszkw6a2gGAYxj4CmFXQuhpvArvodMLLIJVJ0/edit?usp=sharing We look forward to hearding from folks.
lone axle
Would it be relatively easy to add the property os.SEEK_END to the builtin os module? It exists in CPython and returns 2. I'm guessing that it wouldn't be so hard, but I've never tried modifing os or anything that is also a CPython builtin.
Also is a changethat would be wanted in the core? it would make it so that some new code in a PR for requests would be able to match more closely between circuitpython and CPython
tulip sleet
Would it be relatively easy to add the property `os.SEEK_END` to the builtin `os...
there are three such constants, so we'd want to add them all.
but that adds to all builds
i looked at some of the seek() impls and wasn't sure they all implemented SEEK_END, but some do
<@&356864093652516868> meeting starting in two minutes
lone axle
The space was a concern I considered. It's very easy to just use the value 2 or make a SEEK_END local variable for it, and that also works in both CPython and CircuitPython. So it's probably not worth it if it has any chance to push any builds over the limit. (or even if there is more useful functionality that could use the space.)
tulip sleet
that is kind of micropython's attitude. We are more interested in helping folks, but I can't remember seek() coming up in support
slender iron
@lone axle I think I know who travis consulting would be if you need a contact
lone axle
Nice, thank you. I'll reach out if so. Thus far I only saw the listing for it, but haven't tried out any of those prior images yet.
solar whale
tulip sleet
@jerryn could you mute
tiny peak
@slender iron if you want some examples of custom Renode peripherals, xobs has done a bunch of that stuff for Precursor https://github.com/betrusted-io/xous-core/tree/main/emulation
lone axle
I'd lean toward updating and they'll have to change their code. We could bump the major version.
I'll try to go through the open PRs before the change so there aren't merge conflicts.
midnight ember
This is true. The simpletests for sensors are usually serial only.
midnight ember
For beginners they might not know the difference in how to switch between displays and display drivers... there are sooo many different displays.
This would be great for the feather tft's and reverse tft's. i think the learn guides for those boards do exist with display code.
updating label text updating is a very common question in discord
solar whale
a guide to using guides 😉
midnight ember
We need a Displayio Guide to the Galaxy, edition 1.
short tendon
For the WIZNET5K socket.py => socketpool.py, should this be done before or after Jeff's ssl: work on anything (since it will create conflicts with that PR)
gilded cradle
Thanks Dan for hosting
lone axle
Thank for hosting Dan, hope everyone has a great week!
I may try to figure out how to get a bullet list working for the rendering of this file: https://github.com/adafruit/Adafruit_CircuitPython_Bundle/blob/main/docs/drivers.rst The raw text is formatted nicely, but the github front end rendering smashes the list into a paragraph basically instead of 1 on each line.
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/13qSa787bMKb6fA1Z0utzht-5TBOAT-q1_aWa2L-mAaw/edit?usp=sharing
short tendon
@lone axle - zero rush on this. Let me know when I should plow through doing the socketpool PRs for wiznet5k and esp32spi.
- esp32spi has no open PRs so that's easy
- wiznet5k has 5 - 4 are around connection (with one being a dup) and the other is Jeff's SSL one
lone axle
@short tendon do you have an idea of how to test https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/pull/150 ? Honestly the specifics of this change (and a few others it seems) are beyond my knowledge. All I can really do is test them out on hardware and look over to the best of my abilities.
Trying to figure out how to test this one, it's a change inside of __exit__() of socket class from adafruit_wiznet5k_socket.py I'm assuming I need to be using a context manager, but I'm unsure on the proper syntax and not having much luck so far.
short tendon
<@1152454697614069811> do you have an idea of how to test https://github.com/ada...
I don't fully. You could do something like:
with connection_manager.get_socket("www.adafruit.com", 80, "http"):
time.sleep(60)
And while it's waiting disconnect the cable, which would kill the socket...
lone axle
I don't fully. You could do something like:
```
with connection_manager.get_sock...
Okay, I'll give that a go. Do you know if you ended up testing this one out? https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/pull/151 and if so anything about the results?
sleek relic
While exploring CircuitPython apps and libraries, I did a side project. I created PyAPIDiffTool (https://github.com/mMerlin/PyAPIDiffTool), a tool to compare the API of different module implementations. This could be useful to developers working on CircuitPython packages. It has extensive configuration file and command line support to adjust what it reports. It can be used to show differences, not implemented, extensions between the base implementation and the port. It can report or ignore differences in docstrings, typehint annotations that exist in the port but not in the base, When doing differences, it can be told to report exact matches or not.
I wrote it for my own curiosity, so reporting is very rough. Developers working on library code should be able to understand it. That said, I could use some feedback on things I may have forgotten to implement, or better ways to do the reporting. It is now stable enough to write good unittests, but that would be a whole other project. The repo github issues is open for bugs and feedback. Contributions welcome.
As an example, here is the output comparing logging and adafruit_logging. Configured to show minimal information about public matched attributes with differences. It can be a lot more verbose.
14 base attributes skipped, 12 port attributes skipped.
13 Matched, 42 Not Implemented, and 15 Extension attributes found.
Matched in "logging" base and "adafruit_logging" port implementations.
"LogRecord" Differences: ('logging',)¦('adafruit_logging',)
Base detail key class not equal port key namedtuple: cannot compare further
('class', Sentinel Tag: 'Other:Expand')
('namedtuple', "('name', 'levelno', 'levelname', 'msg', 'created', 'args')")
"getLogger" Differences: ('logging',)¦('adafruit_logging',)
Method Parameters:
positional parameter 0 name: base "name"; port "logger_name"
positional parameter 0 default: base "None"; port :str ""
I didn't. I can later today or tomorrow if it's helpful
lone axle
I didn't. I can later today or tomorrow if it's helpful
Any testing or feedback on any of the open wiznet (or other library) PRs is definitely helpful and appreciated. I really don't have much in depth knowledge of ethernet or networking generally. It's possible that you're closer to this code and the surrounding usages of it than I am at this point honestly. Getting more eyes looking at, testing, and reviewing PRs allows us support more contributions and changes.
tulip sleet
While exploring CircuitPython apps and libraries, I did a side project. I create...
I have bookmarked this, thanks!
slender iron
tio /tmp/cp-uart
[15:23:18.247] tio v2.8
[15:23:18.247] Press ctrl-t q to quit
[15:23:18.247] Connected
Auto-reload is off.
Running in safe mode! Not running saved code.
You are in safe mode because:
CIRCUITPY drive could not be found or created.
Press reset to exit safe mode.
Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 9.1.0-beta.0-28-g64312c3b34-dirty on 2024-04-22; Renode Cortex-M0+ with cortex-m0+
>>> print("hello renode")
hello renode
>>>
@lone axle I wonder if it's worth adding a testing readme to wiznet5k on how to test some of these super specific things
lone axle
@short tendon we're down to just this one: https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/pull/150 and your draft one. If you have the chance and are want to take a look at that one it'll be appreciated.
short tendon
<@1152454697614069811> we're down to just this one: https://github.com/adafruit/...
Yeah, I've been watching. I don't know enough on the hardware level to be able to help. I wonder if @crimson ferry can...
lone axle
I'm never really opposed to additional documentation around testing (or much of anything else tbh).
crimson ferry
@short tendon I'm not sure either, I largely treat that lib as a black box. The PR is a one-liner, but I don't know what it's doing or why.
onyx hinge
I read some datasheets & web pages and wrote a semi-educated opinion.
crimson ferry
thanks @onyx hinge, I did the first part of that but couldn't come up with a semi-educated opinion 😉
the good news I think is that this PR should be independent of other planned work for the WIZnet library
short tendon
It will probably have conflicts once I jack it all up for socketpool...
Here's a question, in similar but very different environments, I have setup servers for testing things like timeouts (and other things).
For example you can make a post with the header close_after: 10 which will automatically hold and then close after 10 seconds.
Is this something Adafruit would find worthwhile?
lone axle
The python library findimports parses python code file and finds all of the imports that it uses. Does anyone know of a similar tool that finds "other files" that the python code uses? Like if the script relies on blinka.bmp, arial12.pcf etc... it could return those as a list?
Or if such a thing doesn't exist is there any standardized way to signify required files in a comment near the top of a code file?
# Required Files: blinka.bmp, arial12.pcf
Or similar? I am thinking about this with regards to making tools that load examples onto devices. It's easy to load the example code file, but without some solution to this it won't be able to proper load other required files that the script needs.
short tendon
Or if such a thing doesn't exist is there any standardized way to signify requir...
Nothing that I know of. But I wonder if you could hack around distutils and pass in a manifest file that listed them...
short tendon
@tulip sleet or @onyx hinge, do you know if the SSL for WIZNET5K will go in 9.1?
onyx hinge
<@329766224093249548> or <@426200547024961539>, do you know if the SSL for WIZNE...
No promises until it actually gets merged, but there's nothing that forces it to wait for 10.
short tendon
No promises until it actually gets merged, but there's nothing that forces it to...
Thanks. was trying to figure out if the ConnectionManager support could be merged in, but need a way to know it's there and was going to use version. Is there anything else that is now defined that could be checked?
Could we do something like export a flag in shared-bindings/ssl/__init__.h that if existed we would know it was supported?
onyx hinge
probably not, because that would not be in standard python's ssl module (we don't add things if we can avoid it)
tulip sleet
you can check the CircuitPython version with os.uname() or sys.implementation.
sys.implementation breaks it out into a version tuple
onyx hinge
@tulip sleet the problem being, we can't promise justin what the circuitpython version requirement is
I think the PR will have to be un-merged until we know that
tulip sleet
I would say leave the code commented out or with and false in the library, and then it can be turned on in the library when the ssl PR is merged
short tendon
@tulip sleet so on this PR: https://github.com/adafruit/Adafruit_CircuitPython_ConnectionManager/pull/9/files, comment out lines 136-149, so we have it when we are ready?
tulip sleet
<@329766224093249548> so on this PR: https://github.com/adafruit/Adafruit_Circui...
yes, something like that, or keep it in a draft PR for the CM library
the latter makes it clearer, I think.
onyx hinge
if cp_version >= (9, 1, 0) I think
tulip sleet
yup, that works! Tested in CPy:
>>> (9,0,0) >= (9,1,0)
False
>>> (9,1,0) >= (9,1,0)
True
>>> (9,1,0) >= (9,1,1)
False
>>> (9,1,1) >= (9,1,0)
True
``also test (10,0,0) > (9,1,0) for good measure
tulip sleet
True
short tendon
Oh cool. I'll update it, and make the version a constant. I'll leave it as a open PR for now
tulip sleet
make it draft so it won't be merged by accident
short tendon
@tulip sleet last question on the updating the wiznet5k/esp32spi from socket to socketpool => should ConnectionManager support both (socket and socketpool), or release all 3 together and have it just use the new one?
tulip sleet
<@329766224093249548> last question on the updating the wiznet5k/esp32spi from s...
