fstrings should work exactly the same way .format does (same format mini-language). The % formatting notation is a holdover from Python 2, which is a holdover from C, and unfortunately persists in the stdlib logging module as well.
#circuitpython-dev
rough rapids
But as @jaunty juniper is suggesting the error message seems like it comes from the internal format operator not correctly receiving operands
stuck elbow
the problem is that the adafruit_logging library uses % internally, so you have to pass a format argument to it: https://github.com/adafruit/Adafruit_CircuitPython_Logging/blob/main/adafruit_logging.py#L327
jaunty juniper
yeah I don't know actually, it seemed like it at first glance but I'm not quite sure
stuck elbow
so log.debug("I2cConnection received raw: %s", self._data_to_log_string(result)) should work
willow totem
the problem is that the `adafruit_logging` library uses % internally, so you *ha...
yeah exactly, that sounds like what I suspected, but I haven't checked what the differnce with fstring version or python logging module do
jaunty juniper
the problem is that the `adafruit_logging` library uses % internally, so you *ha...
yeah that's what I thought but:
>>> "no format" % ()
'no format'
Ah ok, that makes more sense
willow totem
so can I fix adafruit logging too, and then my problem goes away 🙂
or are there very good reasons like arduino/samd to use the % version
rough rapids
@Nerdadoc but () isn't a tuple, whereas *args is guarenteed to be at least an empty tuple
stuck elbow
so can I fix adafruit logging too, and then my problem goes away 🙂
fix how it's being called?
stuck elbow
@Nerdadoc but `()` isn't a tuple, whereas *args is guarenteed to be at least an ...
() is a tuple
jaunty juniper
yes it is
stuck elbow
it's a special case
willow totem
fix how it's not compatible with python logging module due to this bug
stuck elbow
normally it should be (,)
willow totem
or special case
rough rapids
ah, my bad, you're right.
jaunty juniper
fix how it's not compatible with python logging module due to this bug
that would be welcome yeah 🙂
willow totem
that would be welcome yeah 🙂
let the yak shaving continue 😉
stuck elbow
something like record = _logRecordFactory(self.name, level, (msg % args) if args else msg, args)
willow totem
oh clever and simple, was assuming id have to dig into python logging module to compare
willow totem
something like `record = _logRecordFactory(self.name, level, (msg % args) if arg...
on python bracket order and if/ternary etc, does it need the brackets around (msg % args)
?
stuck elbow
on python bracket order and if/ternary etc, does it need the brackets around (ms...
even if it didn't, it's easier to read with them
you don't have to guess
to be honest, I'm not sure myself, so I put them in there to avoid confusion
willow totem
oh there was one other technical question, I've had to comment out the future import due to cpy not having absoluteimport, this is the line from __future__ import absolute_import, division, print_function so my question is, is it just for python2 compatibility?
As I know future exists in cpy should I add a stub for absolute import (doing nothing), or if you were me and butchering the library anyway just comment it out everywhere and forgo Python2 compat (what ive done for now).
manic glacierBOT
brazen hatch
The feather tft code only captures print() and not direct serial writes.
That is a known issue I assume? Or is it a non-issue?
(I just got the feather)
slender iron
The feather tft code only captures `print()` and not direct serial writes.
That ...
can you rephrase your question?
you are typing over webserial and it isn't showing on the lcd?
zinc maple
Hey all! Been a while since I've been in here so not sure where is appropriate to put this so feel free to delete/move this post. Have some news regarding EduBlocks (which as you might remember has CircuitPython support) 🙂
https://twitter.com/all_about_code/status/1654196244678430723?s=20
blissful pollen
That seemed to work. On Windows had to connect to it within the OS settings. Then from the browser I could connect
brazen hatch
can you rephrase your question?
stdout.write over regular serial isn’t printed
slender iron
Hey all! Been a while since I've been in here so not sure where is appropriate t...
Congrats!
brazen hatch
wait one moment
oh no it’s just usb_cdc.console not being printed
that’s fine I guess
hmm, supervisor.runtime.serial_connected being always true doesn’t feel right
the usb_cdc stuff work like normal, but the rest are very sus
Well i’m glad I moved to console objects.
idle owl
Hey all! Been a while since I've been in here so not sure where is appropriate t...
This is so exciting! Congratulations on all counts!
slender iron
Well i’m glad I moved to console objects.
what are console objects? are you doing vt100 parsing?
brazen hatch
what are console objects? are you doing vt100 parsing?
With these you can read .connected, and read in/out_waiting with a single object.
So I made my curses library jcurses just accept these for it's data streams.
This way, I can plug a UART serial object to it just the same way.
And this means I can switch between different consoles in runtime.
I did this so I can in the future make diy ssh by a virtual console object.
brazen hatch
Yea makes sense. I had forgor I switched all the prints to that.
Well, when I make it store multiple streams, I will add mirror capability.
I want to eventually make it autoswitch like micropy does.
slender iron
you can make your own terminal grid and feed it your output
brazen hatch
I have enough ansi magic already, any more and my head will explode.
I kinda do mean that.
brazen hatch
This is running on the feather.
The display would have a stroke trying to display it regardless.
It would be funny tho.
Maybe when I finish the ljinux display driver stuff, I can make the display a virtual console object too.
stdin would be broken though.
orchid basinBOT
orchid basinBOT
onyx hinge
huh TIL that this works in circuitpython (but only if it's for ... in range...: : ```>>> for i in range(3.1, 7.9, 2): print(i)
...
3.1
5.1
7.1
it's a bug but an unimportant one
manic glacierBOT
orchid basinBOT
shadow vigil
the range concept is pretty central to python
lone axle
Hey all! Been a while since I've been in here so not sure where is appropriate t...
- Congratulations 🎉 2) The platform you created is amazing, great job and thank you for putting so much effort into making it easier for people to learn programming. 3) Do you have a preferred place or method to search for and/or submit issues? I've noticed the neopixel code that is generated is not complete enough to execute successfully on a device in some cases. Id be interested in attempting to help resolve it as well if you are open to other contributors (though I'll likely need to study up a bit on the JS tools in use). I found the Github here: https://github.com/edublocks/edublocks is that the best place to coordinate development efforts?
zinc maple
1) Congratulations 🎉 2) The platform you created is amazing, great job and t...
Thanks, really appreciate it! There's an in-app feedback section, however in this case I can take a look now at something like this. You can send me a DM pointing me in the right direction if that works for you!
tidal kiln
seeing a RP2040 specific I2C quirk
here's test code:
import time
import board
i2c = board.I2C()
i2c.try_lock()
i2c.writeto(82, b"")
time.sleep(0.1)
i2c.writeto(82, b"")
the first i2c.writeto looks like this:
the second one looks like this:
why this lag in the first one?
test setup for ref:
Adafruit CircuitPython 8.0.5 on 2023-03-31; Adafruit Feather RP2040 with rp2040
>>>
also tested with a feather m4, which does not do that
onyx hinge
@tidal kiln also not sure but I know that on rp2040 a zero-byte read is a special case that's actually using bitbangio, per a source comment: ```c
if (len == 0) {
// The RP2040 I2C peripheral will not perform 0 byte writes.
// So use bitbangio.I2C to do the write.
hmm. so maybe could be due to change over, from hw to bitbang?
curious what a non-zero followed by another zero len looks like....trying that now...
manic glacierBOT
idle owl
Sooooooo... I am working on some CircuitPython wifi code, and this is now what I'm getting when I try to run it. CircuitPython core code crashed hard. Whoops! Fault detected by hardware.
The thing I updated was that I'm running ping in the loop.
Which I'm fairly certain I need to do for what I'm trying to do...
brazen hatch
too many pings == crash
brazen hatch
Even veterans forget.
No problemo.
idle owl
I'm a CP wifi noob. So.. heh.
brazen hatch
I have a fair bit of wifi code, so if I can be of help feel free to ping.
idle owl
Do you know what kind of interval I should use?
brazen hatch
Do you know what kind of interval I should use?
I don't know. I just did 1s to mimic the regular ping command.
I guess, as long as it's not too fast it should be fine
idle owl
And yeah the loop is way too fast, clearly.
Excellent, I'll do 1s.
