Okie dokie, added
#circuitpython-dev
proven garnet
proven garnet
Ah, sadly the rules say drafted PRs can't be accepted ( @ruby atlas )
onyx hinge
feel free to move it out of 'draft', I did that.
@ruby atlas @proven garnet ^
I marked it ready for review again, sorry for causing trouble
and re-marked as accepted
thanks rose!
proven garnet
🥳
idle owl
No, we're evidently leaving it unarchived, but we're not supporting it any further. I will have to ask Limor whether she wants to accept PRs to it, or whether she's ready to archive it to make it read only. This question has come up a few times previously, and the answer was always don't archive. But who knows now.
idle owl
No, we're evidently leaving it unarchived, but we're not supporting it any furth...
@proven garnet Please go through the PRs as you would on any repo. If they're reasonable, we can accept them. If they're not, we don't, as usual. We don't support the library, but we are willing to take updates to it. We will not be archiving at this time.
mystic flume
I'm looking at the asyncio code currently. There seem to be two versions - one is in the main repo under extmod/asyncio, the other is at https://github.com/adafruit/Adafruit_CircuitPython_asyncio
I think I've found a bug (and a fix), but not sure where to file it/ do a PR etc
Anyone got any pointers?
tulip sleet
I'm looking at the asyncio code currently. There seem to be two versions - one i...
Use the library. The extmod version is what is in MicroPython. We should probably delete it from our sources.
mystic flume
Thanks, will do
I think we can probably use the Streams class with UARTs, usb cdc serial ports and ble uarts as well
onyx hinge
@tulip sleet now that our tests use our asyncio module I think we can do that.
@mystic flume there are rough edges so come back and help us solve the problems when you run into them 🙂
tulip sleet
it would be good to track changes to that code, but we can do that in the merges
mystic flume
@tulip sleet I've found several issues in asyncio/stream.py - where is the best place to report these. I have solutions for some but not all of them . Should it be done as multiple issues on github, or discuss here?
orchid basinBOT
ornate breach
<@329766224093249548> I've found several issues in `asyncio/stream.py` - where i...
Best place would be to file an issue in the asyncio repo https://github.com/adafruit/Adafruit_CircuitPython_asyncio
lone sandalBOT
quasi veldt
Is there any plan for USB debugging? Would love to be able to have breakpoints in python and debug interactively instead of with print statements.
Used to that with Atmel studio (and also for web dev or desktop python)
tulip sleet
Is there any plan for USB debugging? Would love to be able to have breakpoints i...
very old issue about this: https://github.com/adafruit/circuitpython/issues/298
MicroPython has some debugging support with sys.settrace(). Local variables are not available for inspection in MicroPython/CircuitPython, so there would be limits on what you could do.
search for other pdb stuff in the MicroPython issues as well
quasi veldt
search for other `pdb` stuff in the MicroPython issues as well
Python + gdb = pdb?
tulip sleet
well, I think just Python + DeBug
gdb is Gnu DeBugger
print statements to the rescue
quasi veldt
I guess you can inject a line to print local vars
Set trace to stop on a line. At that line use repl? Or locals still won't be available
tulip sleet
still won't be available, I think
lavish saffron
Sanity check please!
I've been reading through the code for the DNS server in Adafruit-CircuitPython-wiznet5k, and I think there is a bug. In _build_dns_question() the line self._pkt_buf += host[i] appears to concatenate a str onto a bytearray which should raise a TypeError exception.
Code snippet below (my thoughts added as # > Blah blah
Can someone please confirm that the code runs or have a read and see if you agree?
Thx!
adafruit_wiznet5k/adafruit_wiznet5k_dns.py
class DNS:
"""W5K DNS implementation."""
def __init__(
self,
iface: WIZNET5K,
dns_address: Union[str, Tuple[int, int, int, int]],
debug: bool = False,
) -> None:
...
# > self._pkt_buf is bytearray
self._pkt_buf = bytearray()
...
def _build_dns_question(self) -> None:
"""Build a DNS query."""
# > self._pkt_buf is bytearray
# > (not touched after __init__)
# > self._host is bytes
# > host is str
host = self._host.decode("utf-8")
host = host.split(".")
# write out each section of host
for i, _ in enumerate(host):
# append the sz of the section
self._pkt_buf.append(len(host[i]))
# append the section data
# > concatenate bytearray and string should raise TypeError
self._pkt_buf += host[i]
# > **** ^ ****
# end of the name
self._pkt_buf.append(0x00)
# Type A record
self._pkt_buf.append(htons(TYPE_A) & 0xFF)
self._pkt_buf.append(htons(TYPE_A) >> 8)
# Class IN
self._pkt_buf.append(htons(CLASS_IN) & 0xFF)
self._pkt_buf.append(htons(CLASS_IN) >> 8)
jaunty juniper
Sanity check please!
I've been reading through the code for the DNS server in A...
it should throw an exception, and does in Python 3, but doesn't in Circuitpython because of optimizations that treat strings and bytes and stuff similarly internally or something
so you're right, it's a bug
a silent one in CP, but it should be fixed to be correct C python
lavish saffron
I'll open an issue and submit a pull request with self._pkt_buf += bytes(host[i], "utf-8). Thanks for a quick response!
ancient hemlock
I know (at least I thought I knew...) in C there isn't really a way to make arguments optional. But I see in countio pull argument is optional - how does that work? Or is this a better question for the forums?
onyx hinge
@ancient hemlock I'll try to answer .. but let me check how the code is working in this specific case.
the code you're talking about is in the source file shared-bindings/countio/Counter.c and the function is countio_counter_make_new
jaunty juniper
onyx hinge
there's a structure that describes what is required from Python arguments. An argument can have the REQUIRED flag or not. The pull argument doesn't have the REQUIRED flag: ``` static const mp_arg_t allowed_args[] = {
{ MP_QSTR_pin, MP_ARG_REQUIRED | MP_ARG_OBJ },
{ MP_QSTR_edge, MP_ARG_KW_ONLY | MP_ARG_OBJ, {.u_obj = MP_OBJ_FROM_PTR(&edge_FALL_obj) } },
{ MP_QSTR_pull, MP_ARG_KW_ONLY | MP_ARG_OBJ, {.u_obj = mp_const_none} },
};
if it's not specified, then the default is the Python None value
ancient hemlock
yes, looking for something to model after.
onyx hinge
so later on when args[ARG_pull].u_obj is used, it'll be the default value instead of a user-provided value
ancient hemlock
okay so when you give that to your construct method it will be the none
onyx hinge
and the function validate_pull is prepared to accept the UP object, the DOWN object or the None object, and turn it into the corresponding value in C code: const digitalio_pull_t pull = validate_pull(args[ARG_pull].u_obj, MP_QSTR_pull);
so basically.. you need a special value (good choices might be 0, -1, or None, depending on the situation) that is used as a default if no value is specified in the Python code. Then, in C, you just check for the special value.
ancient hemlock
I think the none object is my missing link. Thank you.
I am modifying pulseio to allow for 32bit long pauses between events to use a hall sensor for slow to fast rotations
onyx hinge
cool, happy if this gets you un-stuck. btw I think asking questions about the core C code is more appropriate here than in help-with-circuitpython, because it is at a far different level than most questions are.
ancient hemlock
I would like to have the default behavior be the original max pulse limit, and be able to specify longer gaps. Thanks for the redirect to the right place also!
jaunty juniper
if you are compiling or writing C, questions go here 😉
onyx hinge
The reason for 16 bit values is probably a combination of wanting to save RAM size and maybe even a limitation of the counter peripheral on some microcontroller like the samd series
but I bet if you're using something like rp2040 you can have a bigger limit easily
ancient hemlock
I have a pi pico that I'm playing with now - exactly
onyx hinge
you might also just code it in pure Python using the rp2pio module and bypass countio / changing core code entirely
ancient hemlock
as in, use my own PIO to read the pulses instead of pulseio?