I don't understand "all 3 together". If you can do ESP32SPI separately, go ahead. That would mean splitting off the wiznet CM changes into a separate draft PR, I think?
short tendon
There would be 3 PRs:
- adafruit_wiznet5k/adafruit_wiznet5k_socket.py -> adafruit_wiznet5k/adafruit_wiznet5k_socketpool.py
- adafruit_esp32spi/adafruit_wesp32spi_socket.py -> adafruit_esp32spi/adafruit_esp32spi_socketpool.py
- having ConnectionManager use xxx.socketpool
tulip sleet
I@onyx hinge's CPy ssl pr is not useful for ESP32SPI, if i understand correctly.
so when 2 is done, can do 3 for 2, but save support for 1 as a separate PR for CM. Does that make sense?
short tendon
Sorry @tulip sleet 2 different conversations:
- about ssl on wiznet
- about making the socketpool a class, incase you have more then 1 radio
tulip sleet
ah, I see, but do you need jeff's PR before you can do 2 and 3?
short tendon
nope
tulip sleet
so is there something you want to add to CM that needs jeff's PR merged first?
crimson ferry
ssl is independent of the socketpool changes, but the SSL PR has implications for ConnectionManager
(to allow WIZnet SSL)
tulip sleet
ok, thanks, I was just connectin socketpool and ssl_context in my head
i am just saying don't be blocked on other things while awaiting the ssl pr
if we don't want two transitions of wiznet code, then wait on some of that
crimson ferry
I think the SSL PR is close (though understandably not highest priority), clients are robust and there's a lot of benefit to enabling that, any server fix could be merged later?
tulip sleet
@slender iron nothing urgent: https://github.com/espressif/esp-idf/issues/4999#issuecomment-2067963620 is an Espressif I2C issue I have been watching for some time. Finally fixed, still awaiting backport to 5.2.x.
tulip sleet
looks like I already did 🙂
yes 🙂 we can upgrade to 5.2.latest shortly before 9.1.0 final
slender iron
we're on the latest release
granite pier
I’m playing around with the open PR to add HTTPS support to httpserver, and I’m trying to understand all the pieces at play. I know CircuitPython uses Mbed TLS, but I’d love to figure out if Mbed uses the hardware accelerated cryptography functions available in at least some ESP32 chips. Is that something someone happens to know?
slender iron
I’m playing around with the open PR to add HTTPS support to httpserver, and I’m ...
I assume the ESP-IDF handles that
@onyx hinge has done the most recent mbedtls work
onyx hinge
We use the esp-idf version of mbedtls on that platform. So if they have hardware accel we should be getting it.
granite pier
heck yeah, thanks! gonna make this server go vroom vroom
short tendon
@tulip sleet - question on wiznet5k.WSGIServer. Currently it uses the fake socket(pool) globally with a call to set_interface. Should we keep it the same, and have that method create a global socketpool, or change it so you pass one in like you do for adafruit_httpserver?
short tendon
@crimson ferry when you have a moment, using main for wiznet5k, can you try the wiznet5k_wsgiserver example and see if hitting /btc works for you? I've tried main and my socketpool branch and it doesn't work (https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/blob/main/examples/wiznet5k_wsgiserver.py#L117)
I get:
Wiznet5k Web Server Test
Open this IP in your browser: 192.168.77.23
Root WSGI handler
Large pattern handler
Static file code.py handler
BTC handler
Traceback (most recent call last):
File "<stdin>", line 137, in <module>
File "adafruit_wiznet5k/adafruit_wiznet5k_wsgiserver.py", line 118, in update_poll
File "adafruit_wsgi/wsgi_app.py", line 70, in __call__
File "<stdin>", line 119, in btc
File "adafruit_requests.py", line 687, in get
File "adafruit_requests.py", line 619, in request
File "/lib/adafruit_connection_manager.py", line 249, in get_socket
File "adafruit_wiznet5k/adafruit_wiznet5k_socket.py", line 187, in getaddrinfo
File "adafruit_wiznet5k/adafruit_wiznet5k_socket.py", line 202, in gethostbyname
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 308, in get_host_by_name
File "adafruit_wiznet5k/adafruit_wiznet5k_dns.py", line 275, in gethostbyname
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 852, in read_udp
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 810, in socket_read
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 1135, in _get_rx_rcv_size
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 1174, in _read_snrx_rsr
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 1121, in _read_two_byte_sock_reg
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 1349, in _read_socket_register
File "adafruit_wiznet5k/adafruit_wiznet5k.py", line 1093, in _read
RuntimeError: pystack exhausted
@short tendon try increasing pystack in settings.toml?
short tendon
<@1152454697614069811> try increasing pystack in settings.toml?
You saved the day again!
crimson ferry
I don't know the wsgi server, but it's not surprising that server calling requests could be some nested calls
short tendon
yes, quite a few... 😉
crimson ferry
meanwhile... I can't connect to wifi with ESP32SPI o_O (haven't used it in a wwhile)
short tendon
That's strange
midnight ember
Airlifts prior to a certain nina version must be updated with the new TLS stuff. They won't work right until updated.
short tendon
meanwhile... I can't connect to wifi with ESP32SPI o_O (haven't used it in a wwh...
What error are you getting?
crimson ferry
"No such ssid", but if I increase the timeout then it's "Failed to connect to ssid" ...trying different APs / encryptions / etc ...I'll figure it out, it's probably something dumb (RSSI is -46)
crimson ferry
Airlifts prior to a certain nina version must be updated with the new TLS stuff....
shouldn't affect AP connection, only certain certificates used to connect TLS
afaik, Airlift boards still ship with 1.2.2, which generally still works fine
midnight ember
yes i just woke up, still drinking my coffee, ignore me.
thorny jay
Hello, I could not find the meeting recording on Spotify (no worries, I have it on YT: https://www.youtube.com/watch?v=TmFugF2c4mk ), but maybe something went wrong.
Also, a message to @tulip sleet , your audio level is always much lower than the other, this is especially visible (and problematic) when you are the host, because you talk more. This is not new, it's been like that since a bit of time. Not sure if it is like that LIVE or only in your local recording... but I think it is generally low.
midnight ember
I had to rewatch it this week since I missed most of the meeting. Yes Dan's mic volume was super low.
Going from Dan to FoamyGuy almost blew out my eardrums because I had to crank the volume to full for danh and then 1/4 volume for foamyguy. Dan's mic is usually fine but was very low for some reason this week.
didn't know it gets syndicated on spotify, neat. only thought it was on youtube.
can't fix the video just something for dan to fix his volume before next weeks meeting.
short tendon
"No such ssid", but if I increase the timeout then it's "Failed to connect to ss...
I usually get this the first time and need to loop. Usually takes 2 tries...
crimson ferry
yeah I have it looping and debug on
short tendon
😦
thorny jay
can't fix the video just something for dan to fix his volume before next weeks m...
I think the solution is a setting or setup on Dan side for the future, the past is the past. 😉
I don't know when (what week) it started, but it is not the first time, I have noticed that "many" times (maybe I should have complained earlier).
onyx hinge
Hello, I could not find the meeting recording on Spotify (no worries, I have it ...
I don't know the details of how the podcast feed is produced, except that someone at adafruit has a script that does it. The canonical URL for the podcast feed is https://adafruit-podcasts.s3.amazonaws.com/circuitpython_weekly_meeting/audio-podcast.xml and it has not been updated this week. It is created/updated by some script from the youtube playlist https://www.youtube.com/playlist?list=PLjF7R1fz_OOUvw7tMv45xjWp0ht8yNgg0 -- the latest meeting is on that playlist so that's not where the problem lies.
If the podcast version of the playlist misses an update next week I'll raise it internally with the script maintainer; you can remind me if so
short tendon
If anyone has time for at least a high level look at the wiznet5k socketpool PR, I would appreciate it: https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/pull/159
midnight ember
I think the solution is a setting or setup on Dan side for the future, the past ...
I have noticed as well but this week he was particularly low and even at full volume I had trouble hearing him.
crimson ferry
@short tendon I should be able to look at the PR after dinner
short tendon
<@1152454697614069811> I should be able to look at the PR after dinner
Wonderful. You can also wait until tomorrow, or the next day or whenever...
Although I would love to get this moving, I don't want to put people out
tulip sleet
@midnight ember @thorny jay I didn't change anything on my side, so I'll try a different mic next time, one that's closer to me. I storted of looked ok in the obs level meters.
midnight ember
<@126221640559362048> <@681528999520108564> I didn't change anything on my side,...
I recommend a cloud flight headset they're wonderful... and wireless with great range and connectivity. I also happen to have a youtube video on replacing the battery with one that lasts twice as long as the stock battery.
I think Jepler also has one.
tulip sleet
I dislike wearing headphones in general because I can't hear myself. Does it have sidetone
Right now I am just using the mic on my Logitech webcam. I have a Zoom H2n I could hook up to be closer
midnight ember
I don't know what a sidetone is. Can you explain?
tulip sleet
that's when you hear a bit of yourself when you're talking. Old-fashioned desk phone do that
midnight ember
It's a unidirectional mic but has a wide view so it's almost omnidirectional.
ohh heck no, can only hear others, cannot hear yourself that would be too distracting.
you can set it up to hear yourself if you want, that's part of audio monitoring that singers need sometimes and is an OS level thing.
tulip sleet
i mean when you put on well-sealed phones you can't hear yourself talk very well, as opposed to when you don't wear headphones. And the sidetone should have no latency, that would be terrible.
it was designed into desk phones from the Bell system
midnight ember
oh that, they're circumaural meaning they go around your ears. honestly they work a bit too well as people can sneak up on me and i can't hear them coming.
well they're not like some military grade ear cups that are completely sound deadening but they're definitely better than most in that dept.
tulip sleet
I don't want that. I want to hear other things. My favorite headphones with original Sennheiser HD414's
I have not found a suitable substitute
midnight ember
ah then they might not be a good fit as they are sound deadening. not like bose electronic sound deadening, those things are amazing.
midnight ember
i know the sennheisers you're talking about they were great. the wireless capability of the cloud flight is better than any of the logitech g-series. i went through 4 sets of those they break easily.
tulip sleet
incredibly comfortable, could wear them for 8 hours while doing homework
i have not found anything else as close to comfortable
midnight ember
those are binaural type where the cushion rests against the ears instead of around it. i thought you were talking about a different type.
crimson ferry
my brother had a pair like that, super light and comfortable
midnight ember
thought you were talking more about this style. my dad had something like these they were amazing.
and i just realized their not a headset just headphones. :/
tulip sleet
<@329766224093249548> - question on `wiznet5k.WSGIServer`. Currently it uses the...