Hmm. time.monotonic() question. If I'm tracking two different time intervals, in two separate time blocks, do I need a separate variable for current_ or initial_time for one of the blocks? Or how does this work. ```py
Initialise time tracking
initial_time = time.monotonic()
Track the first time the code runs.
first_run = True
5 minutes in seconds
time_check_interval = 300
ping_interval = 1
The times at which it should change from blue to red, and red to blue
color_time = [RED_TIME, BLUE_TIME]
Track consecutive ping failures
ping_fail_count = 0
while True:
current_time = time.monotonic()
try:
if first_run or current_time - initial_time > ping_interval:
wifi_ping = wifi.radio.ping(ip=ip_address)
if wifi_ping is not None:
ping_fail_count = 0
if first_run or current_time - initial_time > time_check_interval:
if RED_TIME <= sundial.tm_hour or sundial.tm_hour < BLUE_TIME:
print("Red time!")
pixels.fill(RED)
pixels.show()
if BLUE_TIME <= sundial.tm_hour < RED_TIME:
print("Blue time!")
pixels.fill(BLUE)
pixels.show()
# reset clock
current_time = initial_time
first_run = False
I need a current_time = initial_time in the first block inside the try, but I'm wondering now if that changes how the second block's tracking works.
This might have made more sense in the help-with channel. 😕
brazen hatch
Hmm. `time.monotonic()` question. If I'm tracking two different time intervals, ...
Just store one, and check if time.monotonic()-stored > 1.
And then in the loop's end set it with a new time.monotonic().
idle owl
(That snippet is clearly not the entire program.)
Ohhhhh
I think I get it. Thank you!
idle owl
Just store one, and check if `time.monotonic()-stored` > 1.
And then in the loop...
So, like this? ```py
if first_run or current_time - initial_time > ping_interval:
wifi_ping = wifi.radio.ping(ip=ip_address)
print(wifi_ping)
initial_time = time.monotonic()
Because something isn't right again, and I am trying to figure out where to start.
The ping is happening 1 second apart at least.
brazen hatch
Can be shortened by setting current time as 0 for first run.
The rest seems fine.
Wait, the variable names are kinda flipped, so by setting initial_time as 0.
idle owl
I don't think I understand. 😕
brazen hatch
Yea same.
It's midnight and I have been studying for 5 hours.
stored_time = 0
if time.monotonic() - stored_time > ping_interval:
...
stored_time = time.monotonic()
crimson ferry
The thing I updated was that I'm running `ping` in the loop.
there's an issue https://github.com/adafruit/circuitpython/issues/5980
idle owl
Ah hmm.
brazen hatch
For the first run it will be time.monotonic() > ping_interval.
Wifi connection can take like 7s, so it's not a matter of being in that part in the first second.
idle owl
Hmm ok
first_run is a tracked variable. And you're saying I can do it with time instead?
brazen hatch
Yea same.
It's midnight and I have been studying for 5 hours.
```py
stored_time ...
this, should work.
idle owl
Ohh hm. Ok.
brazen hatch
I see no reason to store a first_run just for this
idle owl
Let me see if I can make that work.
Ooooh.
This is working time-wise now. I think.
Trying to make another part work the way I want now.
brazen hatch
If you want to make it extra fancy you can do the rtt and averages.
idle owl
This is supposed to be simple 😂
Ok, give me a minute to check whether the ping successful code works right. Then is it ok if I post it here for you to take a look at? I'm absolutely certain I'm doing extra that I don't need to, but this is how I know how to do it.
Ugh. Safe mode.
brazen hatch
Well, I suggest you spam a few print(f"CT: {time.monotonic()}") around the code.
idle owl
I'm spamming rando prints at least.
brazen hatch
Ugh. Safe mode.
I have no problemo if you post the code here.
Just make sure to ping me cuz I am alt-tabbing.
onyx hinge
too many pings == crash
is there an issue filed about this?
brazen hatch
is there an issue filed about this?
yep, an old one
brazen hatch
It can prolly be fixed by the same time.monotonic logic, by storing a last_ping_time in the core.
idle owl
I figured it out.
It's doing what I thought it would
the time is getting reset to time.monotonic() in the loop, so the second time interval is never met because it's longer. That's why I thought I might need a separate stored_time.
I think I need one more time variable.
brazen hatch
Not sure I quite understand the issue. As long as stored_time is initially defined outside the loop, it should be fine.
Code?
idle owl
Two shakes
jaunty juniper
I prefer using a variable that is set to when the next event must happen, when you have multiple timed things, each has their "next time" variable
"next time" or "last time" is equivalent, but I find that using the next time is easier to math
idle owl
Ok, @brazen hatch here's my code. It is currently working like I want it to, or so it seems. No safemodes bounces, and it's doing the things when it's supposed to.
jaunty juniper
it also makes it easier to compute next time in different ways depending on whether you want to guarantee a delay or avoid drift
idle owl
That said, I'm almost certain it could be simplified somewhere.
jaunty juniper
That said, I'm almost certain it could be simplified somewhere.
do you want to check the time if the ping fails ? doesn't that mean that you don't have access to the network ?
idle owl
do you want to check the time if the ping fails ? doesn't that mean that you don...
Oh the time thing in the beginning? I'm not even sure it's necessary. This is expected to be plugged into the wall and left alone. I think I did it more for myself than anything.
So in theory that could be removed entirely.
But good call.
One would probably not want to check it if the ping fails.
brazen hatch
That said, I'm almost certain it could be simplified somewhere.
From a quick read:
- For 24/7 uptime with no hiccups,
safemode.py,del&gc.collect()are recommended. - Using regular NTP may be better.
initial_time == 0->(not initial_time).- Using parentheses in the
ifs is also good for readability. - psf/black formatting not applied.
brazen hatch
It was a quick read, logic errors may have been missed.
crimson ferry
From a quick read:
- For 24/7 uptime with no hiccups, `safemode.py`, `del` & `g...
watchdog handy too, and a try/except wrapper for the whole shebang
idle owl
I'm walking the line between "simple" and "will keep working"
crimson ferry
supervisor.reload and microcontroller.reset are two of my best friends 😉
overkill for present purpose, but here are all of the things I could think of for robustness of a CP project: https://gist.github.com/anecdata/cfc4c585406d17985885bfe3a1cc9c73
idle owl
There's a reset on error function that will be switched back to microcontroller.reset()
brazen hatch
overkill for present purpose, but here are all of the things I could think of fo...
This is the manifesto of ljinux code.
Ljinux is and has to be 100% ready for every exception and at the same time affect the stack as little as possible with no leaks.
Tending to gc manually is also critical for performance.
crimson ferry
gc is a bit in heuristic territory I think, someone even mentioned a case where they had to gc twice in a row to make it happen
brazen hatch
Yes, it's a matter of striking a balance.
You need to do just enough gc.collect()'s so that the heap doesn't significantly end up fragmented, or full with stuff to be deleted.
It's faster to clean 2 times 10k of ram, than clean 20kb when stack is full.
manic glacierBOT
brazen hatch
But if you always spam it, it ends up way slower.
Doing 100 times 50 bytes is a million times slower.
This is very obvious in any ljinux tui application.
nano being the biggest mem hog.
brazen hatch
By not doing any, garbage accumulates and then it takes several moments for it to collect all of it.
jaunty juniper
have you measured the gc time ? on boards with 8MB of ram it takes an eternity even if you didn't allocate anything
idle owl
May I suggest temporarily swapping ping with a placeholder print function?
Of course you may, and I will do that.
brazen hatch
have you measured the gc time ? on boards with 8MB of ram it takes an eternity e...
By feel only.
By scrolling screen in nano.
jaunty juniper
I'd love to see some benchmarks
brazen hatch
I plan to make some in the future.
Ljinux has a plethora of funcs and stuff that needs time benchmarks.
It will measure all system times.
I have done some parts in test code.
But never all in one.
I want a stress-ng equivelant.
I do want to see flash & psram performance with different block sizes.
idle owl
@brazen hatch Is the point of this to basically prove that the issue is the ping feature?
I don't get it though. It ran for a while just fine, and then boom safemode again.
brazen hatch
The point of what? Using a placeholder? There is nothing else that can safemode the board in that code, is there?
So by swapping it for now, you get to not have to reboot the board all the time.
idle owl
I see where the safemode.py would come in handy, but if it's dumping into safemode approximately 8 seconds after it starts running every time, that isn't so handy.
idle owl
The point of what? Using a placeholder? There is nothing else that can safemode ...
I don't think there's anything else, no.
I think it's ping making it fail.
brazen hatch
Just swap ping with something like print(f"PING! {time.monotonic()}") and set the result value with a 1 or something.
idle owl
I was printing "ping goes here"
and it was running fine.
I slowed the ping down to 2 seconds, and now it's running ok. But .... nevermind.
Safemode.
brazen hatch
Well, in the log, don't you see many ping messages if you add the print instead?
idle owl
I don't think I follow what you're asking
brazen hatch
Just swap ping with something like `print(f"PING! {time.monotonic()}")` and set ...