ancient hemlock
I think I'm in that position of being very new, where I grabbed on to something that looks familiar-ish
onyx hinge
I guess countio on rp2040 is using the pwm peripheral, I kinda didn't expect that
ancient hemlock
I started with countio, moved to pulseio for the exact pulse times then started to work down at the lower end of the range realizing the counter for pulseio was maxing/rolling over
onyx hinge
https://raspberrypi.github.io/pico-sdk-doxygen/group__hardware__pwm.html#ga157c8b11483967714823187725a5ed5a pwm_get_counter only operates on a 16-bit value so it may be a dead end to do it like countio does now (countio uses pwm)
pulsein is using the pio peripheral, so it shouldn't (knock on wood) have a 16-bit limitation
ancient hemlock
I have it working now so it will count to 32bit. Not being very familiar with the PIO I'm not very confident how it works exactly
onyx hinge
but the countio design needs the IRQ, you'd need a different design of pio program if it was going to work as pure python .. no interrupts from python code
OK, yeah, I think working based on improving pulsein within the core is the way to go!
ancient hemlock
other than it seems like it looks for a transition and then counts the cycles and reports them as microseconds
Not sure if this is something that you (collectively) would want in the main or not but I will play with it some more. I wonder if it was 16bit to be consistent across architectures...
manic glacierBOT
ancient hemlock
going back to pulseio and countio I see now how the arguments work...I just missed the pattern before
`static const mp_arg_t allowed_args[] = {
{ MP_QSTR_pin, MP_ARG_REQUIRED | MP_ARG_OBJ },
{ MP_QSTR_maxlen, MP_ARG_INT, {.u_int = 2} },
{ MP_QSTR_idle_state, MP_ARG_BOOL, {.u_bool = false} },
{ MP_QSTR_max_period, MP_ARG_INT, {.u_int = 65536} },
};
the
there is a default value as the last value of each argument object.
thanks again
orchid basinBOT
vague thicket
Sorry to bother you again, but does anybody know, what I need to change to make built work? When I wrote this (https://github.com/adafruit/circuitpython/pull/7135#issuecomment-1297785568) comment, repos were in sync. Many thanks for any hint.
orchid basinBOT
proven garnet
@onyx hinge that looks not great, is that Python thing gonna be an issue for libraries too?
It looks like it was for building the wheel
onyx hinge
some stuff may come up, but this particular thing is probably unique to esp-idf's requirements.txt
proven garnet
Gotchya, okay thanks. I still want to address issues with those actions so just trying to keep up with that
proven garnet
I thiiiink Python 3.11 is breaking pylint for library CI 😬
@idle owl I can go ahead and move all the libraries over to that new composite action if you want me to just push directly to mains, I can handle that tonight
Looks like wrapt hasn't been updated to stop using removed (previously depreciated) methods yet.
orchid basinBOT
idle owl
<@330227457296957440> I can go ahead and move all the libraries over to that new...
Yes, please.
proven garnet
Yes, please.
Youuuu got it! I'll take care of that ASAP (tonight)
mystic flume
@proven garnet, looks like setting pylint version to 2.13.0 or higher should fix that - seems to work in my recent PR for https://github.com/adafruit/Adafruit_CircuitPython_asyncio
proven garnet
Ahh, a preferable solution I think, thanks!
Is there any reason the core and lib need to (or should) use the same Python version?
I know @onyx hinge had the PR to pin the core's to 3.10
mystic flume
I think this is just the python version used to check code before commits. I think either way works tbh
Shouldn't actually make much difference as CircuitPython is (I think) based on 3.4 (with features from 3.5 to 3.7)
idle owl
@proven garnet Using a newer Pylint might mean another sweep of fixes to the libraries. Be aware of that. Eva can be involved if it turns out to be necessary.
idle owl
Is there any reason the core and lib need to (or should) use the same Python ver...
No. They are two entirely separate things when it comes to this.
proven garnet
<@135415127879974912> Using a newer Pylint might mean another sweep of fixes to ...
I was gonna say I'll run my tool, but adabot does it now daily too! 
I'll let you know, I also am likely to have a list of libraries that need other work too, if they aren't using the standard CI
I'll get that list prior
idle owl
@proven garnet I also think I want a change to the download stats page. And I have no idea how complicated it would be. Out of my immediate wheelhouse anyway, but it might be super simple to you.
Couple changes, maybe, now that I look at it rendered.
proven garnet
Sounds good, I'll take a look if you made them to the PR
idle owl
I'll try to explain in the PR review, but we can chat if what I suggest doesn't make sense.
I don't know how to make the biggest change. So I'll try to explain it instead.
proven garnet
It's worth mentioning the I put the Blinka stats in there not primarily for reference, but because I need it for the report, and so need to write it them to parse later.
It is a good comparison value, I think, but not the primary reason
idle owl
It's worth mentioning the I put the Blinka stats in there not primarily for refe...
Blinka stats can stay, I think it's a good piece of information to have in reference to the library downloads, since they basically go hand in hand (other than when pip installing for your editing environments etc.)
proven garnet
I take it that last one is the complicated one? I think all the libraries have the "same" PyPI name as repo name besides capitalization and dashes/underscores (my code and existing code relies on it even)
idle owl
I take it that last one is the complicated one? I think all the libraries have ...
Yes, the last one. I wasn't sure what you'd have to do to make that happen.
I assumed you could, at worst, parse the package names and use that, but that doesn't do camelcase for the lib name necessarily.
proven garnet
Totally doable 😄
idle owl
Thanks!
idle owl
Hah ok
idle owl
Two thumbs!
proven garnet
Working on moving things over to the composite actions!
proven garnet
Libraries now utilize the composite actions! Those need a couple updates so the libraries show as failing but that's expected right now. Sorry if someone just got 300 emails about it like I did!
In hindsight I should have done that first 
proven garnet
There's a lot to fix, but it all seems to be valid fixes need and not anything bogus from what I can tell
When the website updates library infrastructure issues, that should be the correct list.
Most issues seem to stem from two things:
no-memberraised as a result of CPython inconsistencies likeos.uname
- Names not conforming to the standard, often caused specifically the variable
cswhich is in a bunch of driver libraries for the chip select.
proven garnet
Latest pylint (2.15.5) seems better and fixes most of those werid errors
Yeah, much better, I'll move it to that and finish up this patch - the rest can be cleaned up
turbid radish
@lone axle do you have your newsletter update in?
lone axle
<@382939733107408897> do you have your newsletter update in?
😩 I do not. I am terribly sorry I have slipped on that a few times now. I will submit that shortly, and try my best to get back into the routine of submitting by Thursday evening.
idle owl
@lone axle @onyx hinge Is it one of you I need to talk to if there's an issue with the simulated-CIRCUITPY dynamic screenshot generator? I keep starting with Justin, and keep being reminded I'm asking the wrong person.
lone axle
<@382939733107408897> <@426200547024961539> Is it one of you I need to talk to i...
I believe I made the initial version of it that screenshotted the learn guide projects. I think Jeff adapted it to the library / product guide ones and hooked up the automagic actions running, but I've done one or two tweaks on that side as well iirc.
I will leave it open to Jeff also, but maybe best to start with me and if I hit a snag or something I can't figure out or work through I'll reach out for here for help.
idle owl
I believe I made the initial version of it that screenshotted the learn guide pr...
Keen, ok, so apparently it's not including .csv files. The Project Bundler itself includes the actual file, but the dynamic screenshot doesn't include the file name.
lone axle
Ah, I think there is an allow-list of file extensions that it will include. I'll poke into it this evening and see if I can add CSV to the list
lone axle
I'll look into whether it makes sense to use a deny-list instead of allow as well. Thinking about it now it seems to me there would be a rather short list of files that we wouldn't want to include. .license files being the only that come to mind immediately. Perhaps there were others that came up when it was originally coded. But also possible it was just a consequence of using a dictionary to map extensions to the mock icon images for the different types.
idle owl
Fair enough. I'm happy with whatever you decide is the best route. We can always iterate on it over time as well.
lone axle
@idle owl if you are around still (no worries if not) do you know off hand a learn guide project, or library repo / example that has a CSV in it? I think I've got the change but need to test it.
If not I'll add one temporarily to a project locally and run it there to try it out without committing any changes to the project.
idle owl
<@330227457296957440> if you are around still (no worries if not) do you know of...
Let me check, I think I might have one.
@lone axle I don't. Please add one temporarily to test it. Eva's latest has one, but I have no idea what it's called.
lone axle
Will do, Thank you. CGrover searched some up for me as well to try out just now
idle owl
Excellent!
proven garnet
<@330227457296957440> if you are around still (no worries if not) do you know of...
I'd love to link to one potentially in my circuitpython-csv library!
Also, we're now fully (for real this time) using composite actions for the CI! Feel free to tag or ping me if something seems weird - about ~50 libraries have issues that need to be addressed since we upgraded pylint versions.