I don't really know enough to have an opinion on that. I have been speaking from a high level about trying to make the API's as similar as possible so that code could be shared.
HD414
midnight ember
those remind me of sony walkman kind of style. some of the earlier headsets in the 90's looked like that.
tulip sleet
the foam would rot after a few years and have to be replaced
no ear pain whatsover, and no sweaty enclosed ear box
midnight ember
same is still true today but you can find replacement earcups. same thing different decade.
tulip sleet
yep the cups are still available, and people put them on Grado's
midnight ember
one of the bad things about those is if you get up you yank the cord out and your neck down. one of the advantages of wireless. no more running over the cord with your office chair too.
we need to get you updated to 2020's standards. cloud flight. you think about that and i'll think about running linux someday. 🙂
tulip sleet
🙂 ok, I will check them out at microcenter or something
short tendon
I don't really know enough to have an opinion on that. I have been speaking from...
Sounds good. I passed it in, it felt like the right thing
manic glacierBOT
thorny jay
I dislike wearing headphones in general because I can't hear myself. Does it hav...
I recently tried "bone induction" headset... supposedly for runner, but also great if you want to hear the sound arround you and not have something in or on your ear. So you do hear yourself (and everything else).
But the "problem" is more on the microphone side, I would say.
stuck elbow
why would you want to hear the sound around you, though
fringe yarrow
To avoid bear attacks
manic glacierBOT
onyx hinge
<@329766224093249548> - question on `wiznet5k.WSGIServer`. Currently it uses the...
is there much reason not to use a non-wiznet-specific httpserver now? maybe just deprecate it rather than try to fix it up much.
onyx hinge
Hello, I could not find the meeting recording on Spotify (no worries, I have it ...
The official RSS feed for the podcast version of the weekly meeting has now updated and includes monday's meeting. spotify will likely follow in updating soon if it hasn't already
crimson ferry
WSGI was removed from ESP32SPI last September. adafruit_httpserver is much more feature-rich and well-supported.
short tendon
is there much reason not to use a non-wiznet-specific httpserver now? maybe just...
It only needed a few changes...
It's all working in the PR, and was a good way to make sure I didn't break anything
tulip sleet
@short tendon i agree with jepler and anecdata about removing wsgiserver from wiznet if httpserver can now be supported with it
I made the decision to remove wsgiserver
short tendon
<@1152454697614069811> i agree with jepler and anecdata about removing wsgiserve...
Do you want me to remove it in my socketpool PR?
empty salmon
I would like to submit a PR to have the Adafruit_CircuitPython_Bitmap_Font added to the available set of frozen libraries.
I have not done a PR before. Do I make the change in the main branch of my fork or do I keep the change in a separate branch and make the PR from that?
tidal kiln
separate branch is best. typically a new branch made specifically for the changes related to the PR.
empty salmon
Thanks.
tulip sleet
Thanks.
See https://learn.adafruit.com/contribute-to-circuitpython-with-git-and-github/always-work-on-a-branch and see the whole guide as well.
you'll be adding that library as a new submodule.
empty salmon
Thanks @tulip sleet. I did it locally so I will make a clean environment and do it according to instructions in a new branch of a fork.
slender iron
why would you want to hear the sound around you, though
I like open air headphones so I can hear cars when I'm walking around
tulip sleet
me too, and in the house, i want to hear when someone's calling me, the phone rings, etc.
fringe yarrow
also if there's bears in the house..
stuck elbow
I guess I'm spoiled, I don't walk where there are cars or bears, and I never pick up the phone. But crying children and loudly talking people seem to be everywhere.
tulip sleet
@slender iron someone is #help-with-circuitpython is not getting ble workflow to work. I'm not at all sure what's wrong
slender iron
k, let me look
orchid basinBOT
fringe yarrow
Looks like recent changes have broke the CI flow for the website PRs, as well..
tulip sleet
Looks like recent changes have broke the CI flow for the website PRs, as well..
could you point to that?
fringe yarrow
empty salmon
there appears to be an error in the documentation for adding a frozen library to the circuitpython build. The web page instructions gives the command make update-frozen-modules but the Makefile has the target as update-frozen-libraries.
tulip sleet
there appears to be an error in the documentation for adding a frozen library to...
Fixed!
empty salmon
thanks. I was not sure where to make the request
tulip sleet
thank you
empty salmon
I am still learning all of this stuff
tulip sleet
you can use the "Feedback? Corrections?" link in the left sidebar of the guide
empty salmon
ah. thanks for the pointer
manic glacierBOT
tulip sleet
ah. thanks for the pointer
so are you going to add it as a frozen to an existing board?
empty salmon
I am adding it as a frozen to a new board. I thought I should do a PR for the frozen library before I submitted the PR for the new board
empty salmon
like I said, I am new to this process and wanted to do it is baby steps
tulip sleet
no problem 🙂
empty salmon
kind of fun having multiple windows open and watching this whole process hapenning in realtime
tulip sleet
github is so good at dynamic webpage updating
short tendon
@tulip sleet can I ask a really big favor? I opened up this PR https://github.com/adafruit/Adafruit_CircuitPython_ConnectionManager/pull/12, which is 100% test updates, so I don't need to deal with conflicts later. If you have time, can you look and merge?
crimson ferry
@short tendon I'll look at the ESP32SPI PR tomorrow, I have a config wired up with 2 airlifts, it's working with two copies of the libraries, so should be straightforward to convert the code to use the new socketpool (like WIZnet).
short tendon
<@1152454697614069811> I'll look at the ESP32SPI PR tomorrow, I have a config wi...
Oh that's awesome!
tulip sleet
<@329766224093249548> can I ask a really big favor? I opened up this PR https://...
didn't see this until late
merged
short tendon
Thanks!!
manic glacierBOT
manic glacierBOT
short tendon
Good morning! A few simple PRs in connection libraries:
- https://github.com/adafruit/Adafruit_CircuitPython_ConnectionManager/pull/8 : Different pool, different ConnectionManager
- https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/pull/160 : Remove wsgiserver
- https://github.com/adafruit/Adafruit_CircuitPython_Requests/pull/188 : Use with in examples, remove close in json()
crimson ferry
Good morning! A few simple PRs in connection libraries:
1. https://github.com/ad...
- looks good testing-wise, maybe someone could code review
crimson ferry
@tulip sleet sorry to cross streams with my edit. I noticed it had come out of Draft, so thought maybe we'd be considering to merge. jepler knows best.
tulip sleet
np - I have been relying on you folks to test
manic glacierBOT
onyx hinge
@short tendon ^^^ so yes, (9,1,0) will be the version check for core ssl module supporting wiznet
short tendon
<@1152454697614069811> ^^^ so yes, (9,1,0) will be the version check for core ss...
Awesome! I'll take that PR out of draft
https://github.com/adafruit/Adafruit_CircuitPython_ConnectionManager/pull/9 - Add support for WIZNET5K ssl in ConnectionManager is also ready
tiny peak
@slender iron I made some fpga progress this past week. Spent a long time researching pin constraint files. Reasonably confident that I have one now which is pretty comprehensive and will hopefully let me put the DRAM chip in low power mode without damaging it (it runs at 1.35V rather than 3.3V). Also managed to get a verilog-from-scratch serial loopback bitstream working, including an RX activity LED. Haven't finished writing that up yet though.
slender iron
<@252717193496756235> I made some fpga progress this past week. Spent a long tim...
Nice work!
manic glacierBOT
tulip sleet
@short tendon should https://github.com/adafruit/Adafruit_CircuitPython_ConnectionManager/pull/8 published as a release be delayed until the other PR's are ready, so they are all merged and released at once?
short tendon
<@1152454697614069811> should <https://github.com/adafruit/Adafruit_CircuitPytho...
That's up to you all. I will have 2 more PRs:
- wiznet5k + esp32spi socket => socketpool (in draft)
- update to close out all connections on a pool (in progress)
I think since the wiznet ssl one is merged into main, having ConnectionManager support those seems a good idea. So maybe a minor bump now, and a major with the socket pool changes?
tulip sleet
I just don't know if publishing the new CM PR will break anything or not, or it's safe to do so now.
that's what I'm asking
short tendon
Ahhh, it shouldn't break anything
tulip sleet
ok, i'll release!
short tendon
🚀
tulip sleet
Socket programming q: I am implementing a user-specified timeout for socket connect(). lwip doesn't support this natively, so i use the socket as non-blocking (which is true internally for all sockets in CPy), and just keep checking select() to see if the connection occurred during the timeout. This is fine if the connection happens, but suppose it does not. What should I do to cancel the connection attempt? Should I close the socket?
I've done quite a lot of websearching on this but haven't found an answer. MicroPython does the above to get a user-specified timeout, but doesn't bother to do any cleanup if the timeout happens. Not sure that is the right thing.
answering my own q, esp-idf closes the socket
onyx hinge
standard Python does not close the socket after a timed-out connect. Instead, if the connection is not made or definitely failed before the timeout is up (including if you had a timeout of 0, e.g., non-blocking connect) it's in a state where you can check if it's connected in various ways including getpeername() (raises "Transport endpoint is not connected" if not connected)...
tulip sleet
Thanks, I can do that. It was a struggle to find any discussion of this. I did look at the Python impl and it didn't do anything.
onyx hinge
https://stackoverflow.com/questions/17769964/linux-sockets-non-blocking-connect#17770524 https://stackoverflow.com/questions/36092580/how-to-connect-with-a-non-blocking-socket
tulip sleet
Thanks - I did see both of these, but I don't see that they say explicitly that even if the timeout happens that the connection might eventually happen.
The second has a long discussion about how to check for connection, and I coded based on that and the MPy code
onyx hinge
more on standard python in linux:
import select
import os
import socket
s = socket.socket()
s.settimeout(0.001) # unlikely to succeed in 1ms
err = s.connect_ex(("8.8.8.8", 443))
print(f"connect_ex -> {os.strerror(err)} ({err})")
try:
print(s.getpeername())
except Exception as e:
print(f"{e} trying to get peer name")
_, w, _ = select.select((), (s,), ()) # note: no timeout on select
print("After select...")
print(s.getpeername())
[Errno 107] Transport endpoint is not connected trying to get peer name
After select...