With this in place of the .ping(), what does the serial output look like?
idle owl
Oh man! Turns out I hadn't increased the delay. It doesn't matter, it did enter safemode eventually anyway. But! The ping returned None enough times for the full purpose of the code to work, where it stops showing blue, and blinks red when the ping fails, and then once ping is working again, returns to blue. So.... that's nice I guess.
idle owl
With this in place of the `.ping()`, what does the serial output look like?
Ok give me a moment
idle owl
With this in place of the `.ping()`, what does the serial output look like?
etc.
Interval currently set to 2 seconds.
idle owl
It feels like something is hosed with ping.
brazen hatch
bruh
idle owl
@onyx hinge Can you give me a QT Py ESP32-S2 build? I can't remember where to find the assets or whether they're even built yet.
brazen hatch
My code still works though, so 1s is fine.
onyx hinge
@idle owl yes I can make one then I have to step away for dinner
idle owl
<@330227457296957440> yes I can make one then I have to step away for dinner
Thank you.
I'll test it with my code.
onyx hinge
@brazen hatch empirically it's somewhere between .2 and .3 seconds that it switches to "mostly working, no exception". not sure exactly why that would be.
blissful pollen
If you need I'm about to make another build if needed
idle owl
If you need I'm about to make another build if needed
Cheers.
idle owl
Thank you!!
onyx hinge
afk, you're welcome though
idle owl
Have a delicious dinner!
@brazen hatch OK, so here's the weird thing it's doing now. (Running the code to see if it fails.) Every 7 pings, it returns None. Every single time. One of the times it returned None two times in a row, but otherwise 7 successful, 1 fail, 7 successful, 1 fail..... This was happening earlier, and I thought I did something to fix it, but I a. don't remember what that might have been, and b. am unconvinced I ever did anything to fix this in the first place.
I fixed the thing where it wasn't printing None for every failure already.
brazen hatch
None is a failed ping, is it not?
That is connection related as is specific to your wifi setup.
I have never seen anything as such on any of my boards.
You should prolly just treat failed and None the same way.
idle owl
Ok, trying that.
brazen hatch
I connect my boards to my pc's / pi's wifi hotspot so I can monitor wifi stuff via journal.
This way, severe errors are registered.
idle owl
LLOL...... I ran out of memory??
brazen hatch
LLOL...... I ran out of memory??
u runnin it on a tater?
idle owl
It's still doing the 7 on 1 off thing.
idle owl
u runnin it on a tater?
Possibly.
jaunty juniper
that might be a wrong error diagnosis ?
brazen hatch
Perhaps it's time to begin le gc.collect() spam.
idle owl
That's fair. I happened twice after about 16 pings, but that isn't consistent now.
brazen hatch
could be mem, since I have never seen it.
crimson ferry
but no safemode?
jaunty juniper
the code doesn't allocate anything big, and the QT PY has tons of RAM, was it an ESP memory error maybe ?
idle owl
Correct!! No safemode!!
crimson ferry
the PR only catches the error (to avoid safemode), not change the underlying reason
that's deeper in the ESP IDF
brazen hatch
if esp-idf was stable we would have flying cars by now
jaunty juniper
I'm running your code on my FunHouse btw with a 2s delay and things seem to go well
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.004
0.005
0.007
0.005
etc.
crimson ferry
over 1s should work, but...
idle owl
I mean, this 7 on 1 off thing doesn't really matter - it doesn't do anything until ping fails 5 times in a row.
idle owl
I'm running your code on my FunHouse btw with a 2s delay and things seem to go w...
Switched back to 2. Trying that.
Still getting that None every 7 times. 🤷🏻♀️
idle owl
I mean, this 7 on 1 off thing doesn't really matter - it doesn't do anything unt...
But, I'm annoyed with it. Oooh, that time it None'd after 3 successes.
brazen hatch
Well, I will go sleep now, have fun.
My last suggestion is to test the ping part of the code, in a different code.py, where nothing else is going on.
idle owl
Well, I will go sleep now, have fun.
My last suggestion is to test the ping part...
Thank you so much for all your help!!
brazen hatch
No problem. Glad I could help.
idle owl
It's an espidf memory failure. Line 138 is the ping code. Maybe I merged that PR too quickly?
I need to go eat. I guess I'm done with this for tonight.
blissful pollen
Maybe something isn't cleaning up completely. I think the PR still did fix one issue. Could be more then one was there
jaunty juniper
I think the PR converts the crash into an exception, it's an improvement
onyx hinge
I think the PR converts the crash into an exception, it's an improvement
yes that's exactly right. on my system I started getting a "generic error" instead of the crash; kattni's error is a bit different, maybe because of different memory usage in her code vs mine
idle owl
Hmm.
So am I supposed to catch the error now?
Is that the idea?
Already have it in try/except. I can try an if/else under the except I suppose. I assume that works.
blissful pollen
The try/except should catch the exception if it shows up. You can't catch an exception with if/else normally
idle owl
The try/except should catch the exception if it shows up. You can't catch an exc...
No I mean, inside the except, if the error is the espidf error, pass. And if it's other stuff, reboot the board.
Although I have no idea if it will keep running after that error.
Or if it requires the reload.
Basically this will reload all the time, it seems like. Sigh.
manic glacierBOT
brazen hatch
question, I am making a board port,
this board has a battery hold pin
if it’s not held up, the board will power down when the power button not held and not plugged in.
Should I have circuitpython never reset the pin and hold it up by default?
It’s the m5stack timer camera x, straight from ada store.
It somehow doesn’t have a circuitpy port
It has a builtin camera and is a standalone battery powered package
I have arduino rtsp + webserver code and it can stream 30fps video just fine
manic glacierBOT
orchid basinBOT
brazen hatch
Also, how do I map camera pins to the mp_rom_obj_tuple_t camera_data_tuple that is set for others?
I have no clue. I only have the datasheet and a working arduino example.
brazen hatch
bool espressif_board_reset_pin_number(gpio_num_t pin_number) {
if (pin_number == 33) {
// Turn on BAT_HOLD by default, so that the board does not power off.
gpio_set_direction(pin_number, GPIO_MODE_DEF_OUTPUT);
gpio_set_level(pin_number, true);
return true;
}
return false;
}
Doesn't really work. Not really sure why.
The pin stays unset.
brazen hatch
I added a mp_printf(&mp_plat_print, "Resett.\n");
It gets printed on the correct pin deinit.
mental nexus
Just saw about your desire to use the LCD peripheral on the ESP32-S3 for a parallel display. Definitely doable but will need some work on the CircuitPython core. I made a fork to add a DotClockDisplay using the S3 LCD peripheral, but it was never merged. But had lots of fun hacking on it!
idle owl
I have come to believe that there is a memory leak in the ESPIDF related to WiFi and requests. I'm not even using ping anymore because I couldn't get past 3 pings without it failing with a memory error. Now... it's failing when trying to retrieve the time service from Adafruit IO, saying Sending request failed but, if you ctrl+c in the middle of the reset sequence, it turns out that error is being directly caused by espidf.MemoryError.
onyx hinge
@idle owl I would be happy to look at it in detail with you, maybe monday after the meetings.
idle owl
<@330227457296957440> I would be happy to look at it in detail with you, maybe m...
Yes please.
onyx hinge
I found something hinky in adafruit_io but I don't think it alone is the cause of the problem: https://github.com/adafruit/Adafruit_CircuitPython_AdafruitIO/issues/100
@idle owl which qtpy is your project using?
idle owl
<@330227457296957440> which qtpy is your project using?
ESP32-S2.
onyx hinge
OK, we can talk more Monday!
idle owl
For sure. Thank you!
crimson ferry
ESP32-S2.
I get those routinely in _get_socket just using requests. Retry a couple of times, same. Do some other stuff then come back and do a request later and it's fine again. Checking available espidf memory (QT Py S2) prior to the (successful or failing) socket operations gives something like:espidf.heap_caps_get_largest_free_block(), espidf.heap_caps_get_free_size(), espidf.heap_caps_get_total_size() = 253952, 284736, 405960seemingly plenty of espidf memory.