I'm trying to make sure PRs that have been run since that fail on them try to get fixes sooner
proven garnet
Catching up on the stream, @lone axle do you want to use the latest Python, or keep it pinned to 3.10? I noticed you got the cryptic formatargspec error - I managed to solve it for Python 3.11 with some upgrades to pylint. If you'd rather the latest Python I'm happy to submit those changes as a PR.
Looks like you got it running on 3.10 which is good.
lone axle
Catching up on the stream, <@382939733107408897> do you want to use the latest P...
I don't have a preference / would defer to same as we are doing for other repos. Someone in the chat mentioned that Learn Guide repo had been reverted to 3.10 for the time being, so I tried it to see if the same change would work for screenshotter.
Have you seen the Black / Click issue that I ran into before? that one was new to me.
proven garnet
Yeah, I have - if I remember correctly that was that click implemented a change where they removed that library and black still referenced it. So it was basically a mismatch between the two.
Basically no version of click was pinned, so their upgrade broke things across anything using black
manic glacierBOT
wispy badger
Hello there, I discovered something strange while doing some testing with CircuitPython today.
I was testing this code I wrote some time ago https://github.com/educ8s/CircuitPython-Pi-Calculation-Benchmark
This script calculates the Pi using an approximation algorithm.
I ported the code in C to run on an Arduino Nano and I noticed something strange.
The execution time of the script was linear depending on how many iterations of the algorithm I was doing.
On CircutPython on the other hand the execution time grew exponentially
and the memory usage grew on each iteration
That's the code of the algorithm:
def calculate_pi(n):
"""
Calculate the value of pi using the Bailey–Borwein–Plouffe formula.
"""
pi = 0
k = 0
while k < n:
pi += (1 / (16 ** k)) * ((4 / (8 * k + 1)) - (2 / (8 * k + 4)) - (1 / (8 * k + 5)) - (1 / (8 * k + 6)))
k += 1
return pi
After a lot of testing I found out the culprit. It is this experession: 16 ** k
I replaced it with math.pow(16,k)
and I now have linear performance and no memory increase at all. Also the speed diference is huge.
For 1500 iterations the scipt needed 6.8sec with the ** expression and only 0.2 sec with math.pow function.
I don't know if it is a bug, I just wanted to let you know.
stuck elbow
might be a good idea to report it as an issue on github
wispy badger
Thanks for the feedback. Report it on the main CircuitPython repository or somewhere else?
stuck elbow
I think the main repository is best
wispy badger
@stuck elbow Thanks, I just submitted a report.
onyx hinge
I was testing this code I wrote some time ago https://github.com/educ8s/CircuitP...
The types of the expressions are different. Python and CircuitPython represent ints exactly, but the memory storage and computation time increases according to the length of the number; while floats always use a constant amount of storage but don't represent values exactly: ```py
type(16**k)
<class 'int'>
type(pow(16, k))
<class 'float'>
you can also write 16.**k so that the power operation is done as a floating point calculation rather than an integer calculation
onyx hinge
<@&356864093652516868> hey all! Our weekly meeting is coming up in a few hours (3ish) but remember to double check your local time, as most of the US switched to winter (standard) time over this past weekend. It's a good time to add your notes to the document! See you a little later today. https://docs.google.com/document/d/1BtwVBaXLXcFHM52SGlMX0-7K0m3r_e-T6l6MLfux5KI/edit?usp=sharing
random junco
Thanks jepler!
onyx hinge
hey @random junco, @turbid radish says that the newsletter is ready to crib from for the meeting notes!
random junco
great, thank you!
wispy badger
The types of the expressions are different. Python and CircuitPython represent ...
Thank you very much for your time!
Very interesting, didn't know this difference\
onyx hinge
Thank you very much for your time!
It is not evident, it's something you have to learn/know about python. Glad I could help.
random junco
@tulip sleet: Would you be willing to speak to the core in the weekly meeting?
tulip sleet
<@329766224093249548>: Would you be willing to speak to the core in the weekly m...
i was out a lot of last week, so @onyx hinge would be better this week (if ok with you, Jeff)
onyx hinge
sure!
random junco
great, thank you!
onyx hinge
as though I know what is going on...
midnight ember
2pm or 3pm?
random junco
Now
onyx hinge
We're a few minutes away from starting the meeting
adafruit folks are wrapping up our internal meeting but no reason for paul not to get started 🙂
random junco
no hurry. 🙂
idle owl
Ah Jeff beat me to it. We'll be there soon!
midnight ember
come on ya scallywags
onyx hinge
(swoon emoji)
random junco
I've lost sound
onyx hinge
🤯 * 206300 PyPI downloads over 322 libraries
gilded cradle
Is Paul still using a browser?
idle owl
@random junco Everyone else's sound is working. Are you not hearing us?
onyx hinge
@random junco do you need someone to take over the meeting?
gilded cradle
It's all good @random junco. We've all lost audio at some point.
idle owl
No worries, @random junco! It's happened to all of us. Seriously.
lone axle
it sounds similar to how my sound cut out the last few meetings at points. I was using the browser to record discord and speak. I've changed over to using the app today🤞
midnight ember
yeah that dev setup stream was awesome
been enthralled with all the great stuff happening at supercon this weekend. really great stuff.
onyx hinge
belated hug report for everyone who joined in during hactoberfest
(cgrover, you seem professional enough to me)
errant grail
(cgrover, you seem professional enough to me)
<blush> Still have a long journey ahead of me.
midnight ember
idle owl
Thanks, foamyguy! 🎉
lone axle
I'm not familiar with the MCP2307 chip, so can't be sure. But the Debouncer library I'm pretty sure has ability to interface with arbitrary functions which should make it possible to use for even non-standard digitalio type stuff.
midnight ember
I'll take another look into the library. Figured it would require access directly to GPIO pins like the keypad library code does.
onyx hinge
@midnight ember do you need to set each of the chips to a separate address by soldering at least one of those address pads?
midnight ember
They're on 2 separate busses so I can get away with using the same address. Didn't know I could chain them if I really needed to. Since designing the PCB figured out I can get 8 chips per bus if needed. Right now have 1 chip per bus and still have 10 unused GPIO... god i love expander chips.
onyx hinge
aha ..
midnight ember
whew Kattni's been busy
ember iris
That's a full plate for sure
idle owl
whew Kattni's been busy
Indeed.
idle owl
That's a full plate for sure
Yeah it's enough for weeks-plural.
onyx hinge
it was early christmas in the internal meeting, in terms of work assigned 🙂
lone axle
This cricket example is the one that shows the function technique: https://github.com/adafruit/Adafruit_CircuitPython_Debouncer/blob/main/examples/debouncer_crickit_test.py it uses lambda to define a function that reads from the cricket library, which is seesaw underneath I think. Different expander chips should be able to set up similarly with functions / lambda that call the expander libraries functions.
idle owl
Oh and I will also finally be taking the time to update the Contributing to CircuitPython using Git and Github guide!
midnight ember
I saw that example but decided to go a different route with enumeration which ended up being a dead end. Will give that a shot thank you.
onyx hinge
excellent work on the composite actions, belated hug report @proven garnet
candid sun
i realized after the meeting began that i neglected to add notes, but just wanted to say group hug to all.
onyx hinge
Thanks Paul!
midnight ember
Thank you Paul for hosting.
idle owl
So greatly appreciated! Thanks so much, Paul!
gilded cradle
Thanks Paul
errant grail
Thanks Paul!
lone axle
Thanks for hosting Paul. Have a nice day everyone! 🎉
candid sun
thanks paul!
errant grail
@random junco , I have many stories that would easily eclipse your experiences. It's all part of the game.
ember iris
Thanks all! Have a great week everyone!
random junco
thanks all for the support. It's appreciated. 🙂
midnight ember
Kattni doing an entire session with only her voice sounds like it takes the cake. big oof. lol. these things happen.
this is why we have people here who record backups, these things happen, kind of regularly, just part of internet life.
idle owl
Kattni doing an entire session with only her voice sounds like it takes the cake...
Yeah it was pretty bad. 😄
I could hear everything, OBS was acting like it could, but only my side came through. 🤷🏻♀️
errant grail
I performed a music gig as a duo with new in-ear monitors. Forgot to turn on the PA for the audience for the first two songs.
midnight ember
in OBS when others are talking you want to see that VU meter going back and forth 😉
idle owl
I performed a music gig as a duo with new in-ear monitors. Forgot to turn on the...