('8.8.8.8', 443)
If timeout is 0 instead of very small, I get connect_ex -> Operation now in progress (115) instead.
In either case (fully non-blocking, or timeout-before-connect) the socket stays in a case where the connection can complete, and when it does the socket will be treated as writable by select/poll.
if I try to connect to a non-routable IP the select() "never" completes (or, well, however long the TCP handshake timeout is in the linux network stack)
for "connection refused", ```After select...
Traceback (most recent call last):
File "/home/jepler/socky.py", line 18, in <module>
print(s.getpeername())
^^^^^^^^^^^^^^^
OSError: [Errno 107] Transport endpoint is not connected
I was thinking of doing a getpeername() check in the code I was writing but realized it's really the responsibility of the user code.
short tendon
@tulip sleet when we think about making code and examples simpler, I wanted to share a util I have:
def connect(radio, env_prefix="CIRCUITPY_", retries=3):
It works similar to the ConnectionManager in determining the radio, and thus how to call connect, and then gets the connection vars:
wifi_ssid = os.getenv (f"{env_prefix}WIFI_SSID")
And then tries to connect.
So it takes all of this:
import os
...
wifi_ssid = os.getenv("CIRCUITPY_WIFI_SSID")
wifi_password = os.getenv("CIRCUITPY_WIFI_PASSWORD")
while not radio.connected:
radio.connect(wifi_ssid, wifi_password)
And turns it into:
import connection_helpers
...
connection_helpers.connect(radio)
Does this seem worth adding to ConectionManager?
midnight ember
https://github.com/adafruit/Adafruit_CircuitPython_ATECC/pull/35 I was asked to review a PR and discovered a scenario where a certificate signing request can be allowed to input 1 character into all CSR fields and it will still pass. I wrote up a section for in the weeds in this weeks meeting docs. Can someone take a look and see if my interpretation of CSR fields being required and amending the docstring is correct? It can wait until in the weeds but would be nice to have some input on it for a decision and not have an in the weeds topic.
# 2-letter country code
MY_COUNTRY = "US"
# State or Province Name
MY_STATE = "New York"
# City Name
MY_CITY = "New York"
# Organization Name
MY_ORG = "Adafruit"
# Organizational Unit Name
MY_SECTION = "Crypto"
``` intent, this will pass as an issuer of a CSR# 2-letter country code
MY_COUNTRY = "1"
# State or Province Name
MY_STATE = "1"
# City Name
MY_CITY = "1"
# Organization Name
MY_ORG = "1"
# Organizational Unit Name
MY_SECTION = "1"
``` but so will this# 2-letter country code
MY_COUNTRY = "US"
# State or Province Name
MY_STATE = ""
# City Name
MY_CITY = ""
# Organization Name
MY_ORG = ""
# Organizational Unit Name
MY_SECTION = ""
``` and so will this which I think is a problem.So it's definitely better, as previously the if/else was just around the value of common
I haven't looked in a while, you don't need to have all of them, but not remembering what's actually required
midnight ember
crimson ferry
@short tendon adafruit_connection_manager.get_connection_manager(pool) isn't typically needed by user code, is it?
(I'm getting unexpected behavior trying multiple ESP32SPI with CM#11 + ESP32SPI#198)
only if you wan to directly access the CM for a pool to do something deeper with its sockets?
midnight ember
but it's only checking that all fields have some kind of data in them. The check for 0 length doesn't quite match up with the actual minimum count of 55.
total length should at least be 55. checking against 0 only checks that at least 1 field isn't empty but the script won't allow that anyway since they are required fields... so 0 does nothing.
crimson ferry
@short tendon oh, maybe the issue is that CM #8 and CM #11 are, for now, independent... I need to make a merged copy
(I'm always getting the same ConnectionManager back, for multiple pools of the same type)
but I didn't have this isue with WIZnet
short tendon
It was around all of them <https://github.com/adafruit/Adafruit_CircuitPython_AT...
All the ifs are separate
short tendon
<@1152454697614069811> `adafruit_connection_manager.get_connection_manager(pool)...
It isn't
midnight ember
yes but each adds 11 at a minimum to the total count it's checking against.
so if 5 fields are required then the minimum should be 55 to check against not 0.
short tendon
<@1152454697614069811> oh, maybe the issue is that CM #8 and CM #11 are, for now...
Yeah, I need to get 11 updated with main (now that 8 is merged)
crimson ferry
ok, that makes sense, but now I have to figure out what I did with WIZnet
that seemed to work
and it shouldn't have
short tendon
yes but each adds 11 at a minimum to the total count it's checking against.
if common:
tot_len += 11 + len(common)
else:
raise TypeError("Provided length must be > 0")
Will raise if common is falsy
midnight ember
sorry i meant the PR proposed change ```py
def issuer_or_subject_length(
country: str, state_prov: str, city: str, org: str, org_unit: str, common: str
) -> int:
"""
Returns total length of provided certificate information.
:param str country: Country of certificate
:param str state_prov: State/province of certificate
:param str city: City of certificate
:param str org: Organization of certificate
:param str org_unit: Organization unit of certificate
:param str common: Common data of certificate
:raises: TypeError if return value is 0
:return: Total length of provided certificate information.
"""
tot_len = 0
if country:
tot_len += 11 + len(country)
if state_prov:
tot_len += 11 + len(state_prov)
if city:
tot_len += 11 + len(city)
if org:
tot_len += 11 + len(org)
if org_unit:
tot_len += 11 + len(org_unit)
if common:
tot_len += 11 + len(common)
if tot_len == 0:
raise TypeError("Provided length must be > 0")
return tot_len
if country, state, city, org, and common are all required fields then total_len minimum to check should be 55 not 0?
in this case org_unit is also a required field so it should be 66. but i propose to make org_unit optional.
Seems fairly cut and dry to me. Checking for tot_len == 0 seems rather pointless when it should be 66 for the check to work correctly?
Fields can be blank with "" and still pass. I don't think that's how a CSR should work?
short tendon
So, I would specifically check that the required fields are valid first:
if not required_1 or not required_2:
raise TypeError("Required fields missing")
And if that passes calculate the length
midnight ember
Thank you that's kind of what I was thinking. I wrote some feedback with a script that does that. It's my first PR review so I just wanted to double check that I'm not completely off base with what is really needed to fix that check.
The +11 prepended to the length ensures that it allocates the byte space needed for the crypto chip hash prior to each field.
short tendon
@crimson ferry CM/pull/11 is updated with main. So it should work if you pull it down
crimson ferry
@short tendon w00t! (I remembered that for the multi-WIZ test, I used a pre-CM version of Requests) ...now time to wire up the multi-WIZ / multi-ESP/ +native wifi monstrosity :p
short tendon
<@1152454697614069811> w00t! (I remembered that for the multi-WIZ test, I used a...
Awesome! What PRs is that running on?
short tendon
I'll do esp too
Wonderful! What's the MCU? I'm guessing adding many more will peg what it can do....
I've been using a UMS3
crimson ferry
UM FeatherS3... yeah we'll see about esp-idf memory
crimson ferry
actually, not sure what of all this besides native wifi wouldn't come from PSRAM, so there may be plenty
short tendon
I haven't actually ran tests on a UMS3 for how many sockets I can open when not using the WiFi is (when I am, it's 7)
Because I was like, why would I?
short tendon
<@329766224093249548> when we think about making code and examples simpler, I wa...
@crimson ferry would love your thoughts on this helper too
crimson ferry
#sockets, espressif & raspberrypi: https://github.com/adafruit/circuitpython/issues/7543#issuecomment-1509307970
esp32spi has potentially 10 total iirc
wiznet... depends on which model
crimson ferry
<@360465312518504448> would love your thoughts on this helper too
I need to think about this, a wifi helper could be very useful, just not sure whether it shouldn't be separate from CM
CM is going to be almost-mandatory now with all examples using it... it will be harder to not use it, so may want to keep it focused
short tendon
I need to think about this, a wifi helper could be very useful, just not sure wh...
And that's my counter argument. For the generic user, they will be using CM. So why not throw in a few more helpers to make it super simple...
crimson ferry
but then does it get into APs, scanning, ping, and other wifi features? Pico W is already feeling memory-constrained (see https://github.com/adafruit/circuitpython/issues/9085)
short tendon
And then when we add the ESP64WIZBANG6000, the examples don't change 😉
short tendon
but then does it get into APs, scanning, ping, and other wifi features? Pico W i...
No. Only connection based things
And memory usage should stay small, since it doesn't import anything that wouldn't already be used
I would also wonder if for the Pico W, if we could freeze it, to take up less space...
crimson ferry
does freeezing use less RAM than .mpy on disk?
short tendon
Yes, it's part of the space used by the build
Which is why it's done for some other devices
short tendon
Although, would need someone with a fuller understanding to confirm
short tendon
Pico W is also somewhat flash-constrained
Maybe I'll freeze it and see if it fails...
crimson ferry
@short tendon the wiznet run above, I had it set up wrong... I was creating the pools from my own import and appending them to a list. I'm not geting wiznet or esp32spi working across multiples with new #11 ...I'm not fully grok'ing the library... there's one pool per CM... is there one CM per radio type, or per radio?
short tendon
<@1152454697614069811> the wiznet run above, I had it set up wrong... I was crea...
Should be per pool
It's possible something isn't right. Sadly I only have 1 of each...
crimson ferry
or is it intended that I track my own pool per radio?
short tendon
Thinking....
It may be for this case (using the same radio code), you need to create the pool manually
crimson ferry
ok, that's fine
short tendon
CM doesn't key off the radio (it can't)
So multiple of the same radios would need some advanced code
But you should be able to have just the main py file (not needing to duplicate)
crimson ferry
yeah, I basically do:```
radios.append(ESP_SPIcontrol(spi, esp32_cs, esp32_ready, esp32_reset))
# pools.append(adafruit_connection_manager.get_radio_socketpool(radios[i]))
pools.append(socketpool.SocketPool(radio[i]))
ssl_contexts.append(adafruit_connection_manager.get_radio_ssl_context(radios[i]))
requestss.append(adafruit_requests.Session(pools[i], ssl_contexts[i]))
where is ConnectionManager class actually instantiated?
oh, must be in the requests etc libraries
short tendon
oh, must be in the requests etc libraries
Yeah, let me write something to do it differently...
Wait, this should work. As long as the radio is different, you should get a different pool.