I also have CIRCUITPY_RESERVED_PSRAM=262144, but that may not be working as we think??
or it's not actually an espidf memory error??
lavish saffron
To enable the WIZNET5K driver to be compatible with the Adafruit Adafruit_CircuitPython_HTTPServer the socket.socket module needs to raise OSError exceptions, since these are trapped in try: except: statements. What is the best source for these error types? I've been looking at the BSD Unix socket man pages. Do you think I'll get close enough using these?
onyx hinge
@lavish saffron interesting, our errno module seems to be undocumented ... ```py
MicroPython 8.1.0-beta.1-4-g1ca435c5e3-dirty on 2023-04-05; linux version
Use Ctrl-D to exit, Ctrl-E for paste mode
import errno
raise OSError(errno.EAGAIN)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
OSError: [Errno 11] EAGAIN
Ah, that's perfect. I'll see what I can do. Thank you!
crimson ferry
Traceback (most recent call last):
File "/Users/a/Documents/GitHub/circuitpython/ports/espressif/../../tools/gen_web_workflow_static.py", line 7, in <module>
import minify_html
ModuleNotFoundError: No module named 'minify_html'
make: *** [build-espressif_esp32s2_devkitc_1_n4r2/autogen_web_workflow_static.c] Error 1
is this something not installed as part of the circuitpython / espidf environment?
ditto jsmin
crimson ferry
oh, that's a note in the "Manual Setup", which the macOS setup says to skip over
but I guess build error counts as "If the setup instructions above don't work for your particular OS setup, for whatever reason"
I build so infrequently, it's always a new experience 🤪
jaunty juniper
yeah if you didn't build since things were added, you'll be missing stuff
my build script runs the pip install every time, it takes 2s when there's nothing new
crimson ferry
yeah I think the requirements stuff is new since I last did this
make -j got errors, so I'm doing no -j 😦
brazen hatch
I suggest keeping an env for circuitpython compiles, seperate from idf's
It's also that arch is now changing to py3.11, so a lot of stuff will break.
idle owl
Can you put symbols in a computer hostname? Specifically _
onyx hinge
@idle owl I think that letters, numbers and "-" OK, but "_" is not. Assuming you mean a computer name like you might have in a URL, "example.com" and so forth.
you can have "sub-domain.example.com" but not "sub_domain.example.com"
onyx hinge
yes, try that
idle owl
That's equally fine. I'm not married to _. I'm simply used to it from file naming.
Thank you!
wraith crow
Is the safemode.py feature merged into the 8.0.5 line?
Based on a search of the github release notes, I'd say not...
jaunty juniper
no
wraith crow
Thanks 😄
jaunty juniper
yeah, https://github.com/adafruit/circuitpython/commit/d8231f15886ef43cf3a052c50bca49ec6dd150ff only lists 8.1.0-beta.2 8.1.0-beta.1 8.1.0-beta.0 8.1.0-alpha.2
manic glacierBOT
lone axle
Any thoughts on this? https://github.com/adafruit/Adafruit_CircuitPython_Colorsys/issues/28 (or perhaps it's good for weeds?)
We have a library that is a partially complete implementation of colorsys from CPython, but it was altered to have different return types for the functions that it has (ints 0-255, whereas the original CPython is floats 0-1). I think it makes sense to change it to be floats so it will match the CPython implementation to make code using it more portable. I had a preliminary look around Learn repo and Library Bundle examples and found no usages of the Circuit Python library specifically, but any code that is out there that uses it would need to get adjusted if we did make this change.
slender iron
Any thoughts on this? https://github.com/adafruit/Adafruit_CircuitPython_Colorsy...
It should definitely match CPython if it shares a name
crimson ferry
what's the magic these days to make space for a debug build? (esp32-s2 4MB flash) every combination I've tried gets some undefined reference
lone axle
what's the magic these days to make space for a debug build? (esp32-s2 4MB flash...
I think I disabled ulab last time I needed to make a debug build. perhaps one or two others but I don't recall them for certain.
jaunty juniper
did you make clean ?
crimson ferry
I always make clean
I don't even see ulab in mkconfigport.mk ...do you do that on the command line?
jaunty juniper
CIRCUITPY_ULAB = 0 in the mpconfigboard.mk
crimson ferry
oh, board level
jaunty juniper
or the command line
gmake BOARD=adafruit_feather_esp32s2_tft -j6 DEBUG=1 CIRCUITPY_ULAB=0
that was in my history
(not really, I have a script that makes those for me, but that's the result)
crimson ferry
-j has been giving me errors, so I've been doing make and then do something else for a while
jaunty juniper
you can only override what is not set in stone in the board file from the command line though
(and what is defined in py/circuitpy_mpconfig.mk)
stuck elbow
unless they use ?=
crimson ferry
what is set in stone?
jaunty juniper
yeah board files don't use ?= maybe they should, I assume it also defeats the github custom build
jaunty juniper
what _is_ set in stone?
everythin that is in mpconfigboard.mk that doesn't use ?=
jaunty juniper
then you're free to play with anything
manic glacierBOT
stuck elbow
I usually try to pin down all the possible settings in mpconfigboard.mk, otherwise someone always comes and changes the global defaults, and I only realize half a year later that it stopped working
crimson ferry
this is very arcane stuff
slender iron
I usually try to pin down all the possible settings in mpconfigboard.mk, otherwi...
did you see jepler turn of synthio on one of your boards?
stuck elbow
did you see jepler turn of synthio on one of your boards?
no, I didn't
slender iron
k, I pinged you on github but you must have missed it. his changes have made it better and larger. not sure what you want to do about it
stuck elbow
well, I'm getting abuout 40 e-mails a day from github
turbid radish
The newsletter is ready to crib from in GitHub for the 2pm meeting
slender iron
The newsletter is ready to crib from in GitHub for the 2pm meeting
thank you!
slender iron
well, I'm getting abuout 40 e-mails a day from github
https://github.com/adafruit/circuitpython/pull/7862#discussion_r1184001712 Ones with mentions are designated with a cc to a mention address https://docs.github.com/en/account-and-profile/managing-subscriptions-and-notifications-on-github/setting-up-notifications/configuring-notifications#filtering-email-notifications
stuck elbow
thanks
I was hoping to use synthio in my games, but to be honest I never had the time to actually try it
maybe with the size gains of the newest micropython it will fit in there again, when they are merged
slender iron
¯_(ツ)_/¯
stuck elbow
progress
slender iron
it does sound way better
stuck elbow
for sure, I was planning to try and make something like the bfxr for circuitpython for years now
slender iron
bfxr?
stuck elbow
it's for generating sounds for games
I guess it's replaced with sfxr now that flash is dead
basically you set about a dozen parameters, and you get a cool sound
slender iron
nice!
manic glacierBOT
slender iron
<@&356864093652516868> We'll have our weekly meeting in just over an hour 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/1y5tGjTm1zDpbaSjTIQLytXE_Eg4mmfI4kf95AclnJ2Q/edit?usp=sharing Thanks! Talk with you soon.
idle owl
or it's not actually an espidf memory error??
I'm wondering if it's a memory leak in the ESPIDF, but this is certainly not my wheelhouse. Jeff and I are planning to look into it, I believe.
@onyx hinge Are you still up for looking into the wifi / requests issue today? Worth noting: I have no reason to believe that my code hasn't been running for a solid day or two at this point. That said, I have no way to know that it didn't fail and a soft reload resolved it, and that event was off my scrollback. Thing is, when it was failing previously, it would pretty consistently keep failing after that and a supervisor.reload() did not resolve it. So, I'm wondering if something I did is causing it to avoid failing now. I don't know anymore, heh.
onyx hinge
@idle owl we can sure look at it still
idle owl
<@330227457296957440> we can sure look at it still
Thanks so much!
slender iron
brb
manic glacierBOT
devout jolt
Rust, clearly
idle owl
Scott - something is rubbing against your mic, and it's causing staticky type sound from you.
👍🏻
turbid radish
😁
idle owl
Hey I'm on the review list on the core for once! 😄
devout jolt
brb
willow totem
tektriks got self review labelled today for my adafruit_circuitpython_logging pull request so thank you
idle owl
🎉
silver tapir
@onyx hinge FIR for ESP32. I bit old, but it should be a nice example on how to use the espressif stuff to get an FIR working.
https://github.com/bobh/ESP_DSP/blob/master/ESP_DSP/ESP32FIR/README.md
onyx hinge
self.plot_line_pint: List[Tuple[int,int]]] = []
silver tapir
@gilded cradle Good topic for on the weeds, this is a complex issue for riscv boards as well, as some board makers publish distros only sometimes, and normal distros like ubuntu/debian/fedora work only sometimes. A good example are the D1 boards, which all share a distro, and don't have device tree files for each boards so they might have i2c pins on the wrong place.
This is also a complex issue as it cannot be tested on something like Github's CI and even if the distros existed, they wouldn't run on the real hardware. Maybe github self-hosted runners could help with testing, but I'm getting off-topic here.
So far a markdown doc with a nice table should be good enough.
silver tapir
No, there is nothing specific about the distribution, so the code doesn't use systemctl or the package system.
But it is a good idea to know what has been tested so far, in case you are having problems, to just check the doc.
And yeah, circuitpython.org is a good idea as well.
idle owl
@gilded cradle If there's documentation elsewhere pointing to circuitpython.org, you'll cover a wider user-base.