See? Happens to everyone in some way!
@errant grail Don't forget to disconnect from the voice chat to avoid surprises.
errant grail
Yes, and that's just one of many similar experiences.
idle owl
in OBS when others are talking you want to see that VU meter going back and fort...
It was, that was the problem. 🙂
midnight ember
as a DJ who wore headphones, imagine sliding into the next track that is muted. sounds fine in the headphones... also why i'm deaf in 1 ear now.
errant grail
<@339922377746022402> Don't forget to disconnect from the voice chat to avoid su...
And there's another...
proven garnet
excellent work on the composite actions, belated hug report <@135415127879974912...
Thanks! Definitely going to make some necessary changes to the CI much more manageable
random junco
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/1_pM6rZNwdHXKLP69lmqDRBxVVnlMj-TsIRuRjjSCyEU/edit?usp=sharing
random junco
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is...
Ok, I think all the post-meeting stuff is done, just need this PR merged if someone gets a chance: https://github.com/adafruit/adafruit-circuitpython-weekly-meeting/pull/90 - thanks!
idle owl
Ok, I think all the post-meeting stuff is done, just need this PR merged if some...
Done!
random junco
awesome, thank you
trim elm
@proven garnet Hihi. I heard we are working on fixing the pylint stuff and was told to ask you for more info on what specifically the problem is
manic glacierBOT
minor plume
My apologies for missing the meeting today, all. Despite having a block on my work calendar for that time, I was scheduled for TWO conflicting work meetings during that time. 😕
jaunty juniper
@proven garnet I noticed in help-with that the circup instructions for circuitpython_csv are incorrect, likely an effect of: https://github.com/adafruit/cookiecutter-adafruit-circuitpython/issues/207
onyx hinge
@tulip sleet do you know if we have a USB HID descriptor for an absolute pointing device (custom descriptor etc)? (I ask due to the above issue)
proven garnet
<@135415127879974912> Hihi. I heard we are working on fixing the pylint stuff an...
Yup yup! I'll message you about it now
proven garnet
<@135415127879974912> I noticed in help-with that the circup instructions for ci...
Thank you! I'll file an issue and fix it!
thorny jay
<@329766224093249548> do you know if we have a USB HID descriptor for an absolut...
I did something like that some time ago. Doing copy cat of someone else code. There was code in boot.py because I think this is where the description must occurs. Let me search my gist or tweet.
onyx hinge
I did something like that some time ago. Doing copy cat of someone else code. Th...
if you find it can you comment on the issue?
jaunty juniper
I turned it into a library
https://github.com/Neradoc/CircuitPython_Absolute_Mouse
@onyx hinge 👆 (and @thorny jay )
thorny jay
onyx hinge
thank both! I went ahead and put those links in
thorny jay
thank both! I went ahead and put those links in
Great collaborative work. I was typing, so I edited to only keep the added value part.
onyx hinge
I keep meaning to look at the macros and figure out why my mistake want a compile error
amber sundial
Yesterday I successfully followed this guide for Building CircuitPython using Windows 10 (21H2 Build 19044.2130) and Windows Subsystem for Linux (WSL) running Ubuntu. Hooray!! It went pretty smoothly and the guide is excellent. https://learn.adafruit.com/building-circuitpython/build-circuitpython
Some of the steps were slightly different for me though:
https://learn.adafruit.com/building-circuitpython/windows-subsystem-for-linux
Some added comfort:
- WSL 2 is now default with a new install, it didn't need to be seperately enabled
- Moving Files to Windows has become much easier: you can call
explorer.exe .from WSL and copy the files via the GUI - There easy to setup intergration with VSCode following these instructions https://learn.microsoft.com/en-us/windows/wsl/setup/environment
Rust needed to be installed
- in order to install requirements-dev.txt and requirements-doc.txt, I needed to install Rust first (didn't come with Ubuntu):
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh```
You can verify the installation by running:
```source $HOME/.cargo/env #Import the environment config for rust
rustc --version```
https://learn.adafruit.com/building-circuitpython/linux
**Some unzip weirdness**
- in WSL2, this only worked without errors using sudo: `sudo tar xvf <name of the .bz2 or .xz file you downloaded>`
**.profile and .bashrc**
- .bash_profile is just .profile in WSL2
- path format needed to be `export PATH="$HOME/bin/gcc-arm-none-eabi-10.3-2021.10/bin:$PATH"`
- in order for the paths to work I had to restart WSL2 Ubuntu
I ended up adding these three lines to .profile and .bashrc:
```. "$HOME/.cargo/env"
export PATH="$HOME/bin/gcc-arm-none-eabi-10.3-2021.10/bin:$PATH"
export SETUPTOOLS_USE_DISTUTILS=stdlib```
Maybe one or two of these things might be worth adding to the guide. Other than that, the build process went smoothly and the .uf2 work on the boards I had. 🍩 🍩 🐘
tulip sleet
Yesterday I successfully followed this guide for Building CircuitPython using Wi...
Thank you for the detailed feecback! I've recorded it in the Learn Guide with the "Feedback" link.
tulip sleet
Yesterday I successfully followed this guide for Building CircuitPython using Wi...
Some unzip weirdness
- in WSL2, this only worked without errors using sudo: sudo tar xvf <name of the .bz2 or .xz file you downloaded>
that seems like you were in a non-user directory when you tried to un-tar? Could you go to your home directory and tar xvf without sudo? Or maybe you downloaded with different permissions?
amber sundial
that seems like you were in a non-user directory when you tried to un-tar? Could...
it was in the bin directory, but I had to create that
tulip sleet
/usr/bin ?
amber sundial
yes
tulip sleet
or ~/bin ?
amber sundial
that seems like you were in a non-user directory when you tried to un-tar? Could...
let me try this real quick, I have it open
tulip sleet
ah yes, you shouldn't try to unpack there, it's for sudo apt-installed stuff. Instead you can create ~/bin, say, and then add that to your PATH. I'll make this clearer: I was assuming the user was more famliar with Linux directory layout
regular users should not write in /usr/bin or /bin, etc
amber sundial
ah I just put the .tar.bz2 there because that's where I wanted the files to end up
FWIW it worked even if unorthodox
tulip sleet
it did work, but the danger is accidentally overwriting an existing system file, so I'll make clear to put it in a directory under your home directory
amber sundial
OK that makes sense
tulip sleet
analogy is that you don't want to put stuff in the C:\WINDOWS folder
again, thanks for the feedback, it is very helpful, because I make unknowing assumptions.
amber sundial
regular users should not write in /usr/bin or /bin, etc
ah hang on, you were right, the file is in ~/bin
that's why I had to set the path to PATH="$HOME/bin/gcc-arm-none-eabi-10.3-2021.10/bin:$PATH"
I'm just discovering the various bin directories
fun 😀
learning has occurred
lone sandalBOT
proven garnet
@jaunty juniper I think I fixed the installation instructions for the cav module and confirmed using circup.
Thanks again!
onyx hinge
aha the voltage still seems a bit low because it's one diode below VUSB, albeit with the very low forward voltage MBR120VLSF. So 4.8 is plausible.
midnight ember
any idea who does the graphics for pretty pins for the boards in product pages? i'd like one made for the MCP23017 learn guide.
idle owl
any idea who does the graphics for pretty pins for the boards in product pages? ...
We don't do it for sensors/breakouts/etc. We do them for microcontrollers. If there are any on sensors/breakouts, they were generated by someone else most likely and included following that.
midnight ember
i mean for the bare chip because adafruit does sell that and the way the library uses it is different than most pinout diagrams you find on it
idle owl
Right, I'm saying we don't do PrettyPins for chips, we do them for microcontroller boards.
It's not designed to work with chips as it is.
midnight ember
ok can i recommend a graphic for the MCP23017 learn guide because the pinouts are very confusing to use the first time. they start on the right side of the chip.
onyx hinge
the software "prettypins" is https://github.com/adafruit/PrettyPins fwiw
midnight ember
manic glacierBOT
manic glacierBOT
tulip sleet
@onyx hinge could I have an audio chat with you about translations and the compression mechanism?
onyx hinge
<@426200547024961539> could I have an audio chat with you about translations and...
sure! give me a couple minutes to transition to my computer with good mic.