Can you share the whole loop when you have a moment?
crimson ferry
@short tendon see the commented lines, 48-49: https://gist.github.com/anecdata/8e8db7d325df23f438e039b0ec8f615b If I get the socketpools from CM, I always get the same ConnectionManager and only the first radio gets used. If I allocate the socketpools myself and feed those to Requests, they each get a unique ConnectionManager and both radios get used.
short tendon
<@1152454697614069811> see the commented lines, 48-49: <https://gist.github.com/...
Okay, let me do some troubleshooting in the morning. Since I don't have 2 of something, I'll need to fake it a bit. But I'm guessing something isn't right.
If you have time before I look, I would see if:
radios[0] != radios[1]
And
pools[0] != pools[1]
If either of those is false it will tell me where to look
crimson ferry
yeah, I'll check and post, I'm about to wrap up for the eve too
pools.append(adafruit_connection_manager.get_radio_socketpool(radios[airlift]))
#pools.append(socketpool.SocketPool(radios[airlift]))
⋮
True radios[0] != radios[1]
False pools[0] != pools[1]
#pools.append(adafruit_connection_manager.get_radio_socketpool(radios[airlift]))
pools.append(socketpool.SocketPool(radios[airlift]))
⋮
True radios[0] != radios[1]
True pools[0] != pools[1]
also in the first case (CM), connection_managers[0] != connection_managers[1] is False
two devices of the same class need separate pools, but if that needs to be manual, that's ok, it's not a common use case ...moving discussion to the PR
fringe yarrow
@gilded cradle Is there something I'm supposed to be doing with that circuitpython-org PR of mine that keeps failing? It doesn't look like the failures are on my end, but rather the CI flow.
gilded cradle
@fringe yarrow as far as I can tell, it doesn't appear to be on your end either. Every so often jekyll or ruby needs some weird update and I think this is one of those cases.
I re-ran the job just in case it was only a hiccup in the system or something.
manic glacierBOT
short tendon
@tulip sleet would love thoughts on this PR: https://github.com/adafruit/Adafruit_CircuitPython_ConnectionManager/pull/13. It's the last CM one I want to get in before opening up the SocketPool ones. It makes no breaking changes, just adds some small functionality
tulip sleet
this seems fine. I was waiting for the testers to finish.
you were awaiting feedback from them
crimson ferry
fine to me too, I'll defer to Retired Wizard to test since it their use case
short tendon
Sounds good!
manic glacierBOT
slender iron
@tulip sleet any chance you want to switch all of the circuitpython code over to #pragma once?
tulip sleet
<@329766224093249548> any chance you want to switch all of the circuitpython cod...
I do want to do that. Also MicroPython dropped STATIC in favor of static. And we talked about changing the header on non-MicroPython-derived sources, since those sources don't come from MicroPython files. I think Damien was in favor of that.
slender iron
I do want to do that. Also MicroPython dropped `STATIC` in favor of `static`. An...
I'm getting tired of doing pragma once piecemeal
tulip sleet
i can do this sometime soon. I can do it when you're away. Also maybe there's an upstream merge soon?
tulip sleet
it will arrive when it arrives. But these other things don't need to wait
slender iron
it may give you something a little larger to work on too
manic glacierBOT
short tendon
@tulip sleet I was looking for what else I needed to change for the SocketPool update and
found this: https://github.com/adafruit/Adafruit_CircuitPython_WSGI/blob/main/adafruit_wsgi/esp32spi_wsgiserver.py
and we removed https://github.com/adafruit/Adafruit_CircuitPython_Wiznet5k/blob/5.2.1/adafruit_wiznet5k/adafruit_wiznet5k_wsgiserver.py
In Adafruit_CircuitPython_WSGI Should I
update: WSGIServer.__init__(self, port=80, debug=False, application=None)
to WSGIServer.__init__(self, socket_pool, port=80, debug=False, application=None)
so it can be generic? This is what I did first in wiznet5k before we removed it...
tulip sleet
<@329766224093249548> I was looking for what else I needed to change for the `So...
I don't know anyone who uses this, but sure, it can be a major version revision. Or we could deprecate this server in favor of HTTPServer? But I don't know if this has features that HTTPServer doesn't.
short tendon
I don't know anyone who uses this, but sure, it can be a major version revision....
Totally your call. I'm 100% okay with updating anything that needs it, but don't want to create reviews, releases, etc for something nobody uses
HTTPServer is updated a lot more
tulip sleet
I'd just leave this alone, then.
short tendon
👍
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
midnight ember
I finally sat down and played with the 800x480 7" TFT from Adafruit (purchased last year). It seems the RA8875 driver module was missed with displayio integration. Is this type of display not capable of displayio? Seems like a fairly normal SPI display though it is huge at 800x480. Found some posts dating back to Adafruit carrying them since 2014 so it pre-dates most of displayio (about 10 years ago) to the point people were using them Arduino Uno and Due. Seems most the driver is a port from Arduino around the time of the M4 Feather. Fast forward a couple of years and now the S3 drives it well enough to consider adding displayio. https://github.com/adafruit/Adafruit_CircuitPython_RA8875
The problem with the current RA8875 driver is it uses all native RAIO (chip manufacturer) registers to display images and text which is very reminiscent of Arduino's GFX library from a decade ago. It can do layers but nothing like displayio can. Would love to see this driver updated with displayio. Unsure if it's possible or how to start. Any advice appreciated.
Here is the 800x480 next to a 480x320 TFT Featherwing.
The 24-bit background image took about 3 minutes to draw to the screen line by line.
Without displayio this is about as useful as it gets. Drawing text is equivalent to Arduino GFX with a background color behind the text. There are native methods to draw text but not to remove it for layer updates, displayed text eventually turns into a jumbled mess.
short tendon
@lone axle that was fun! https://github.com/FoamyGuy/Adafruit_CircuitPython_Requests/pull/3
tulip sleet
@short tendon After the review I did this morning, I started looking at the ConnectionManager code in more detail (which I had not really done before -- I was taking a more managerial approach). I see some refactoring that could reduce the code size slightly, but I also wanted to ask you some questions. Are you up for a brief audio chat?
short tendon
<@1152454697614069811> After the review I did this morning, I started looking at...
I am
tulip sleet
can meet you in Amelia Earhart
short tendon
@tulip sleet for when you are around, this looks a lot better. Thoughts?
result = self._get_connected_socket(
addr_info, host, port, timeout, is_ssl, ssl_context
)
if isinstance(result, Exception):
# Got an error, if there are any available sockets, free them and try again
if self.available_socket_count:
self._free_sockets()
result = self._get_connected_socket(
addr_info, host, port, timeout, is_ssl, ssl_context
)
if isinstance(result, Exception):
raise RuntimeError(f"Error connecting socket: {result}") from result
Yes! I think that makes it clear.
short tendon
Awesome. and it will make it easier to handle specific exceptions if wanted
tulip sleet
Awesome. and it will make it easier to handle specific exceptions if wanted
This is tested twice: if isinstance(result, Exception):. is that what you meant? or is there some code elided here?
short tendon
yes, because if there are available sockets, we try again and have a new result
Code is pushed up for when you have time to look/comment on it
tulip sleet
oh, I missed the second assignment to result 🙂
arctic cliff
Has anyone already tried to build CircuitPython under Ubuntu 24.04?
I always get the following error message:
xyz@MARTIN-BARNER:~/circuitpython$ pip3 install --upgrade -r requirements-dev.txt
error: externally-managed-environment
Under Ubuntu 22.04 I did not need a virtual Python environment.
THX
tulip sleet
Has anyone already tried to build CircuitPython under Ubuntu 24.04?
I always ge...
Yes, this is going to be true, due to PEP 668. This already came up for people on Raspberry Pi OS and Debiain Bookworm. Go ahead and use a venv. I am going to install 24.04 relatively soon and will write something up in the build learn guide about this.
midnight ember
@short tendon sorry I didn't get to help test your CM changes. Looks like you got the feedback you needed though.
short tendon
<@1152454697614069811> sorry I didn't get to help test your CM changes. Looks l...
Thanks and no worries. If you want, check out the helpers example. I updated it with some info when using with vs not
midnight ember
I looked at a lot of them. Great work. Thank you for taking that on. It's nice not being the only one submitting stuff to those examples anymore. Seems like CM is also moving in a good direction again. Looking forward to seeing it all come together.
arctic cliff
Yes, this is going to be true, due to PEP 668. This already came up for people o...
Thank you!
tulip sleet
Thank you!
There is a writeup here for RPI OS, which is applicable: https://www.raspberrypi.com/documentation/computers/os.html#python-on-raspberry-pi
PImoroni has worked on an "easy as possible" approach: https://github.com/pimoroni/boilerplate-python/pull/13
virtualenvwrapper looks interesting: https://github.com/raspberrypi/bookworm-feedback/issues/100#issuecomment-1772164305
And we have a whole guide here, which is more about RPI OS than CircuitPython build environments: https://learn.adafruit.com/python-virtual-environment-usage-on-raspberry-pi
arctic cliff
And we have a whole guide here, which is more about RPI OS than CircuitPython bu...
Many thanks for the information and links. I'm on vacation next Thursday and will try it out.
arctic cliff
Many thanks for the information and links. I'm on vacation next Thursday and wi...
It works with this bash script under Ubuntu 24.04 . It's not pretty, but it should work as a temporary solution. Here I have used the solution from Pimoroni.