Keeps your external docs simpler, but still gets folks where they need to go.
silver tapir
Also, there are 100's SBC distributions, so making a new functions for each would be weird.
idle owl
Exactly!
silver tapir
And the other problem is that it kinda has to be a matrix of distribution and boards, so the outcome will be quite large.
Debian might work on one board, but not on the next one.
idle owl
Entirely unrelated, but I forgot to share earlier: I received my new MacBook Pro this morning. M2 Max with 96GB of RAM. Super excited!
gilded cradle
Sweet @idle owl. I'm still on the one I got in 2019.
random junco
that's a lot of RAM!
idle owl
Sweet <@330227457296957440>. I'm still on the one I got in 2019.
If it's working for you, stay with it until it's not enough!
idle owl
that's a lot of RAM!
Right?
silver tapir
Thanks all.
gilded cradle
If it's working for you, stay with it until it's not enough!
Yeah, that's what I was thinking too.
ember iris
Thanks all! Have a great one!
gilded cradle
Thanks everyone
random junco
Thanks everybody!
turbid radish
I have 64GB on my Threadripper
devout jolt
96GB of RAM? You'll be able to run Chrome!
idle owl
Mine was still working quite well, but Rose needs an upgrade, and it makes more sense for me to upgrade and give her my current one.
turbid radish
Not since the Arduino Edison
silver tapir
@royal rover You can try following this stuff to add them to blinka. If they work with the linux gpio library, they almost for sure work with blinka.
https://learn.adafruit.com/adding-a-single-board-computer-to-platformdetect-for-blinka
https://learn.adafruit.com/adding-a-single-board-computer-to-blinka
slender iron
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/1xektzKK1vh8mua913H68WG-104GH4q00L6EmL87ZRGE/edit?usp=sharing
idle owl
@lone axle Maybe you know. How do you format the :return: in this? It returns a color tuple. ```py
"""
Function description.
:param hour: Parameter description.
:return: ???
"""
I think Tuple[int, int, int]
idle owl
I think `Tuple[int, int, int]`
Oh weird. OK.... I think that's too much for this code. I thought maybe it was a sentence or something explaining what it returns. Nope! Thanks so much!
lone axle
@idle owl I think it can have an english description after the colon as well. Some like "Tuple containing 3 int values 0-255 representing red, green, and blue" Would be good if it needs to be comprehensive or is aimed at someone who may not know.
left of the colon would be where the Tuple[int, int, int] lives I think. And right of colon is human readable.
idle owl
Oh.
That makes a little more sense.
It's within a code.py
So I'm trying to document, but also, it will have a guide so....
Maybe don't overcomplicate the inline documentation.
@onyx hinge Can I do if thing or second_thing and third_thing: and have it execute either if thing is true, OR if second_thing AND third_thing are both true?
Or do they need to be nested ifs.
Or can you do and twice with three conditions?
(Separate question)
I think I have to nest to do what I want.
I'll try that, and you can correct me tomorrow.
lone axle
Is there any official or unoffical supported version of CPython for the libraries? The closest thing I know of is Blinka declaring a minimum of 3.7: https://github.com/adafruit/Adafruit_Blinka/blob/b84de4800b2c31023f29a536d26209bd7517d504/setup.py#LL52C1-L52C31 Is that considered the minimum supported version? i.e. would we want to hold off accepting a change that requires python 3.8+?
midnight ember
Just finished up watching this weeks meeting. A DVI featherwing that offloads DVI onto a 2nd dedicated RP2040 is something I would definitely be interested in for taking HDMI to the next level stacked with an Airlift. Definitely looking forward to that becoming a thing. Baby steps. 🙂
crimson ferry
this link in this guide https://learn.adafruit.com/feather-rp2040-dvi-video-synth links right back to the source guide
orchid basinBOT
orchid basinBOT
hidden rain
Good Morning All ! I am using analogbufio repeatedly in an application. While there is a deinit, I am uncertain about memory resources. I was hoping to re-use the same object repeatedly, but that returns messed up data. This feature might benefit from having a reset feature. Comments, Opinions?
orchid basinBOT
onyx hinge
@hidden rain Are you saying two readinto calls gives a bad result for the second call? that sounds like a bug. Otherwise, can you be more specific about what the Python code you're trying to run looks like..?
hidden rain
<@314934150647906305> Are you saying two `readinto` calls gives a bad result for...
yes. the data the second or nth time around is total trash. I'd like to request a resource object reset to this feature.
I forgot how to do that (feature request) or open an issue.
hidden rain
Perhaps someone would confirm analogbufio.readinto() failures on use beyond first.
jaunty juniper
yes. the data the second or nth time around is total trash. I'd like to request ...
have you tried manually zero-ing the buffer ? Not good for performance, but to see if it has to do with the issue ? This does it without allocating a new buffer:
for i in range(len(buffer)):
buffer[i] = 0
opening an issue is here:
https://github.com/adafruit/circuitpython/issues/new/choose
hidden rain
have you tried manually zero-ing the buffer ? Not good for performance, but to s...
Yes. The first thing I tried; pretty much just as you wrote it. I did the experiment several times. Before zeroing, the buffer is motly noise with a small reflection or shadow of what was previously good data.
orchid basinBOT
slender iron
who was talking about a streaming json parser?
the pirate weather api response is 30k+
manic glacierBOT
slender iron
is there a plan to update pylint in libraries? 2.11.x doesn't work with python 3.11
idle owl
is there a plan to update pylint in libraries? 2.11.x doesn't work with python 3...
There's always a plan to update Pylint. 🙂
Eva's usually the one to do it, but she's been out. I'll sort out who can.
proven garnet
tekktrik has entered the chat
I can get a patch ready this week, either to run this week or next.
idle owl
I can get a patch ready this week, either to run this week or next.
That's not the only thing though. There will be updated needed to conform with the new version. 😉
Basically, x number of libraries will fail to build due to new Pylint errors.
And someone will need to deal with all of that.
proven garnet
I can handle the unfortunate fallout
idle owl
Excellent. Thank you so much!
slender iron
thank you!
manic glacierBOT
idle owl
To everyone who was helping me with my ping and time-check code, thank you. After banging my head against this for longer than I care to admit, it turns out my code runs flawlessly on the QT Py ESP32-S3. I have no answers for why, and at this point, I need this project to run, so I don't care. But I wanted to let you know that I found a workaround, and things are happy now.
I was using the ESP32-S2.
And couldn't ever figure out making it work right. And it continued to fail in the weirdest ways.
@jaunty juniper and @brazen hatch It was apparently a hardware issue. So I either have a bad board, or there's something up with ESP32-S2 and the features of my code. Either way, it's working now on the ESP32-S3, and I've spent too much time on it already. So I'm handing it off with the hardware change, and that's what we'll be using for the guide.
brazen hatch
Hmm, I should investigate, but schedule's rather full now.
idle owl
No worries.
jaunty juniper
yeah it's a weird issue, I did not have ping errors on the FunHouse
idle owl
Well, ping works great on its own on the S2.
It's something about ping and io.receive_time() together that end up failing in all sorts of weird ways.
crimson ferry
@idle owl I also have network code that runs much better on S3 than S2, no idea why
idle owl
<@330227457296957440> I also have network code that runs much better on S3 than ...
Fair enough. I suppose at least it's not only me.
crimson ferry
adafruit_httpserver doesn't show up on the PR page? https://circuitpython.org/contributing there is an open PR
it's gone in and out of draft, maybe that's an issue?
manic glacierBOT
manic glacierBOT
proven garnet
I might be late to the party, but just a heads up that GitHub is having several issues all around. Particularly, the GitHub Actions tubes are clogged: https://www.githubstatus.com/
tulip sleet
@slender iron I found one use of plain "PyRate": https://github.com/dsilvestro/PyRate
also "Pyrate" is considered an alternate spelling of "Pirate" in English in general
slender iron
huh, interesting
I'll leave naming up to pt and limor
@tulip sleet this is what I reimplemented: https://github.com/daggaz/json-stream/
I reimplemented it because I use the json module for value parsing
idle owl
The upcoming collab project with Noe requires a CP fix that is currently only on main. What are the plans for the next release?
Or at least it's possible that it requires it. The failure that requires it isn't happening so far on S3, but I still think it's worth having so if it does happen, my code will keep running.
tulip sleet
<@329766224093249548> this is what I reimplemented: <https://github.com/daggaz/j...