OK I'm ready @tulip sleet
make BOARD=adafruit_proxlight_trinkey_m0 TRANSLATION=fr
https://github.com/adafruit/circuitpython/pull/7181/commits/f5b20a3ef0917d62cc836356c24649d611098b9d
tulip sleet
Plus d’un tiers
onyx hinge
Although ubiquitous in typeset material, the typographic apostrophe ( ’ ) is rather difficult to enter on a computer, since it does not have its own key on a standard keyboard. Outside the world of professional typesetting and graphic design, many people do not know how to enter this character and instead use the typewriter apostrophe ( ' ). The typewriter apostrophe has always been considered tolerable on Web pages because of the egalitarian nature of Web publishing, the low resolution of computer monitors in comparison to print, and legacy limitations provided by ASCII.
tulip sleet
0000000 P l u s d 342 200 231 u n t i e r 0000020 s \n 0000022
jaunty juniper
yeah, nobody uses it
tulip sleet
I just pasted that from a lemonde.fr article
but you don't type it on the keyboard, you are saying, so when does lemonde fancy it up
jaunty juniper
oe is also good enough for œ, likewise it's annoying to type on a keyboard
tulip sleet
does the Academie still care?
jaunty juniper
there's setups in software like Word, and web CMS to convert automatically
(and for example putting a non-breaking space before : as is required by the typographical rules)
tulip sleet
is there a provision in weblate for remapping certain character sequences?
onyx hinge
En français, l’utilisation de la ligature œ n’est pas esthétique mais linguistique, contrairement à d’autres ligatures (par exemple, fi) : œ et oe ne se lisent pas de la même manière, la deuxième forme notant un hiatus (cœlacanthe ≠ coefficient).
french wikipedia says that the ligature is "not aesthetic but linguistic"
tulip sleet
i was giving @onyx hinge the example of ligatures being important in at least some languages, and offensive if not used, like in English, w could be vv but everybody would say that's weird
jaunty juniper
I don't think we should aim at the typographical quality of prestigious newspapers
also it's mis en œuvre for implemented when implémenté is used somewhere else
onyx hinge
I'm sure the consistency of the translations could be better!
jaunty juniper
(borrowed from the english, it has entered the French language)
onyx hinge
which is .. better ?
tulip sleet
feel free to modify the french translation if you have ideas about that
onyx hinge
or do a corsican translation 🙂
jaunty juniper
yeah
by the way, things not to do:
mp_raise_msg_varg(&mp_type_RuntimeError, translate("%q is %q"), MP_QSTR_storage, MP_QSTR_extended);
worth noting that in an issue, say
jaunty juniper
yeah I'm going through untranslated strings and taking notes
there's Expected a %q in many places and one Expected an %q in alarm 🤔
tulip sleet
could be Expected %q
onyx hinge
locale/circuitpython.pot:msgid "Expected a %q or %q"
locale/circuitpython.pot:msgid "Expected an %q"
those may be convertible to mp_obj_t mp_arg_validate_type(mp_obj_t obj, const mp_obj_type_t *type, qstr arg_name)
which will say mp_raise_TypeError_varg(translate("%q must be of type %q"), arg_name, type->name);
I can put back the oe with the single offset block now
jaunty juniper
why not, but I'm removing the one occurence from the translation anyway 😉
onyx hinge
good, thank you
jaunty juniper
I might be increasing the translation size by adding translated strings and fixing some stuff 🤫
orchid basinBOT
jaunty juniper
and I logged in with my github account to discover that I already made a weblate account, and to associate the old one with my github I delete the other one, so now the first changes I made today are credited to "deleted" 😛
onyx hinge
thanks @ deleted for your first contribution to circuitpython!
jaunty juniper
does that need a translate ?
for (size_t i = 0; i < COMMON_HAL_MCU_PROCESSOR_UID_LENGTH; i++) {
mp_cprintf(&mp_plat_print, translate("%02X"), raw_id[i]);
}
(is there an advantage, like compressing ?)
onyx hinge
@tulip sleet am I wrong to expect a subclass of aesio.AES to work does subclassing built in types work generally? ```import aesio
b = bytearray(16)
a = aesio.AES(b'1' * 16)
print(a.mode)
a.encrypt_into(b'a' * 16, b)
print(bytes(b))
print()
class A(aesio.AES): pass
a = A(b'1'*16)
print(a.mode)
a.encrypt_into(b'a' * 16, b)
print(bytes(b))
I started by removing the "I thought it was unneeded" type check at the top of several of the methods, but then the encryption is just .. not even performed and the mode is not set...
1
b'\xc6\xb9\xd1\x02\xdd\xea\xb1i\xf3$\xcb\xdb\xe73\xf0\n'
0
b'aaaaaaaaaaaaaaaa'
yes that translate() apparently saved space, it was part of a crop of space savings that dan committed about a month ago
tulip sleet
it is really weird that it made a difference; I think it may have to do with how the surrounding code was compiled and optimized
gilded cradle
Could I get a review on https://github.com/adafruit/Adafruit_Python_PlatformDetect/pull/257 ? It's holding up other PRs.
tulip sleet
<@329766224093249548> am I wrong to expect a subclass of `aesio.AES` to work doe...
I thought the native C code had to anticipate subclassing and doing special lookups, e.g., see examples in displayio
jaunty juniper
as far as I know subclassing builtin types is complicated because you can't call their dunder functions like __setitem__ since they are not exposed as properties of the class, and other side effects like that
onyx hinge
@gilded cradle looking .. does it matter that _LIBRE_COMPUTER_IDS is just a string, but you're using 'in' on it? It might give an unexpected result, say if self.id was a short string like C?
gilded cradle
Let me check...
onyx hinge
-_LIBRE_COMPUTER_IDS = AML_S905X_CC
+_LIBRE_COMPUTER_IDS = (AML_S905X_CC,)
```maybe you want/need something like this.the removed function property wasn't considered part of a public API?
gilded cradle
No, I already changed it in Blinka
gilded cradle
Perfect, thanks
cedar veldt
Sorting the GitHub issues by "Recently updated" seems to be giving strange results recently? There are quite a lot of issues near the top where I can't see anything having been updated.
random junco
Maybe because the hacktoberfest label was removed?
gilded cradle
I think recently updated means any update such as a comment, label change, or whatever.
cedar veldt
Maybe because the hacktoberfest label was removed?
I've seen that before, but it normally says when a label was added/removed in its update history
gilded cradle
Which repo?
cedar veldt
gilded cradle
It looks like the top issues were added to a specific milestone.
cedar veldt
Hmm, adding to a milestone was also shown in the update history before. But maybe GitHub isn't showing that any more.
e.g. https://github.com/adafruit/circuitpython/issues/4303 says in the index it was updated 12 hours ago, was added to a milestone on Mar 2, 2021, and had hacktoberfest label added Sep 29
gilded cradle
Ok, good example. I'm not sure why it's showing updated 12 hours ago.
Possibly a comment was added then removed
cedar veldt
There's about 15 of them at the moment
tulip sleet
the most recently updated ones all have a Hacktoberfest label, so seems related to that
except for 3 actual new ones
cedar veldt
Maybe the Hacktoberfest label itself was edited?
tulip sleet
yes, maybe that was it, though I don't know what it would have been. I think it was already capitalized. Maybe the description was changed.
cedar veldt
Could be the description. I think the same thing happened a week or two ago.
tulip sleet
I'm looking in the audit log for the organization, but don't see anything immediately. However, it has tons of stuff in it
i think I will give up for now
manic glacierBOT
onyx hinge
@tulip sleet do we have some boilerplate about please don't add frozen modules? frozen modules that were not added as submodules are the reason https://github.com/adafruit/circuitpython/pull/7135#issuecomment-1309356235 is failing. I'd like to recommend the submitter not freeze the modules, because I think that's our standard line.
tulip sleet
<@329766224093249548> do we have some boilerplate about please don't add frozen ...
There are some folks who do want to freeze modules, based on the use case of their board (e.g., for education). So we don't recommend it in general, but will defer to their judgment.
In this case, if this is a badge for a conference or whatever, frozen might be appropriate.
onyx hinge
they are having enough trouble with github that I'd have to step in and do all the submodule stuff 😕
which I guess is fine, I shouldn't grump about it
tulip sleet
well... I think they should learn how to do it: https://learn.adafruit.com/building-circuitpython/adding-frozen-modules
if you're willing to do it, OK, but first maybe ask about whether they really want the frozen modules. Let me find another discussion of that.
tulip sleet
@onyx hinge https://github.com/adafruit/circuitpython/pull/6526 is all I could find
glass kite
Hello everyone 👋 ! I just joined the server so hopefully this is the right place to ask my question
stuck elbow
you won't know until you ask it
glass kite
I have a good few hundred neopixel LEDs and I was wondering if I could write the code for my project on my PC and control GPIO pins on the raspberry pi that's on my local network.