#!/bin/bash
PY_ENV_DIR=$HOME/.virtualenvs/adafruit
if [ ! -f $PY_ENV_DIR/bin/activate ]; then
printf "Creating user Python environment in $PY_ENV_DIR, please wait...\n"
mkdir -p $PY_ENV_DIR
python3 -m venv --system-site-packages $PY_ENV_DIR
fi
printf " ↓ ↓ ↓ ↓ Hello, we've activated a Python venv for you. To exit, type "deactivate".\n"
source $PY_ENV_DIR/bin/activate
read -p "Press any key to update Ubuntu"
sudo apt update && sudo apt full-upgrade -y
read -p "Press any key to update CircuitPython"
cd circuitpython
git pull
git checkout main
read -p "Press any key to fetch all submodules"
make fetch-all-submodules
read -p "Press any key to update requirements and update ARM path"
pip3 install --upgrade -r requirements-dev.txt
pip3 install --upgrade -r requirements-doc.txt
export PATH=/home/$USER/bin/arm-gnu-toolchain-13.2.Rel1-x86_64-arm-none-eabi/bin:$PATH
export PATH=$HOME/circuitpython/mpy-cross/build:$PATH
read -p "Press any key to compile mpy-cross"
make -C mpy-cross
read -p "Press any key to compile Raspberry Pi Pico"
cd ports
cd raspberrypi
make clean BOARD=raspberry_pi_pico
make -j16 BOARD=raspberry_pi_pico
cp build-raspberry_pi_pico/firmware.uf2 /mnt/c/Users/pasca/Documents/CircuitPython/CircuitPython\ Firmware/Raspberry\ Pi\ PICO/
read -p "Press any key to compile Raspberry Pi Pico W"
make clean BOARD=raspberry_pi_pico_w
make -j16 BOARD=raspberry_pi_pico_w
cp build-raspberry_pi_pico_w/firmware.uf2 /mnt/c/Users/pasca/Documents/CircuitPython/CircuitPython\ Firmware/Raspberry\ Pi\ PICO\ W/
read -p "Press any key to exit script"
I still have a problem with the Espressif MCs because the Espressif build environment also uses a venv. I still have to find a solution for Ubuntu 24.04.
onyx hinge
It's possible to just make pip stop with its advise about the "externally managed environment", if you prefer to keep the old behavior where pip can install packages into the user's home directory without asking extra nicely. Article's written about Debian but should apply to Ubuntu, because it relates to apt & dpkg which Ubuntu is (afaik) using to package & publish python: https://emergent.unpythonic.net/01678908675
it's not Adafruit's advice but it's what I do on my own systems
the specific command may need changing if 24.04's python is 3.12 instead of 3.11.
tame creek
It's possible to just make pip stop with its advise about the "externally manage...
it seems unfortunate that a well-intended change breaks pip install --user, but it also sounds like that it might be hard to avoid doing that while accomplishing the other desired changes
tulip sleet
I started upgrading my regular dev machine to 24.04 today, and expect to figure out the easiest way to do this (at least for me). I don't expect to override the "externally managed environment" requirement, but instead will try to make using the virtual environment as easy as possible.
small comet
I feel I've run into this before but can't remember the solution 🙂 is there a "standard" way to install python module dependencies just for tests for adafruit (pure python) libraries? Is it just a matter of adding it to optional_requirements.txt or is there a different approach you'd recommend?
midnight ember
Speed improvement for drawing BMP's on the 7" RA8875 driver. https://github.com/adafruit/Adafruit_CircuitPython_RA8875/pull/31
before & after
midnight ember
I'm trying to get raw jpeg decoding working and it's been a rough night of going in circles with chatgpt. Seems like jpegio is only for displayio? Any chance someone has a raw jpeg decoder? Jepler perhaps?
I'm using a 192x128 24-bit JPG. I have the window padding showing up and I can see some color data but it's all scrambled and I cannot figure out why.
What I have so far. My first time with jpg headers so no, I do not know what I'm doing. I've been at it for about 6 hours now.
BMP works great, JPG has been a struggle.
lone axle
It seems that the report at https://circuitpython.org/contributing/library-infrastructure-issues is a week or so out of date.
small comet
BMP works great, JPG has been a struggle.
the jpeg data itself is compressed, in a somewhat complicated way. While not impossible, might be hard/tedious to do in pure python. While I haven't used it, is using https://docs.circuitpython.org/en/latest/shared-bindings/jpegio/index.html not an option?
midnight ember
the jpeg data itself is compressed, in a somewhat complicated way. While not imp...
I don't think so because I think jpegio uses displayio? The RA8875 display does not work with displayio... otherwise I wouldn't even be attempting this kind of thing.
The amount of info out there for decoding jpgs with python is pretty slim. I'm hoping Jeplers work on it is recent enough that maybe there's a raw decode function to jpeg decode that I've completely missed.
99% of searches resulted in either PIL or numpy. We do have numpy so that might be an option but one that is way over my head.
lone axle
It seems that the report at https://circuitpython.org/contributing/library-infra...
Looks like the build actions have indeed been failing on circuitpython-org for a few days. This was the first failed run. It's error seems to be about node JS versions.
Ah, maybe that node.js bit is just a warning and not the root cause of the action fail.
build
Node.js 16 actions are deprecated. Please update the following actions to use Node.js 20: actions/checkout@v3. For more information see: https://github.blog/changelog/2023-09-22-github-actions-transitioning-from-node-16-to-node-20/.
Ack. Out of curiosity, is piggybacking off of https://github.com/adafruit/Adafruit_CircuitPython_ImageLoad/tree/main a different option? Won't get jpeg, but at least bmp, gif and png seem feasible. (It seems to just need something that pretends to be a displayio.Bitmap in terms of API, all the functionality in python.)
lone axle
but in the log for the failed run it looks like this is the root cause of it:
An error occurred while installing sass-embedded (1.58.3), and Bundler cannot
continue.
In Gemfile:
jekyll-get-json was resolved to 1.0.0, which depends on
jekyll was resolved to 4.3.3, which depends on
jekyll-sass-converter was resolved to 3.0.0, which depends on
sass-embedded
I don't have much knowledge or experience with Ruby / Gems, but I'll try to figure out how to get this straightened out. If anyone with more experience sees this feel free to ping me if you happen to know what it means / how to resolve.
midnight ember
Ack. Out of curiosity, is piggybacking off of https://github.com/adafruit/Adafru...
Don’t know. Will look into it. I thought palettes and shaders were displayio things.
slender iron
@small comet you were interested in renode right?
small comet
<@1130377638523977739> you were interested in renode right?
Ah, yes - a while back. I floundered trying to get something setup but didn't make much progress. Any new developments?
slender iron
yup! I have a PR adding a renode port
I was at a conference where renode folks helped me out
Hopefully the PR will go in today
small comet
Fantastic! Looking forward to it - will definitely check it out and scratch my itch to run a few more CI tests!
slender iron
I didn't implement digitalio yet but it shouldn't be too bad
It sets up a fake SoC to run on instead of trying to run on a real one
small comet
Ah, gotcha. Once you've blazed the initial path, hopefully more pieces will get added to it. That's awesome
slender iron
next on my wish list is to enhance tracing to include uart output and memory values
lone axle
It looks like the version of Ruby being used inside the actions is 3.3.1 and the datestamp on it shows it was released at basically the same time that the build stopped succeeding. https://github.com/adafruit/circuitpython-org/actions/runs/8814751421/job/24195299302#step:5:27. My local version is ruby 3.0.2p107 I think I just got it with apt install ruby but I'm not 100% certain. I'll try to figure out how to get a newer version. It seems we'll need to do some kind of updating to get things working under that newer version.
small comet
next on my wish list is to enhance tracing to include uart output and memory val...
Starting to poke through https://github.com/adafruit/circuitpython/pull/9199 - very cool 🙂 do you already have something in the works to get renode running your port in the CI?
slender iron
Starting to poke through https://github.com/adafruit/circuitpython/pull/9199 - v...
I don't. no
tulip sleet
It seems that the report at https://circuitpython.org/contributing/library-infra...
is the report in S3 also out of date?
no, that looks ok
lone axle
The library report one for the meeting?
tulip sleet
yes
lone axle
it doesn't really matter, but it's interesting that mine are sorted into slightly different order
tulip sleet
firefox vs chrome!!
lone axle
I've looked a bit further into the failed library infrastructure issue (There are a few follow up comments above with what I've found). I do think that issue affects specifically the circuitpython-org build.
Ahh
tulip sleet
i would not expect that difference. That listing is generated by javascript
tulip sleet
I've looked a bit further into the failed library infrastructure issue (There ar...
https://github.com/adafruit/circuitpython-org/actions shows failed actions
some ruby library version issue
lone axle
Yep, I think more specifically it started using ruby version 3.3.1 and that one isn't able to install the dependencies
tulip sleet
Yep, I think more specifically it started using ruby version 3.3.1 and that one ...
do you have time to open an issue and tag @jwcooper on it?
lone axle
Yep, I can do that.
<@&356864093652516868> Reminder: The weekly meeting will occur here on discord at the usualy time of 2pm Eastern / 11 am Pacific. That is about 1 hour and 15 minutes from now. Feel free to fill in hugs and status reports to the shared document ahead of time if you can https://docs.google.com/document/d/13qSa787bMKb6fA1Z0utzht-5TBOAT-q1_aWa2L-mAaw/edit?usp=sharing. We look forward to hearing from everyone!
blissful pollen
Don’t know. Will look into it. I thought palettes and shaders were displayio thi...
I don't really have time to look too closely but after doing a lot of the gif work I think some of that would be similar, so can help if you want someone to look over some code.
That said palettes and shaders are displayio concepts vs jpeg. What I did for GIF was load the pixel data into a Bitmap and then used a palette or shader as required
gilded cradle
@lone axle I'm having internet issues, so likely won't be able to read my notes today.
slender iron
@onyx hinge looks like xobs did renode esp32-s3 simulation: https://git.xobs.io/xobs/renode-esp32s3/src/branch/main
midnight ember
I don't really have time to look too closely but after doing a lot of the gif wo...
I did think about converting it to a bmp but I still have to get jpeg working correctly before I can think about stuffing it into another format. The compression is very nice to have as the original BMP is about 147KB and the same image inJPG is 13 KB. Will have to learn about diffie hellman algorithm.
blissful pollen
I did think about converting it to a bmp but I still have to get jpeg working co...
You will need that full size to display it though. Or would have to decode / display it on the fly which may be hard with displayio
stuck elbow
(it's not the same image)
onyx hinge
@tulip sleet what's the error you're seeing with setuptools & virtual environments?
tulip sleet
example:
pip3 install --upgrade cpp-coveralls
Collecting cpp-coveralls
Using cached cpp-coveralls-0.4.2.tar.gz (15 kB)
Preparing metadata (setup.py) ... error
error: subprocess-exited-with-error
× python setup.py egg_info did not run successfully.
│ exit code: 1
╰─> [1 lines of output]
ERROR: Can not execute `setup.py` since setuptools is not available in the build environment.
[end of output]
note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed
× Encountered error while generating package metadata.
╰─> See above for output.
note: This is an issue with the package mentioned above, not pip.
hint: See above for details.
setuptools IS installed, and is latest version
onyx hinge
what's the whole thing you run to get to that point?
tulip sleet
that was just the operative bug in pip3 install -r requirements-dev.txt.
requirements-doc.txt installs completley.