Did you implement the "visitor pattern"? That would let you check each parsed entry aginst all the required paths, so order wouldn't matter. This is all very reminiscent of SAX-style XML parsers, which maybe you were able to avoid in your work life up to now.
tulip sleet
The upcoming collab project with Noe requires a CP fix that is currently only on...
no firm plans, but the last release was two weeks ago, so I wouldn't want to wait more than a couple more weeks for a release. we can do a release at any time
idle owl
no firm plans, but the last release was two weeks ago, so I wouldn't want to wai...
Excellent. Cheers.
slender iron
Did you implement the "visitor pattern"? That would let you check each parsed en...
I didn't. ijson was another option
tulip sleet
saw that too -- the visitor pattern could be implemented at a later time if we felt necessary
blissful pollen
Has anyone been able to run the latest builds of the M7? Anything after May 1st is not even bringing up the CP drive. Tried with my own build and same thing
proven garnet
GitHub is back up, and all libraries should have a newly updated .pre-commit-config.yaml! black and reuse in particular are having new issues, please be aware. I'm currently running pre-commit on all the repos to fix the formatting issues, so if you're working on any libraries in the Adafruit bundle, you may want to merge/rebase accordingly. I'll follow up with fixes for REUSE and pylint as well over the evening and next couple days.
proven garnet
Libraries are patched! Minor fixes inbound for a handful of libraries that now have newly showing pylint issues, but otherwise everything is all set!
proven garnet
All post-patch fixes submitted as PRs! ( @idle owl )
proven garnet
Github is once again down 
manic glacierBOT
brazen hatch
S3 for some reason really dislikes nano
slender iron
Has anyone been able to run the latest builds of the M7? Anything after May 1st ...
Jeff has a PR out with fixes
@onyx hinge I was hoping you'd figure out the reason but I can look if you don't want to.
Seems worth understanding
tidal kiln
general tooling question - does this effectively centralize the CI stuff for CP?
https://github.com/adafruit/workflows-circuitpython-libs
proven garnet
general tooling question - does this effectively centralize the CI stuff for CP?...
Yup!
At least the most common shared parts for building and releasing
The libraries just call out and run those
tidal kiln
@proven garnet hey. thanks for responding. actually asking this as part of looking at the arduino CI stuff - and the impending end of node12, etc. i think you actually brought this up? i got word of it via kattni.
proven garnet
<@135415127879974912> hey. thanks for responding. actually asking this as part o...
Oh yeah, I did. It future proofs a lot of the issues regarding upgrading in the future.
Yeah, this was part of the fix for CircuitPython for the Node issue, or I did it right after.
tidal kiln
cool. yah. that's what i was hoping / thinking.
current idea was to batch update all the repo's yml files
since it's a simple edit
but now wondering if doing something like this might be better
for now and future
proven garnet
Yeah, it's already paid dividends since.
You don't have to pack everything in it, but if it's shared across the whole codebase it at least gives you a single point to edit.
Since its a composite action you can still locally do extra stuff as well
tidal kiln
neat. yah. this seems like a much better way to deal with this.
are there any limitations? like issues with keys or anything? not seeing anything obvious. seems like its just a different way to organize.
muted current
@tidal kiln, sorry for pinging you, but I have a very good question (in my opinion, you might think its stupid and not worth your time), why were these discontinued
I want a pair of these so I can make a one eyed wearable display like zack freedman but I cant find them anywhere, not even EBAY
proven garnet
are there any limitations? like issues with keys or anything? not seeing anythin...
Yeah, potentially. I think I always pass secrets to be safe, but each repository will need that secret stored (or potentially an org secret)
I think actually passing them may not be required, but definitely nice to see in the file in each repository explicitly what is being passed.
tidal kiln
@muted current not sure. but not really #circuitpython-dev related. also no need for the ping.
muted current
sorry, I thought this would be the right place for this thing
where can I take this to???
tidal kiln
probably #help-with-projects since it seems general
muted current
thank you, and sorry for ping
idle owl
All post-patch fixes submitted as PRs! ( <@330227457296957440> )
Thanks so much!
slender iron
@onyx hinge want me to look into the usb pins on imx after I'm back from lunch?
onyx hinge
@slender iron my day is pretty shot at this point so if you can! (but for good reason, was helping my friend who had to move his metalworking shop on short notice last summer .. it's still not working yet but we made big strides)
yes moving that far is exciting 🙂
slender iron
Kk will do once I’m back at my desk
proven garnet
neat. yah. this seems like a much better way to deal with this.
I meant to say, I'm mostly out this weekend but can answer questions so feel free to ping or DM me so I see it!
tidal kiln
I meant to say, I'm mostly out this weekend but can answer questions so feel fre...
cool. thanks. i think plan is to near term just do a simple edit of all the ymls in the arduino lib repos to bump the action versions used. so things wont break when node12 support goes away. but then also, longer term, take a look at changing to a setup like what CP is doing.
blissful pollen
@devout jolt I was looking at your synthio test YT videos and you have, I think, an I2S to headphone jack breakout board? Wondering what it is. Would be nice to hook up to slightly better speakers
devout jolt
<@352910176736772096> I was looking at your synthio test YT videos and you have,...
yeah I am a big fan of these cheapie PCM5102 boards you can find on Amazon or Ali. Here's one link: https://amzn.to/3M0gQMp Those are officially line-level out, but I think they drive most headphones
blissful pollen
Cool thanks. I have those little stereo 4ohm speakers and besides my shoddy wiring they're a "bit" tinny
devout jolt
it's really hard to get those I2S amps sounding good. You really need to be driving speakers in enclosures (turns out boxes are useful, not obvious to me) and of course, larger speakers == better sound, but now you have the power problem of how to drive those speakers off your USB. I like offloading all the audio amplification to a known-good solution of your standard cheap bluetooth speaker w/ aux in
blissful pollen
Makes sense to me. Not a be all/end all type thing but would be nice to have it sound a little bit better. Maybe I should also 3d print a box 🙂
devout jolt
Makes sense to me. Not a be all/end all type thing but would be nice to have it ...
I think anything will sound better than a bare speaker!
lone axle
Is it possible to include non-python files in libraries that get distributed in the bundle. Like font5x8.bin file that some of the older pre-displayio libraries utilize by default i.e. https://github.com/adafruit/Adafruit_CircuitPython_framebuf/blob/7c59940395afe427f2282bc648abd39cf3b64b7b/adafruit_framebuf.py#L493 That file is included in the examples folder of this library, but not other libraries that extend the main driver. They need to be copied over separately from the examples, but it'd be nice if the default would be included in the installed "package" so that it can work out of the box without hunting down other files.
jaunty juniper
I think the bundler would work ? I'm not sure if it only takes python files, but probably not.
onyx hinge
@lone axle I don't know, but I'm nearly sure that non-python files can't be frozen in
I be filterin' yer square waves, fir sure
blissful pollen
After all, it's hip to be square
onyx hinge
old and busted
jaunty juniper
Is it possible to include non-python files in libraries that get distributed in ...
so that would actually require changes to circuitpython-build-bundles
RuntimeError: ('mpy-cross failed on', 'libraries/helpers/framebuf/adafruit_framebuf/font5x8.bin')
and I think too that it would make it un-freezable (it's not frozen in anything currently)
an alternative being to make it into bytes in a py file
blissful pollen
Would it make sense just to include it in all the libraries that need it? I know it is duplicating the one file, but I don't think it is something likely to change
jaunty juniper
in the examples ?
jaunty juniper
that would definitely be easy to do
blissful pollen
Yeah just include it in any example that needs it. Probably less risk then trying to include it in the library itself
lone axle
Yeah, I am in favor of including it in all libraries that ultimately default to using it which is pretty much anything from old non-displayio type of drawing on displays and adafruit_epd, everything with display.text() AFAIK.
encoding it as bytes in a py file so it can be mpy-crossed and included with the current tooling is an interesting idea.
tulip sleet
@blissful pollen @lone axle How about making it its own library? Then the other libraries would depend on it, and there would be only one copy to update if that was needed.
lone axle
Sounds like a nice idea to me. adafruit_bin_font or something. It could be added to if ever there are more (although newer displayio supports different font formats). The logic in frambuf could maybe be tweaked to avoid importing it if the user does happen to set their own font file.
jaunty juniper
I don't think you would want to add more fonts in it, it would blow up in size
tangentially related, I do think that we need an Adafruit repo of selected (open source licensed) fonts, referenced in related guides (like the custom fonts guide and adafruit_display_text labels guide)
tulip sleet
also if more than one dependent library wanted 5x8, there would be only one copy on CIRCUITPY
blissful pollen
I agree a separate library is a good idea. The import failing will provide a clear clue as to what is wrong.
(I may have not been able to find this exact file at least twice I recall when I was starting out)
tulip sleet
the library might just provide a directory to store the font file. Or it could have a small amount of code that would return the file path.
jaunty juniper
hey I would like to check that it's not my browser or content blockers, but does the search also not work in the library docs ?