@stuck elbow sorry man took me a while to write it 🤣
stuck elbow
that's more of #help-with-circuitpython question
glass kite
aight, will ask it there, thanks!
stuck elbow
or even #help-with-linux-sbcs
since you don't necessarily have to use circuitpython for that
glass kite
yeah I mean I wouldn't mind switching to arduino at all, as long as it works
it's actually quite funny because the library is called adafruit-circuitpython-neopixel, so... yeah, I'm sure I'm not the only one who's a bit confused 🤣
stuck elbow
no problem at all, it's just that you are likely to get better answers there
cedar veldt
I was wondering about that. The CircuitPython repo has a custom system for getting submodules, but it still works to add more using git submodule add as in the guide?
manic glacierBOT
Actually this test-setup works fine with the build in board.I2C()
With this code:
And it actually worked 5333 seconds without a error!
This is the CP-Version I am using.
proven garnet
Is there a list or place I can find all the different bootloader mode names for boards (like QTPYBOOT)? Looking to write a program that helps with loading firmware but need to know what to look for.
spiral elk
You'd find most of them in the tinyuf2 repo, but some boards have a hardware uf2 or other bootloader and wouldn't be found there.
jaunty juniper
and tinyuf2 is only for ESP and STM, there are other repos for atmel and NRF
or you can use discotool 😉
jaunty juniper
Is there a list or place I can find all the different bootloader mode names for ...
how exactly do you want to do it ?
you can identify a BOOT drive by the presence of INFO_UF2.TXT in it, which will also contain basic information about the board ID, but of course on RP2040 there is no way to find the board from that
spiral elk
I don't think there'd be any way at all at the bootloader level to distinguish between rp2040 boards
jaunty juniper
yep
discotool tries to provide an identifier for the USB port that the board is connected to, so that you can have a Circuitpython board, make it reset in UF2 mode and find it again by the port it's connected to
but that assumes you already had CP on it
proven garnet
how exactly do you want to do it ?
I was actually going to borrow the same code that circup uses, and just add a parameter for those specific names instead of just CIRCUITPY.
spiral elk
with rp2040 you could read out the contents of flash and try and make some sense of it somehow.
proven garnet
or you can use discotool 😉
Oh, does discotool already do this?
spiral elk
since the rp2040 bootloader allows you to extract a uf2 image off of the device.
jaunty juniper
discotool doesn't care about the name CIRCUITPY, if finds drives connected to microcontrollers, which are defined as: USB devices with a serial port OR USB devices with a VID in the list of known VIDs
there is:
❯ picotool info
Program Information
name: CircuitPython
version: 8.0.0-beta.2-9-g5192082e6
web site: https://circuitpython.org
features: double reset -> BOOTSEL
that's not good enough 😦
spiral elk
You could relatively easily use picotool to read the non-CIRCUITPY flash contents and compare hashes with a list of official releases, but that would only work for official builds.
proven garnet
Ahhh, discotool is likely what I want, thanks for the tip!
jaunty juniper
here is the code I wrote for auto installing UF2s (updating CP) and code.py on CP boards, but I didn't actually use it so it's more like an idea:
https://gist.github.com/Neradoc/64795d4c9b7b5ecd12217b773fdbad92
proven garnet
here is the code I wrote for auto installing UF2s (updating CP) and code.py on C...
Ah, perfect, I'll take a look for reference
manic glacierBOT
oblique iron
Just shouting out that I am the one that submitted this, in case it comes up for discussion on what might need to change.
manic glacierBOT
tulip sleet
Just shouting out that I am the one that submitted this, in case it comes up for...
FYI, the usual reviewer for the nrf52 bootloader repo is out for a few days
oblique iron
FYI, the usual reviewer for the nrf52 bootloader repo is out for a few days
appreciate the heads up. this isn't super time sensitive
brazen hatch
is there any way for me to build circuitpython with debug symbols and actually do meaningful debugging on picow?
The pins are there, I have a pi400, I can debug it with native gpio.
I have used gdb like 2 times in my life, but I really need to extract data if i'm going to implement the picow-ap
I have openocd configured and ready
I have tested it with a pico, and I can halt/reset it
but I don't know how to actually debug circuitpy with it
tulip sleet
is there any way for me to build circuitpython with debug symbols and actually d...
I have not tried this on a pico w; @onyx hinge might have. I do plenty of gdb debugging on atmel-samd boards. you can do a lot of printf-style debugging:
mp_printf(&mp_plat_print, "i is now: %d\n", i);
onyx hinge
No, I haven't used gdb debugging with pico / pico w lately. I mostly do printf debugging
brazen hatch
Alright, thanks, I will stick to printf spam!
onyx hinge
You have several sub problems:
- getting gdb to talk to your hardware
- learning the restrictions (such as small number of breakpoints) of the debugger implementation
- getting useful debug information into firmware.elf
For the former, you should be able to follow instructions from raspberry pi and learn about any caveats from them.
For the latter, we never strip circuitpython by default so you should have function-level information. you can make DEBUG=1 to include additional debug information such as line-level information (equivalent to gcc -g3); you must make clean and then consistently specify make DEBUG=1 or things get messed.
but yeah it's not how I work so I can't give you 'in the trenches' info
Verifying that debug info is present in firmware.elf: $ arm-none-eabi-gdb ports/raspberrypi/build-raspberry_pi_pico_w/firmware.elf GNU gdb (GNU Arm Embedded Toolchain 10-2020-q4-major) 10.1.90.20201028-git … (gdb) list *main 0x10026f3c is in main (../../main.c:916). 911 return exit_code; 912 } 913 914 int __attribute__((used)) main(void) { …
brazen hatch
doesn't sound ver complicated tho, I will try it, and if it ends up working well, I will do a writeup
I have openocd working already with rp2040
brazen hatch
those picow debug pins were such a pain to solder
and its gonna be an even bigger pain to clean
tulip sleet
those picow debug pins were such a pain to solder
did you solder them underneath?
no real reason to clean if it's just flux
brazen hatch
no real reason to clean if it's just flux
mine is conductive for some dumb reason
cleaning
brazen hatch
mine is conductive for some dumb reason
had 3 boards crashing due to flux bridging stuff, now I am kind of a maniac with the flux removal sprays and stuff
onyx hinge
oof, I'd discontinue using that flux!
orchid basinBOT
tulip sleet
what kind of flux is that?
brazen hatch
tulip sleet
I am using leaded rosin-core solder for non-SMD work still
that kind of flux is for soldering pipes and stuff
brazen hatch
???
uhh what?
tulip sleet
don't use it on electronics. It's acidic and will eat away at your connections
did you buy it at the hw store?
brazen hatch
literally at my microcontroller retailer
brazen hatch
DON'T USE IT. do a search for zinc chloride flux
over 10 boards are soldered with it already
painn
tulip sleet
clean them really well but if there's flux inside the connections, you may have trouble later
use rosin core flux or other flux marked for electronics soldering, and tell your store not to sell that stuff
it's for plumbing, jewelry, stained glass, etc, NOT for electronics
brazen hatch
Alright, thanks a lot man! I would have never known..
tulip sleet
shame on that store
brazen hatch
This explains as to why its also conductive..
tulip sleet
oh man, glad we caught this early
tulip sleet
This explains as to why its also conductive..
what kind of solder are you using? Most electronics solder has a rosin core built into the solder wire
brazen hatch
what kind of solder are you using? Most electronics solder has a rosin core buil...
dis, its a simple lead-free I got from the same store
tulip sleet
can you find a link for that?
brazen hatch
the leaded stuff smell nasty
tulip sleet
you should buy solder that's specifically marked for electronics
no part number on that pic; is there another label?
tulip sleet
that company sells a lot of odifferent kinds of solder
brazen hatch
maybe we should go to a different channel as to not pollute the dev channel
tulip sleet
go to hw design yah
thorny jay
DON'T USE IT. do a search for zinc chloride flux
Thank you, I have been using that since a bit of time... Found it in a hardware store and I was super happy to finally find "Flux" as before that I was using none.