I'm in a venv in ubuntu 24.04
onyx hinge
but you ran some commands to create & activate the environment
tulip sleet
sudo apt install python3.12-venv
python3 -m venv ~/.py
source ~/.py/bin/activate
this is a brand new clean install of ubuntu 24.04
(.py) halbert@cod:~/repos/circuitpython$ python3 --version
Python 3.12.3
(.py) halbert@cod:~/repos/circuitpython$ pip3 list |grep setuptools
setuptools 69.5.1
setuptools-scm 8.0.4
(.py) halbert@cod:~/repos/circuitpython$ pip3 list |grep pip
pip 24.0
(.py) halbert@cod:~/repos/circuitpython$
or somehow it is "locked out" of the build env
right, seemed like it might be something like this, but my websearching has so far not turned up the problem
onyx hinge
$ docker run -it ubuntu:24.04
root@f8fca6e6316a:/# apt update && apt install --no-install-recommends --no-install-suggests -qy python3 python3-pip python3-venv && python3 -mvenv _env && _env/bin/pip3 install cpp-coveralls
...
Successfully installed certifi-2024.2.2 charset-normalizer-3.3.2 cpp-coveralls-0.4.2 future-1.0.0 idna-3.7 requests-2.31.0 urllib3-2.2.1
```I'm not getting that outcome here in a docker image fwiw
silver tapir
I just tested it on a WSL ubuntu 24.04 and it worked as well, and I do not have setuptools installed.
tulip sleet
maybe I will uninstall setuptools. Will try more after the meeting
spare jacinth
right, seemed like it might be something like this, but my websearching has so f...
i mean, it seems very counterintuitive to do something like "nah, we dont want this package when building", but my understanding of the text was that...
it does not say "not installed" but "not available on build"
wording is hard anyway, perhaps it means not installed but without the best words for it 😛
i have no idea how eggs/wheels/whatever work, so i cant give any feedback other than reading it too, lol
tulip sleet
```
$ docker run -it ubuntu:24.04
root@f8fca6e6316a:/# apt update && apt install...
uninstalling setuptools causes
ModuleNotFoundError: No module named 'distutils'
error when trying to install cpp-coveralls
i can try the docker locally
midnight ember
I like how the weather forecast icons are circular around a clock, that's a great idea.
tulip sleet
if I avoid cpp-coveralls, the next module that fails, in the same way, is jsmin
onyx hinge
oops I forgot to join the voice channel 
turbid radish
Please tell your colleagues, friends, students to subscribe
tulip sleet
aha:
ython 3.12.3 (main, Apr 10 2024, 05:33:47) [GCC 13.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import setuptools
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/halbert/.py/lib/python3.12/site-packages/setuptools/__init__.py", line 9, in <module>
import distutils.core
ModuleNotFoundError: No module named 'distutils'
wasnt distutils moved into setuptools (or something like that) on a recent-ish Py update?
gilded cradle
I can read
onyx hinge
[huh I've heard of others having this problem (not hearing specific users) but I couldn't hear deshipu talk. linux, using discord app]
tulip sleet
it was really faint for me
silver tapir
LOL on that 
midnight ember
happens to everyone
slender iron
I had tested it on one board and turned it on there
onyx hinge
textual is pretty cool
onyx hinge
I'll be there!
tulip sleet
@stuck elbow will you be going anywhere else in the US?
stuck elbow
<@351136009922019331> will you be going anywhere else in the US?
No, only Pittsburgh, not enough vacation time to do anything else
onyx hinge
I don't think anybody's done that (I haven't done that)
midnight ember
silver tapir
need to drop for another meeting, I'll check in the weeds on the recording.
stuck elbow
people will still install those libs from pip without reading the docs
tulip sleet
people will still install those libs from pip without reading the docs
i wonder if pypi has any way of causing a warning to be printed
slender iron
looks normal time to me
midnight ember
Thank you for hosting @lone axle have a great week everyone.
short tendon
Thanks all!
solar whale
@onyx hinge with a few simple changes for the pinouts, the "ascii-mirror" demo works on the pico/picowbell -- progress.
lone axle
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/1PhxHRJjWKLOBTZeKI6XBk54aA3Bz_tPxidR6AKUuY6U/edit?usp=sharing
lone axle
I found that the PyCamera library is "circuitpython-pycamera" inside of RTD instead of "adafruit-circuitpython-pycamera". The latter I believe is the naming followed by most of the libraries (but I'm not certain). I've submitted a PR to change the badge link URL in the readme: https://github.com/adafruit/Adafruit_CircuitPython_PyCamera/pull/38 but it occured to me after submiting maybe we'd want to change the name inside of RTD instead if the latter convention is more typical and if that is an option. So I'm posing that here.
tulip sleet
I found that the PyCamera library is "circuitpython-pycamera" inside of RTD inst...
it is a mistake that it is called circuitpython-pycamera if it is an adafruit repo
lone axle
It is an Adafruit repo. The name in github (and pypi) do both contain the adafruit part, it looks like just the RTD name is missing the "adafruit" prefix.
tulip sleet
can you change the rtd name or do you need someone with more privs to do it?
lone axle
can you change the rtd name or do you need someone with more privs to do it?
I don't believe I have access in RTD. Will need someone with higher privilage in there.
tulip sleet
I don't believe I have access in RTD. Will need someone with higher privilage in...
i did something. it's confusing whether I did what's necessary, but there is this URL: https://docs.circuitpython.org/projects/pycamera/en/latest/
what url do you need?
and this now works: https://readthedocs.org/projects/adafruit-circuitpython-pycamera/
lone axle
That appears to have done the trick 🎉 the badge URL https://readthedocs.org/projects/adafruit-circuitpython-pycamera/badge/?version=latest is working now on the readme page to show the current badge status.
tulip sleet
rtd's project struture is not obvious
okee dokee, not sure if it's listed twice now, but I think not
lone axle
Agree'd
lone axle
Can anyone think of anything else that would be good to mention, or any improvements to the wording of this message for creating 'good first issue's from?
We would like to have a basic displayio example for this library. The example should be written for micro-controllers with a built-in display. At a minimum it should show a Label on the display and update it with live data readings from the sensor.
The example should not be overly complex, it's intended to be a good starting point for displayio based projects that utilize this sensor library. It should be written to support as many micro-controllers as is feasible. Try to keep all visual content as near to the top left corner as possible in order to best facilitate devices with small built-in display resolutions.
The new example should follow the naming convention `examples/libraryname_displayio_simpletest.py` with "libraryname" being replaced by the actual name of this library.
You can find an example of a Pull Request that adds this kind of example here: https://github.com/adafruit/Adafruit_CircuitPython_BME680/pull/72
We have a guide that covers the process of contributing with git and github: https://learn.adafruit.com/contribute-to-circuitpython-with-git-and-github
If you're interested in contributing but need additional help feel free to join the Discord server to ask questions: https://adafru.it/discord
Looks really good, I wouldn't change any of the wording. One minor thing, I'd remove the - from microcontroller.
lone axle
Thank you!
short tendon
Question, does Adafruit pay for GitHub, I ask because since the libraries all have the same linting and such (or should), things could be cached to run faster, and if applicable save money
tulip sleet
We don’t pay for the open-source public repos. But if you have som suggestions about saving time, we’re all ears
short tendon
Sounds good. I'll look through some of the tooling repos and see what options I can come up with. There's so much downloading of libraries, that caching seems like it would save a good chunch of time...
manic glacierBOT
on ESP32, ADC2 is not available when WiFI is in use. I originally though ADC1 was not connected to any pins, but was used only internally. I think it was my misinterpretation of this table from the ESP32 Technical Reference manual:
tulip sleet
@onyx hinge I created another fresh user on my 24.04 machine and that user can install cpp-coveralls in an identical venv. So I thought it might be something in my PATH, but pruning that down did not seem to help. I did some strace comparisons, but it's getting out of hand. But in any case seems to be something about my own environment.
it's useful to remove .cache/pip to start from scratch. I'll try some more "bare environment" things and see if I can get it to work.
I also compared the file tree in the two venv's and they were functionally identical
tulip sleet
@onyx hinge it was because of this:
https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/check-your-code#workaround-for-pre-commit-issues-on-ubuntu-and-debian-3086216
I had added that export to my .bash_aliases.
If I take it away the pip install is fine, though now pre-commit complains, whether or not it's running in a venv.
tulip sleet
If I update the pylint in a .pre-commit-config.yaml (for instance, in Adafruit_CircuitPython_HID) then errors go away. This is after installing the latest pre-commit into my venv
onyx hinge
😕 we are a bit stuck due to whatever the exact incompatibility is of that old pylint.. is work on switching to ruff ongoing or stalled?
short tendon
@onyx hinge did I drop the ball? I thought ruff was in your court?
We can also just bulk update pylist to get us past this point
onyx hinge
me personally, or adafruit's court generally? I haven't been working on it.
short tendon
My impression was you personally. But I have no idea how priorities work. I can go update all the libraries and make PRs (and automate what I can)
onyx hinge
oof, OK. no, this isn't something Adafruit's prioritized for me to work on. 😕
short tendon
okay. @tulip sleet or @onyx hinge how can I help
We could update the pylint version and do the patch. Any repo that then fails tests I can pump through updating
tulip sleet
@lone axle has tools to make these sorts of bulk changes
we don't have to do this now. This is just a local problem. We have pinned python to 3.11 in the repos
this was just trying to figure out why i was having trouble installing things in a 3.12 system. It was due to a workaround for the pre-commit problem, which is due in part to the older pylint. So it's a multistage issue, but not urgent. I can always use a 3.11 venv locally to run pre-commit.
short tendon
okay... I still think getting pylint updated or switched to ruff would be a good idea. Is there a way I can help there?
tulip sleet
we can wake up the ruff work, but it's not urgent. We can ask foamyguy tomorrow or later.
short tendon
👍
manic glacierBOT
lone axle
<@382939733107408897> has tools to make these sorts of bulk changes
@short tendon, @onyx hinge. IIRC from when Ruff was discussed during a meeting (or perhaps before/after in discord) we were going to start moving forward with rolling it out. I think we got @proven garnet involved at that time.
Tekktrik is this something you have the time to help with (rolling out Ruff to libraries?).
If not I can try to take a look into it with the tools that I've got, but I'm not sure that the way I've done things in the past will be able to fully automate the updates to Ruff compliant syntax. It could update the config files, but there would still be manual work to update the code to make it pass the new actions. I haven't really attempted, but this is my best guess based on how I've done some of the automation in the past.
fringe yarrow
Are we going to have a celebration when support hits 500 boards? Only 5 more to go..
fringe yarrow
I've already done all the stuff I had in my collection.. Was hoping I might get number 500, but I'm out of boards to do..
tulip sleet
there will always be more boards 🙂
fringe yarrow
I have the Heltec Wireless Paper board, which has an epd that's supposedly supported by CircuitPython, if the compatibility matrices between epd models are to believed.. But I can't get it to refresh without locking up.. Similar to that last submitted Issue, but instead of once in a while, it's every time..
tulip sleet
every board that gets added adds to the build times, so I encourage people to add boards they are going to use, not just boards for boards' sake (i.e. not random aliexprss junk)
fringe yarrow
Understandable.. the only one I'm struggling to find a use for that I've recently submitted is that 7" IPS display ESP32S3.. It's too big for most projects..
proven garnet
<@1152454697614069811>, <@426200547024961539>. IIRC from when Ruff was discussed...