Something like that:
https://docs.circuitpython.org/projects/httpserver/en/latest/search.html?q=html&check_keywords=yes&area=default
lone axle
hey I would like to check that it's not my browser or content blockers, but does...
same for me. It's missing jQuery
I remember a similar thing from the core docs. Sphinx-rtd-theme or something it relies on changed something about how jquery was (or isn't any more) included. It needs a seperate pip requirement for it now iirc.
jaunty juniper
I remember a similar thing from the core docs. Sphinx-rtd-theme or something it ...
yeah you mentioned there that jquery is injected in the page, but too late
which matches my observations
(see the issue) 👆
manic glacierBOT
slender iron
@tulip sleet haha me too. the ones that keep time in actual calendar values are terrible
tulip sleet
<@329766224093249548> haha me too. the ones that keep time in actual calendar va...
Any conversion time saved has been used up by all the engineers getting it to work or undoing the time format into something easily manipulable.
slender iron
100% I assumed it was kept for backwards compatibility
jaunty juniper
I shall retest and pay more attention this time
onyx hinge
hopefully this is the last for now ^^ 🙂
@slender iron sorry for making your review harder, I know I should force-push less than I do but my mind was totally somewhere else than thinking about my reviewer
slender iron
np 🙂
blissful pollen
I am hoping to test some of the synthio stuff this weekend (new midi controller was just delivered!) so if there is anything in particular Jeff you'd like me to look at or try just let me know
onyx hinge
@blissful pollen just have fun...?
blissful pollen
<@228322543507865601> just have fun...?
Will do! 🙂
onyx hinge
I have not tested all the combinations of things. the newest things are stereo/panning, ring modulation, and fir filtering so those probably have problems to be found. for sure if you give too many FIR filter entries your microcontroller will bog down and the audio will start skipping but there's really no fix for that
onyx hinge
if you collect any good groups of settings that'd be cool to start a list of somewhere
idle owl
@onyx hinge I have a note here to add a warning to the M7 guide about PWM Audio only working on A0 and A1, does this sound right to you?
I'm not sure where to add the warning. I have to update the template, and so I want to get it right on the first try.
Maybe after the wiring diagram section.
This is the page in question, if I understand it right. https://learn.adafruit.com/adafruit-metro-m7-with-airlift/pwm-audio
Doing it that way. Hope I'm on the right track here, heh.
slender iron
@tulip sleet think we can do 8.1 rc next week?
tulip sleet
kattni would like me to do a beta.3 today (or over the weekend), to help a guide that will come out soon. I am not sure we have all the 8.1.0 issues covered yet
but the remaining issues are network crashes, not sure we can do anything about those
so maybe yes, and push those forward
I am a little mystified by #7932 (AiThinker C3). what is different about this vs C3 QT Py?
@idle owl when is the guide you need the beta for being published?
idle owl
<@330227457296957440> when is the guide you need the beta for being published?
Checking, I have dates in Basecamp.
@tulip sleet Further out than I thought. May 24th.
So no rush, apparently. 😄
tulip sleet
well, then if we do an rc (see above) next week, will be in time no problem
idle owl
I don't know what related issues are already filed regarding what I was running into with this code on ESP32-S2, but I imagine someone might want to eventually look into it. I don't want to file a duplicate issue though, which is why I haven't. I don't have the cycles right now to go through what's already there and figure out how it's related, especially since I don't quite get what was going on in the first place.
tulip sleet
I don't know what related issues are already filed regarding what I was running ...
was that the ping crashing?
we fixed something about pings crashing
idle owl
was that the ping crashing?
Yes but seemingly only when having Adafruit IO time code involved.
You fixed that the espidf exception is now caught, so it doesn't dump into safe mode every time.
Meaning at least my code stopped being in safemode every 3 pings.
But the issue itself was not resolved.
It was reporting espidf Out of memory error every few pings or so. Which I passed on, and the code kept running, but I can't put that in a guide.
tulip sleet
idle owl
So when we figured out the issue wasn't present on S3, I switched and moved on.
idle owl
ok, so it's <https://github.com/adafruit/circuitpython/issues/7940>
Yeah I believe so
tulip sleet
yeah, I don't think we would hold up 8.1.0 for that, since it can be worked around
idle owl
I wouldn't think so
Also, S2 only.
So.....
¯_(ツ)_/¯
Typo of the day: PWN audio.
I don't even want to fix it.
crimson ferry
sounds like a pirate station
idle owl
Certainly does.
slender iron
@tulip sleet I was going to check another C3 and punt if it works.
lone axle
@proven garnet (he/him) I think the fix to this: https://github.com/adafruit/circuitpython/issues/7896 is a relatively simple minor addition to docs-requirements and a list inside of docs/conf.py. I'm going to try to confirm that on one library this evening.
Just wanted to check in with you to see if you already have anything in the works for this one so we don't duplicate effort. If you don't, I'll get prepared for a sweep to patch that in the libraries for next week.
(sorry if you got double pinged, I put this in the wrong room initially)
blissful pollen
@devout jolt just checking for your MIDI tests you go MIDI controller->computer computer->CircuitPython device?
devout jolt
<@352910176736772096> just checking for your MIDI tests you go MIDI controller->...
Correct. I'm using a DAW (Ableton Live in this case) just as a MIDI forwarder. I have also just used OS-level tools to do that too, but a DAW is usually easier. You could also use a hardware dealie like https://retrokits.com/shop/rk006/ Or have hardware MIDI interface, which is just as easy in the software but a bit more of a pain to wire up
blissful pollen
Thanks, just got an Akai MPK miniplus and been watching your synthio YT videos so want to see what this can do 🙂
devout jolt
Thanks, just got an Akai MPK miniplus and been watching your synthio YT videos s...
Totes! The MPK comes with a free DAW if you're on Mac/Windows that should work for MIDI routing if you want to stay all USB
proven garnet
<@135415127879974912> (he/him) I think the fix to this: https://github.com/adafr...
Nope, didn't notice a double ping, thanks for this! That sounds good, I didn't have a fix in mind and my web development is rusty to non-existent, so glad you found a solution 😁
Happy to help with any of the patching though, feel free to ping and/or DM me about it. I actually have a decent list of patches to apply so I'm also happy to incorporate it into any of those if it takes a load off your plate.
lone axle
Happy to help with any of the patching though, feel free to ping and/or DM me ab...
perfect, thank you! This change ended up doing the trick: https://github.com/adafruit/Adafruit_CircuitPython_Logging/pull/50/files I tested it on the Logging library specifically and it has resolved the problem on the docs page. Those changes are ready to roll out whenever you do the next one.
proven garnet
perfect, thank you! This change ended up doing the trick: https://github.com/ada...
Excellent! I'll prepare a patch and add you as a reviewer to confirm.
jaunty juniper
so I had issues with fetch-submodules and the silabs build, had to install git-lfs (has that been added to the guide ?) and now with a fresh new clone, make fetch-submodules pulls 9GB instead of ~3.5 in the past (when exactly I'm not sure)
wraith crow
so I had issues with fetch-submodules and the silabs build, had to install `git-...
git-lfs is included in the ports/silabs/README.md but I don't think there's a specific silabs build guide in the learning system yet.
jaunty juniper
the thing is, it seems that its absence makes fetch-submodules fail, and leave the submodules in a dirty state (I'm not trying to build silabs)
and that
❯ du -ksh ports/*
61M ports/atmel-samd
333M ports/broadcom
22M ports/cxd56
205M ports/espressif
200K ports/litex
673M ports/mimxrt10xx
33M ports/nrf
71M ports/raspberrypi
3,5G ports/silabs
781M ports/stm
400K ports/unix
Oh, I see what you mean. I've been blaming it on my ancient setup, but I usually have to run fetch-submodules/remove-submodules a couple times to get a clean state even with git-lfs installed. Perhaps the larger downloads is part of that problem as well though
onyx hinge
synthio feedback and questions
tulip sleet
so I had issues with fetch-submodules and the silabs build, had to install `git-...
Are you using a recent git that can do the partial clones?
jaunty juniper
I'm on 2.40.0, a fresh clone after I installed git-lfs worked to get a clean status
tulip sleet
also trying a fresh clone...
... it's taking a long time in git-lfs when cloning the silabs submodules
ok, yes, same thing in silabs port, 3.6G, and it took minutes to do that part
@jaunty juniper I'm not sure what we can do about this. Perhaps we could make make fetch-submodules be per port somehow.