Now I need to remember and check the board where I have used that... 😭
tulip sleet
these fluxes should be labeled "NOT FOR ELECTRONIC USE". sad that they are not
brazen hatch
Well I mean mission success
alright someone left broken debug code in picow radio
it doesnt compile with DEBUG=1
more fixing to do.. I will continue tomorrow
onyx hinge
It was probably me. feel free to submit a fix. copypaste code from lwip seems to use the DEBUG macros as well, and the debug messages are just broken
lone sandalBOT
midnight ember
don't feel bad @brazen hatch I did the same thing when I first started. Was using plumber flux for soldering copper pipes commonly found in hardware stores. Started noticing all the joints on my boards looked like they were dissolving in acid after 3 months, because they were. Joints looked like they were 30 years old not 3 months old. Got some chip-qwik flux and everything good since. Still need to clean with alcohol though after use.
brazen hatch
don't feel bad <@614096041083404300> I did the same thing when I first started. ...
Huh? My boards look nice and shiny somehow..
They have been soldered for over 5 months now..
To it's defence though, I am literally spamming Kontakt U (it is a spray, not a kde app) and distilled water.
I do 3-4 passes till it's perfectly shiny and all the laser etchings are visible.
midnight ember
it's very possible at the time I wasn't cleaning the flux off. :/
midnight ember
oof
brazen hatch
Though all were revived after cleaning
some were powering off after like 40s
some did random reboots
some reset
midnight ember
kinda same, i was having all kinds of weird issues. posted on the forums and someone helped me learn the error of my ways.
brazen hatch
and stayed with reset.
Anyways, now I have one morbiliion boards cuz of this
All were fixed with like 3€ worth of sprays and a toothbrush
midnight ember
https://forums.adafruit.com/viewtopic.php?t=155352 2019, i've learned a lot since then
if you're not getting oxidation on your joints then probably nothing to worry about it. the amount of scale buildup was visible and obvious.
i'll have to look into deoxit stuff. i have some but have been using 99% IPA, probably does about the same thing.
brazen hatch
All yours
blissful pollen
@onyx hinge For the NeoPxl8 was it designed (or testing done) on less then 8 strands? Noticed the PIO program always defaults to 8 despite what you set num_strands to. Looking at using it (for the PIO) revamping a project I have but not going to (yet) resolder things to multiple pins
onyx hinge
@blissful pollen yeah I bet it is only exactly 8 strands right now, pull request welcome 🙂
I was noticing that a qtpy rp2040 has 5 consecutive pins available so .. it'd be nice if that worked!
blissful pollen
<@228322543507865601> yeah I bet it is only exactly 8 strands right now, pull re...
Thanks, just wanted to check before I got into it, just finding space for 300 pixels on my desk first
onyx hinge
NeoBFF5
blissful pollen
That would be cool.
One other quick question is the pixel order still strand0.0,1,2,3... stand1.0,1,2,3... or stand0.0, strand1.0, strand0.1, strand1.1, etc. (if that makes sense)
onyx hinge
pixels[1] is strand 1 pixel 0
and pixels[8] (if you have 8 strands) is strand 0 pixel 1
in the one example you can see how to construct pixel maps that give you back the individual strands
blissful pollen
okay cool thanks. And I just loaded that example up
My past project used adafruit_pixel_framebuf and I think should just be able to take advantage of NeoPxl8 with the map
onyx hinge
that would be very cool!
speaking of which, I'm supposed to do a small text scroller for the neopixel bff (5x5) -- did you have font code that might help me out?
manic glacierBOT
@jepler Many thanks for your help! I tried to fix it yesterday at my local build, but without success. Do you have more info about that script to fix this issue? I saw this information at page, which you linked earlier, but I did not figure out how to make it work.
It would be good to mention that script in the learn guide, so peple will not strugle to fix same issue.
blissful pollen
Yes I do, let me see if its on Github
blissful pollen
speaking of which, I'm supposed to do a small text scroller for the neopixel bff...
https://gist.github.com/gamblor21/25a94e286eb95047157c434ff1eab32b
It was fast enough on a ItsyBitsyM4. If the BFF helper is 5 high something from the IS31FL3741 project may help the glasses were only 5 high (but for that it was a core update I made for speed)
onyx hinge
yeah this is strictly 5x5 so it'll display about one letter at a time
I was joking about the "neobff5" (hypothetical product for 5 LED strips in bff form factor) but the 5x5 "neopixel bff" product is real
blissful pollen
Yeah I did basically realize that (though not really familiar with the 5x5).
blissful pollen
I am curious the speed diff for 300 pixels if it is say 1 vs even 2 data lines.
onyx hinge
❤️ @tulip sleet thank you for making that quick improvement on the guide!
tulip sleet
Some folks here mentioned they seemed to need rust to install minify_html, which is listed in requirements-dev.txt. But it's a Python-only package, as far as I can tell. I don't have rust installed and I am able to use the package.
under what circumstances does it need rust, or am i misunderstanding?
onyx hinge
@tulip sleet it probably depends whether there is a wheel available for your system. The description on github says "A Rust HTML minifier meticulously optimised for speed and effectiveness, with bindings for other languages."
so it's probably affecting m1 mac folks for instance, but not linux folks?
onyx hinge
https://pypi.org/project/minify-html/#files there are only wheels for x86_64 macosx, not arm macosx
tulip sleet
the user in q needed it for WSL2, which is a bit strange
onyx hinge
they have wheels for 3.8 through 3.11 manylinux x86_64 which should cover wsl2 unless it's a very old or very new python
part of me thinks we could investigate other minifiers that don't depend on native modules, but .. for most users it's OK. I've never tried to install rust on my own system fwiw
I thought the first report came from a mac person but maybe I jumped to a conclusion or maybe there are multiple people hitting trouble
tulip sleet
part of me thinks we could investigate other minifiers that don't depend on nati...
I have a guide feedback for WSL2.
its main virtue is that it's faster, not sure it matters given how little we have
anyway I am documenting the rust thing for now
onyx hinge
https://github.com/adafruit/circuitpython/pull/7160 rose also(?) ran into it. I was thinking she was on mac but she didn't say.
I think you can also end up with a python/pip setup that's not capable of downloading wheels even if they're available? "Support is offered in pip >= 1.4 and setuptools >= 0.8." .. but those are both super old versions now, setuptools is over 60 and pip is over 20 in versions now
onyx hinge
https://gist.github.com/gamblor21/25a94e286eb95047157c434ff1eab32b
It was fast ...
What "core update for speed" are you referring to?
jaunty juniper
there was that person that mentioned rust #circuitpython-dev message
onyx hinge
jaunty juniper
- in order to install requirements-dev.txt and requirements-doc.txt, I needed to install Rust first (didn't come with Ubuntu):
onyx hinge
@blissful pollen did you specifically have a 5 pixel tall font or did you use a bigger font via the optional 3x scaling?
blissful pollen
What "core update for speed" are you referring to?
Sorry I meant the IS31FL3741_FrameBuffer module vs the pure py thon one
blissful pollen
<@228322543507865601> did you specifically have a 5 pixel tall font or did you u...
I think I used a 15 pixel and scaled it. There was a tomthumb.bdf (i think?) font PaintYourDragon had that was 5 pixels. let me check
blissful pollen
<@228322543507865601> did you specifically have a 5 pixel tall font or did you u...
https://github.com/apparentlymart/ledpixels
It is 4x6 (but can ignore the top or bottom line). I think the original was an Arduino font.
https://github.com/adafruit/Adafruit-GFX-Library/tree/master/Fonts
Has the font in Arduino form
onyx hinge
cool, thanks for the links
manic glacierBOT
cedar veldt
Hi, I was wondering about drive strength on RP2040. Not found anything about it in the source code yet.
Datasheet says "The GPIOs on RP2040 have four different output drive strengths, which are nominally called 2, 4, 8 and 12mA modes"
tulip sleet
Hi, I was wondering about drive strength on RP2040. Not found anything about it ...
we set it to 12ma in DigitalInOut and UART
cedar veldt
What about PulseOut and PIO?
tulip sleet
it is not set in those
digital/DigitalInOut.c line 107 and busio/UART.c line 135
If you'd like to make a PR, that would be great
cedar veldt
Do you mean it should be 12mA for everything?
tulip sleet
I'm looking up some PR's
tulip sleet
Do you mean it should be 12mA for everything?
in general we set it to strong by default at least for DigitalInOut. The UART change was made as part of RS485 support, it looks like.