Hi, sorry I have not due to outside circumstances. I do plan to work on it in the coming weeks but at this time it may be better for someone else to take the helm.
short tendon
@onyx hinge sorry, too many new people, I got confused on who was working on it...
onyx hinge
@short tendon no problem, and of course the fact is I do drop the ball more than I'd like to admit
short tendon
<@1152454697614069811>, <@426200547024961539>. IIRC from when Ruff was discussed...
@lone axle can we run it against a subset of repos? And then open up PRs with the file changes (to do them in chunks)?
solar whale
I'm trying to use a PicoW with AIO and am running into memory errors in the connection manager -- am I doing something wrong? Here I am jsut trying to access a feed ```
Adafruit CircuitPython 9.1.0-beta.1-14-g126a1a4154 on 2024-04-29; Raspberry Pi Pico W with rp2040
import aio
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "aio.py", line 33, in <module>
File "adafruit_io/adafruit_io.py", line 748, in get_feed
File "adafruit_io/adafruit_io.py", line 578, in _get
File "adafruit_requests.py", line 591, in get
File "adafruit_requests.py", line 525, in request
File "adafruit_connection_manager.py", line 290, in get_socket
File "adafruit_connection_manager.py", line 215, in _get_connected_socket
MemoryError:
code runpy
import os
import ssl
import wifi
import socketpool
import adafruit_requests
from adafruit_io.adafruit_io import IO_HTTP, AdafruitIO_RequestError
WiFi
Add settings.toml to your filesystem CIRCUITPY_WIFI_SSID and CIRCUITPY_WIFI_PASSWORD keys
with your WiFi credentials. DO NOT share that file or commit it into Git or other
source control.
Set your Adafruit IO Username, Key and Port in settings.toml
(visit io.adafruit.com if you need to create an account,
or if you need your Adafruit IO key.)
aio_username = os.getenv("ADAFRUIT_AIO_USERNAME")
aio_key = os.getenv("ADAFRUIT_AIO_KEY")
pool = socketpool.SocketPool(wifi.radio)
requests = adafruit_requests.Session(pool, ssl.create_default_context())
Initialize an Adafruit IO HTTP API object
io = IO_HTTP(os.getenv("AIO_USERNAME"), os.getenv("AIO_KEY"), requests)
found_feed = False
Adafruit IO feed configuration
try:
# Get the 'camera' feed from Adafruit IO
feed = io.get_feed("picocam")
print("Got feed")
found_feed = True
except AdafruitIO_RequestError:
# If no 'camera' feed exists, create one
# it neese to be amnually created
print("Create a feed named picocam and turn off history")
My preference is to ideally spend as little time as possible with some libraries one way and other libraries another way. During that time where some are one way and others are different it will make submitting and reviewing PRs much more convoluted I think.
IIRC when it was discussed last I also thought we had talked about commiting changes directly rather than having to make PRs with the changes since it would end up being 1 per repo basically which will be a load of extra clicking around in the front end interfaces to get PR created / reviewed / merged.
tulip sleet
I'm trying to use a PicoW with AIO and am running into memory errors in the conn...
more RAM got used up at one point. We turned off some things, I thought, to make it better. What does help("modules") show? I want to see which native modules are still in the build
short tendon
IIRC when it was discussed last I also thought we had talked about commiting cha...
This isn't something I can help with... We could:
- apply the new files - commit
- run pre-commit a few times - commit
- any libraries that are failing, I'm happy to open PRs for
tulip sleet
usb_host and usb_video are turned on, we wanted to turn those off. They grab ram
solar whale
@tulip sleet - sorry -- I lost track of the time and have to run out. I can try a build removing them later today or maybe tomorrow -- Thanks.
tulip sleet
<@329766224093249548> - sorry -- I lost track of the time and have to run out. ...
np -ttyl, I thought there was a PR 👋
solar whale
I'll check on that
tulip sleet
@solar whale for reference:
PR was here: https://github.com/adafruit/circuitpython/pull/9116 but was not adjusting Pico W.
Pico W RAM discussion is here:
https://github.com/adafruit/circuitpython/issues/9085
@short tendon so I will make the ESP32SPI and Wiznet libs get a major version bump after merging your PR's, right?
short tendon
<@1152454697614069811> so I will make the ESP32SPI and Wiznet libs get a major v...
Correct
short tendon
Also, what do you want to do on that learn guide? Because it doesn't work now anyway. adafruit_azureiot dropped support for the ESP32SPI
Although, it wouldn't be hard to add in
ConnectionManager should also get a major bump
tulip sleet
oh, really, ok, maybe guide should be redone. I can mention it to Brent, the guide's author
what is the connectionmanager incompatibility that I should note in the release notes?
i thought it was only additional support for ESP32SPI and Wiznet, so I thought it was a minor bump
short tendon
it's changed to use the new methods for ESP32SPI and Wiznet
I guess it can be a minor
tulip sleet
so if you were using CM for ESP32SPI, that changed. (Now I forget whether it worked at all with ESP32SPI before.)
if that changed, yup, that's an incompatibility. no problem bumping the major veresion
short tendon
it worked with it before. but if you did an update with circup I think I would want it to ask for updating both
tulip sleet
i will do major, np
summary
2.0.0 - updated to handle changes in ESP32SPI and WIZNET5K API's
@short tendon ok all three release with major version bumps. Going for a walk.
short tendon
WOOT!!!
slender iron
every board that gets added adds to the build times, so I encourage people to ad...
This is a problem we can fix. We shouldn't turn people away who want to add boards. Lots of people use boards off of aliexpress.
short tendon
@tulip sleet you can see all the old socket code that's left is:
- an Azure Learn Guide
- the
AzureIoTexamples - the deprecated
wsgiapp
https://github.com/search?q=org%3Aadafruit+"adafruit_esp32spi_socket+as+socket"+language%3APython&type=code
tulip sleet
<@329766224093249548> you can see all the old socket code that's left is:
- an A...
Thanks, I will mention the azure stuff to Brent. You said it already dropped support for ESP32SPI somehow?
short tendon
Thanks, I will mention the azure stuff to Brent. You said it already dropped sup...
Yes 2 years ago: https://github.com/adafruit/Adafruit_CircuitPython_AzureIoT/commit/75af72b6b80388f2365c736039deb8507b11451d. It was hard-coded to use ssl. I fixed a few things recently that made it work in general: https://github.com/adafruit/Adafruit_CircuitPython_AzureIoT/commit/10bb5307e030e18edf723bfda20e9addaf9e5c2c, but didn't spend the time to get it working for ESP32SPI .
Not that it would be hard to add
tulip sleet
Thanks - I will bring this up.
short tendon
Cool
slender iron
@tulip sleet around for a quick design chat?
tulip sleet
sure
prime flower
@short tendon Hi! You mentioned you fixed a few things that made the AzureIoT library work in general, namely fixing ssl and set_socket. What work would be required to get this library working with the ConnectionManager/ESP32SPI like we have in other libs. like Adafruit IO CircuitPython?
manic glacierBOT
solar whale
<@343570430042308609> for reference:
PR was here: <https://github.com/adafruit/...
Thanks for the links. I’ll try the “slim build” tomorrow to see if that helps. My goal is to be able to use the OV5640 camera as well so memory will be a challenge…
short tendon
<@1152454697614069811> Hi! You mentioned you fixed a few things that made the A...
So looking at iot_mqtt.py:
we have:
def __init__(
self,
callback: IoTMQTTCallback,
socket,
iface,
hostname: str,
device_id: str,
device_sas_key: str,
token_expires: int = 21600,
logger: Logger = None,
):
and then later:
self._mqtts = MQTT.MQTT(
broker=self._hostname,
port=8883,
username=self._username,
password=self._passwd,
client_id=self._device_id,
is_ssl=True,
keep_alive=120,
socket_pool=self._socket,
ssl_context=ssl.create_default_context(),
)
The ssl.create_default_context() is the problem.
If we updated it to take a SocketPool (which is what self._socket is already) and a ssl_context (instead of iface which isn't even used anymore), if should work.
And then the same thing in device_registration.py
manic glacierBOT
prime flower
So looking at `iot_mqtt.py`:
we have:
```
def __init__(
self,
...
Do you have an azure IoT account to test this change with?
short tendon
Do you have an azure IoT account to test this change with?
I do. I created a demo one to fix the other bug that was found
prime flower
I do. I created a demo one to fix the other bug that was found
Okay, I see where the issue is.. I'm wrapped up in a large project currently and may not have time to implement this. If you are willing to make the changes to update the ssl.create_default_context() calls, and tag me on a PR, I can review and test. When changes are in-place, I'll also do a pass over that planter guide I wrote to ensure it's working.
solar whale
@tulip sleet FYI, I had bit of time, so I tried turning off PICODVI and USB_HOST,```GEN build-raspberry_pi_pico_w/genhdr/qstrdefs.generated.h
/Applications/ArmGNUToolchain/13.2.Rel1/arm-none-eabi/bin/../lib/gcc/arm-none-eabi/13.2.1/../../../../arm-none-eabi/bin/ld: warning: build-raspberry_pi_pico_w/firmware.elf has a LOAD segment with RWX permissions
Memory region Used Size Region Size %age Used
FLASH_FIRMWARE: 1367472 B 1532 KB 87.17%
RAM: 89940 B 256 KB 34.31%
SCRATCH_Y: 0 GB 4 KB 0.00%
SCRATCH_X: 0 GB 4 KB 0.00%
Converted to uf2, output size: 2735104, start address: 0x10000000
Wrote 2735104 bytes to build-raspberry_pi_pico_w/firmware.uf2
jerryneedell@Mac-mini raspberrypi % scp build-raspberry_pi_pico_w/firmware.uf2 gjnpi5nvme.local:picow.uf2
thanks for testing. as bablokb mentioned, there is still a lot of increased ram consumption
solar whale
This does not bode well for using the camera with the pico-w.... too bad.
crimson ferry
@solar whale what does the RAM section there mean? On a fresh reload, I get gc.mem_free() of 126160 with CP 9, 16K more on CP 8 (Pico W). I think the assumption was that it was due to those two modules, but maybe there's something else going on.