For TinyUSB, Thach got rid of a lot the submodules and now has a Python script to fetch stuff
and I have 300Mb up/down Internet connection
jaunty juniper
yeah I don't know if we can define the "bigger ports" or "less used ports" and have them be retrieved only if explicitly asked for
something like fetch-submodules gets the most common, fetch-submodules <portname> retrieves the missing parts (and fetch-submodules all)
and remembers for subsequent calls ?
or each port's makefile detects if it was fetch-submoduled and warns ?
tulip sleet
it's possible to specify paths for git submodule so in theory could work fine
automating it through the makefile sounds interesting
tulip sleet
it could do the fetch if needed, and we could remove the need to do make fetch-submodules up front
manic glacierBOT
jaunty juniper
say, I don't see GPIO4 referenced in the pins.c of the Feather RP2040, which is what the BOOT button is connected to in the revision, what do we do for that ? Add it as BUTTON (knowing that it's not available on older models) or GP4 ? (There is a D4 which is GP6)
https://github.com/adafruit/circuitpython/blob/main/ports/raspberrypi/boards/adafruit_feather_rp2040/pins.c
crimson ferry
No such file or directory: 'arm-none-eabi-gcc' I thought that was what I just installed
wasn't there some discussion a few weeks ago about setting up a way to build for a specific board on github?
jaunty juniper
well that would be a path issue ?
jaunty juniper
wasn't there some discussion a few weeks ago about setting up a way to build for...
do you mean that ?
https://github.com/adafruit/circuitpython/pull/7594#issuecomment-1432525280
crimson ferry
do you mean that ?
<https://github.com/adafruit/circuitpython/pull/7594#issuecom...
yeah, that, I guess that should work?
jaunty juniper
the mac install doesn't give any instructions other than "Download and install the appropiate .pkg file for your Mac" but maybe they need adding to the PATH ?
I don't do it like that anyway, I have the zipped versions in a directory, and my build script manages the path as needed
crimson ferry
i don't know about that, I'm trying the GitHub action
worst case, I make a PR and test the artifact 🤪
YOLO
hopefully I didn't break the espressif local build with all this
crud, it did break
ok, that was obvious enough to fix
plucky tulip
Hello, I was trying to do something like JavaScript's setTimeout in CP, I found this article about multitasking https://learn.adafruit.com/multi-tasking-with-circuitpython/all-together-now, but the solution there is very manual. Is there a library for doing something like this (image below), and if not, would this be a good candidate for a new lib e.g. adafruit_eventloop? I would be happy to make it if that is the case.
jaunty juniper
Hello, I was trying to do something like JavaScript's `setTimeout` in CP, I foun...
like a helper for asyncio or something different ?
there's the circuitpython_schedule module too
https://cognitivegears.github.io/CircuitPython_Schedule/
plucky tulip
there's the `circuitpython_schedule` module too
<https://cognitivegears.github.i...
Thank you 
brazen hatch
This should be api-compatible with console objects and be able to be used by esp32 and esp32c3.
manic glacierBOT
torpid mantle
brazen hatch
Can someone in the following days help me finalize the m5stack timer camera circuitpython port?
It's a board sold by adafruit (yes, https://www.adafruit.com/product/4959), yet has no circuitpython port.
I have finished the rest of the port (led, psram, i2cs), I just need to do the camera part.
I have never used espcamera so it's really hard for me.
manic glacierBOT
manic glacierBOT
brazen hatch
It's simple enough I should be able to make it a console object.
From there, plugging it into the os is a piece of cake.
lone axle
Are there special steps needed in espressif port to have a debug build that can connect to WIFI? It seems that CIRCUITPY_WIFI_SSID and password in the settings.toml is not auto-connecting in the debug build. Trying to connect manually in user code reports this error: ConnectionError: No network with that ssid but I know the ssid does exist and has the correct name. The non-debug build on the same device does connect successfully to that network.
brazen hatch
Are there special steps needed in espressif port to have a debug build that can ...
Hmm, I in the past used to only have debug builds on my boards (I blame stackless), and all that worked fine.
What board are you using?
lone axle
Feather S2 TFT
brazen hatch
Hmm, no idea.
I was also using s2 back then.
It's been a good while since I last used debug though.
If you scan, will it show up?
lone axle
I will try to find the scanning code and give it a try. Up to now I've just copy pasted the connect code, and eventually removed that when automatic connect from settings.toml start happening.
brazen hatch
From ljinux wifi:
def scan(self):
"""
scan and store all nearby networks
"""
if wifi.radio.enabled:
net = dict()
for network in wifi.radio.start_scanning_networks():
sec = None
if len(network.authmode) is 3:
sec = str(network.authmode[0])
sec = sec[sec.rfind(".") + 1 :]
else:
sec = "Unknown"
net.update({network.ssid: [sec, network.rssi]})
del sec
wifi.radio.stop_scanning_networks()
return net
return list()
@lone axle does the ConnectionError: No network with that ssid persist with retries? I can try a debug build and see what happens.
lone axle
It does seem to persist. I tried maybe 3-5 times only though before switching back to non-debug to confirm
Tried scanning, but have fallen a little deeper into the rabbit hole: I got a single result the first time with a network that is visible around me, but not the one I was using. It did not show the actual network that I've been using. I was trying to get it to retry and now I've ended up into a (different from the one I was hunting I think) hard fault
brazen hatch
I remember there was an issue back in the day that reduced effective range, but that was really old.
It was way before stackless.
lone axle
In my case, based of the signal strength reported in my PC at least the real network I am using should be closer than the one that it did happen to list when I scanned.
brazen hatch
lone axle
Interesting. I have had another handful of retries now back in debug mode. I've seen 2 networks a few times, but vast majority of scan attempts only show 1. And neither of the 2 shown have been the actual one that I use.
I will try to move it closer to the router, but I'm a bit skeptical of the distance being a factor. It's already only ~6ft or so from the router, and I'm fairly certain that the network that is showing up most commonly in the scans is one that isn't from an AP inside of my house (I hope anyway 😅 ).
crimson ferry
this debug build is a few days old, but```
W (120290) CP wifi: connected
I (120310) wifi:<ba-add>idx:0 (ifx:0, 00:1a:dd:re:da:ct), tid:0, ssn:1, winSize:64
I (120330) wifi:AP's beacon interval = 102400 us, DTIM period = 1
I (121290) esp_netif_handlers: sta ip: 192.168.6.148, mask: 255.255.252.0, gw: 192.168.4.1
W (121290) CP wifi: got ip
I haven't seen an issue with running wifi code
jaunty juniper
does the debug build use a different tx_power ?
lone axle
Weird, now it's about 10 inches away from the AP and I am consistently getting an empty dict back from scan.
brazen hatch
does the debug build use a different `tx_power` ?
Doesn't seem the value is set by CP at all, just went to ports/espressif/common-hal/wifi/ and I can't find much with tx_power.
It's probably a default
crimson ferry
are you basically at tip of main?
lone axle
is there any kind of 5G vs. 2.4G that comes into the mix (forgive me if it's an ignorant question tbh I treat WIFI as mostly magic)?
I don't normally mess with it at all, and I'm not sure which my AP is.
I am at a fully updated main.
brazen hatch
I will build debug for all my esp boards.
brazen hatch
is there any kind of 5G vs. 2.4G that comes into the mix (forgive me if it's an ...
'retty sure esp only does 2.4
crimson ferry
yes, only 2.4
I'll test a fresh rebuild
what did you disable to get the debug build to fit?
lone axle
what did you disable to get the debug build to fit?
CIRCUITPY_ULAB = 0
crimson ferry
same
lone axle
My phone seems to think the network (that I'm trying to connect to) is 5G, though I would guess perhaps the AP outputs both. I'm trying to find a way to confirm that.
I would assume it does a compatible one though because networking does work fine for me in standard builds on this device.
crimson ferry
I tried that, but there seem to be a lot of dependencies there
brazen hatch
Fair nough.
lone axle
Okay, yep. I've confirmed with a different scanning utility I have 2 networks with the same name, one 5G and one 2.4G.
brazen hatch
CP will only see the 2.4 one, the other should not exist as far as it's concerned.
crimson ferry
takes me a while to build, can't do -j because... reasons
brazen hatch
Built on my s2 board, loading now.
crimson ferry
(errors with -j, some script issue isn't managing the parallelism right?)
brazen hatch
(errors with -j, some script issue isn't managing the parallelism right?)
top of main is fine for me
I did -j12 rn.
Using my pc, not the poor pi400 rn.
crimson ferry
Adafruit CircuitPython 8.1.0-beta.2-30-gca01200b4-dirty on 2023-05-15; ESP32-S2-DevKitC-1-N4R2 with ESP32S2 DEBUG build connects fine, scan shows a few dozen APs (most of them mine) ...RSSI values from -33 to -86