Limor's philosphy is to set it to strong by default, makes for fewer support problems. .e.g https://github.com/adafruit/circuitpython/issues/1270#issuecomment-463457634 (on nRF)
cedar veldt
I see the default is 4 mA
I have some IR things on PulseOut and PIO where 12mA would be plenty but 4 is perhaps not quite doing it
And a visible LED with PWM that might be dimmer than expected?
tulip sleet
yes, if there's no API to change it, it should be the max. There is a bit of an exception on the nRF chips, where high drive strength on pins near the radio can cause interference, so they don't recommend it.
but we know about that
last forum
Hey, there’s a really interesting confluence going on right now. The graphical, sound-focused programming language plugata just gained the ability to compile into Python. That puts it in sniffing distance of working on any CircuitPython board, which means that DIY synth builders can build on embedded platforms and, probably most excitingly, it’s an editor that can reach a whole new kind of brain for any digital, embedded project.
(Sorry to interrupt. I’m really excited.)
cedar veldt
digital/DigitalInOut.c line 107 and busio/UART.c line 135
Ah yes I see it. Tried searching for "drive" or "strength", but the comment says 'Turn on "strong" pin driving' 😉
last forum
Also, I didn’t know you could switch current supplies on pins and that’s really cool.
cedar veldt
Also, I didn’t know you could switch current supplies on pins and that’s really ...
There isn't a CircuitPython API for it (yet?) so CircuitPython needs to decide what to set it to. And they were saying maximum is best except where there's a particular reason.
Higher drive strength than needed isn't normally a problem as long as your circuit is OK, while lower than needed can mean that stuff doesn't work.
last forum
Higher drive strength than needed isn't normally a problem as long as your circu...
It’s a normal mA capacity, right? So you’re saying it’s probably better to design the peripheral circuit to not draw too much current so you don’t just starve your peripherals by accident. Still, built-in current limiting seems like it could matter in the wild.
plugdata x CircuitPython
mystic flume
Hi all, I'm currently looking atulab: (https://github.com/v923z/micropython-ulab); it's got a bunch of extra functions in python files in the snippets directory. These aren't automatically imported when circuitpython is built, but I think they could be really useful. I've tried setting FROZEN_MPY_DIRS to the snippets directory, but I just get errors with relative imports, however, the bin file is about 18k bigger, which sounds about right. ere to
I've tried also creating a separate frozen dir and creating a separate ulab dir there with it's own __init__.py, but that doesn't seem to import at all well
Any suggestions?
When I say the second attempt didn't import I mean that I couldn't see any of the snippets I was hoping would be included
mystic flume
https://github.com/timothyschoen/PlugData
Ooooo shiny
last forum
@mystic flume RIGHT?
It’s got 20 years of codebase, but Timothy’s work on it from just the last two years turning into his fork is monumental. It’s like he’s made a whole new thing in two years with all the benefits of people collaboring on it since 1992 or something.
Holy crap 1992 is 30 years ago, not 20. I keep forgetting.
mystic flume
Holy crap 1992 is 30 years ago, not 20. I keep forgetting.
I'm sure 1992 was a couple of years ago.
jaunty juniper
so I bought 2 pybstick boards, one is RP2040 with 1MB of flash and the other is a ESP32C3 with 2MB of flash, dunno if I should bother trying to add them to CP with those limitations...
last forum
I just checked: it was first released in 1996, so at least that’s only 26 years ago.
mystic flume
so I bought 2 pybstick boards, one is RP2040 with 1MB of flash and the other is ...
I'm just building CP for the Pico
comes in at 719kb flash usage for circuitpython
so you've got about 300k-ish of space for the flash drive etc - enough to play with
jaunty juniper
yeah that's what it has in MP
cedar veldt
It’s a normal mA capacity, right? So you’re saying it’s probably better to desig...
Right, say if you have a LED and you want 8mA, and the pin has the 12mA drive strength, you just choose your resistor get 8mA as normal.
jaunty juniper
(they ship with MP)
mystic flume
Dunno about ESP32 - I presume needs more space for bluetooth/TCP/IP stack?
last forum
Right, say if you have a LED and you want 8mA, and the pin has the 12mA drive st...
Makes sense! Why would you want to limit on the μC end? Low part count?
jaunty juniper
ah no I think the C3 is 4 MB, I misread, that's more standard at least
cedar veldt
Makes sense! Why would you want to limit on the μC end? Low part count?
To reduce chance of damage in case of mistakes I guess.
Including rules about "total current drawn from all GPIOs"
Limiting the current for your circuit by using the drive strength doesn't seem like a good idea, since it's not that well-defined
mystic flume
Here's some info from broadcom about the raspi:
Two reasons:
1. The raspberry-Pi 3V3 supply was designed with a maximum current of ~3mA per
GPIO pin. If you load each pin with 16mA the total current is 272mA. The 3V3
supply will collapse under that!
2. Big current spikes will happen especially if you have a capacitive load. That will
"bounce" around all the other pins near it. It is likely to cause interference with the
SD-card or even the SDRAM behaviour.
What is a safe current?
All the electronics of the pads are designed for 16mA. That is a safe value under which you
will not damage the device. Even if you set the drive strength to 2mA and then load it so
16mA comes out that will not damage the device. Other than that there is no guaranteed
maximum safe current.```Unfortunately seems like a lot of data sheets don't seem to state max current source/sink per pin these days
I'm looking particularly at the nordic chips here Grrr
cedar veldt
So it is about interference. Not totally straightforward then.
onyx hinge
well that's weird. I tried reactivating WFI in port_idle_until_interrupt on rp2040. My picow starts and resets just fine but either timekeeping or sleep() time are messed until I do something like write to CIRCUITPY! my main loop looks like this: while True: pop.frame() pixels.show() time.sleep(.10) i += 1 t1 = adafruit_ticks.ticks_ms() diff = adafruit_ticks.ticks_diff(t1, t0) if diff > 1000: print(f"{i * 1000 / diff:.0f}fps") t0 = t1 i = 0 and before any change it consistently prints 10fps. This makes sense because the sleep of .10 seconds is huge compared to the computation time. But right after the change, it prints 7FPS! until I write to CIRCUITPY, and then it prints 10FPS! ```diff
diff --git a/ports/raspberrypi/Makefile b/ports/raspberrypi/Makefile
index 5095c4c1f8..7f9a3a09fe 100644
--- a/ports/raspberrypi/Makefile
+++ b/ports/raspberrypi/Makefile
@@ -127,6 +127,7 @@ INC +=
-I../shared/timeutils
-Iboards/$(BOARD)
-Iboards/ \
- -isystem ./../../lib/cmsis/inc
-isystem sdk/
-isystem sdk/src/common/pico_base/include/
-isystem sdk/src/common/pico_binary_info/include/
diff --git a/ports/raspberrypi/supervisor/port.c b/ports/raspberrypi/supervisor/port.c
index 781d2b11d2..7a5ed9ccb4 100644
--- a/ports/raspberrypi/supervisor/port.c
+++ b/ports/raspberrypi/supervisor/port.c
@@ -72,6 +72,9 @@
#include "supervisor/serial.h"
+#include "tusb.h"
+#include <cmsis_compiler.h>
+
extern volatile bool mp_msc_enabled;
STATIC void _tick_callback(uint alarm_num);
@@ -269,10 +272,9 @@ void port_interrupt_after_ticks(uint32_t ticks) {
void port_idle_until_interrupt(void) {
common_hal_mcu_disable_interrupts();
- if (!background_callback_pending()) {
-
// TODO: Does not work when board is power-cycled. -
// asm volatile ("dsb 0xF" ::: "memory"); -
// __wfi();
- if (!background_callback_pending() && !tud_task_event_ready()) {
-
__DSB(); -
}__WFI();
common_hal_mcu_enable_interrupts();
}
@onyx hinge given that reset_cpu() is not really a reset, I don't see a way to prevent a single instruction window that would lose an interrupt. Maybe a restructuring of when the alarm pin interrupts are actually set up would help, but it's kind of a major change. The PR is still much better than the previous situation.
onyx hinge
this is with the serial repl open
onyx hinge
well that's weird. I tried reactivating WFI in port_idle_until_interrupt on rp2...
It's definitely sleeping over-long.
lone sandalBOT
lone sandalBOT
analog bridge
whoops! 😅
manic glacierBOT
manic glacierBOT
digital shoreBOT
adafruit
