#circuitpython-dev
manic glacierBOT
thorny jay
@thorny jay Is this what you meant? https://github.com/boochow/micropython-raspberrypi
@lapis hemlock Yes, I tested that and that was a revelation for me (the I tested bcm64 and I started to love bare metal). That particular port of micropython was advertised on Adafruit blog and/or newsletter. [It also got me interested in some maker from japan, not that I understand anything they write, but some picture speak for themself.] I may have an ethical issue with Raspberry Pi and the fact you cannot run on that hardware without some binary blob (the GPU code?) and that mean it is not totally free/open. I may have similar issues with Expressif and the ESP, I am not up to date, but IFAIK there are openness or licence issue with the fact that you run under their operating system somehow. So I am very currious at Adafruit positioning on this topic.
manic glacierBOT
I'm still not fully familiar/comfortable with the transitions among these states; the naming is a bit different in each port and for each type of managed resource and I'm never entirely confident the semantics are the same. (As a larger issue, it wouldn't hurt to document this for developers reference and to ensure consistency)
My question is whether the state transtion ...
onyx hinge
thanks dan
onyx hinge
with an alpha3-ish build on trinket m0, I'm seeing a problem where IF my code contains a sleep() call, then it's likely to NOT auto-reload. Does this ring a bell for anyone?
stuck elbow
what does it do instead?
onyx hinge
continues running until I hit ctrl-c
onyx hinge
actually it seems to eventually respond, it just takes quite awhile
crimson ferry
@onyx hinge I had noticed something similar on esp32s2, hadn't tied it to alpha-3... it seemed to me that in the past, ctrl-C would interrupt blocking processes (like sleep), but now they don't.
I see it most on wifi activity ...could be a completely separate thing
lapis hemlock
@lapis hemlock Yes, I tested that and that was a revelation for me
@thorny jay I definitely see your reasoning as to why the raspberry pi could be interesting, while I have a couple of misgivings. One is the current consumption, the second is that the chip itself is not accessible to mere mortals. I believe, it is sold exclusively to the raspberry foundation. Not that it would be too useful anyway, given that it is a BGA chip, so good luck with developing your own board.😉> So I am very currious at Adafruit positioning on this topic. @slender iron pointed out that this is not their priority.
lone axle
Is there some thing that needs to be done to "release" SCK? I'm winding up with ValueError: SCK in use that is resulting from the use of board.SPI()
it happens sometimes when my device resets due to saving new code. But doesn't happen every time.
lone axle
My trouble is potentially related to the use of sdcardio. In my case I've got a display and sdcard both on SPI. After unplugging the sdcard I haven't had it happen again yet through many restarts.
stuck elbow
do you have a full traceback? there shouldn't be anything using SCK explicitly
lone axle
code.py output:
Traceback (most recent call last):
File "code.py", line 30, in <module>
File "/lib/adafruit_featherwing/keyboard_featherwing.py", line 60, in __init__
ValueError: SCK in use
this line in keyboard_featherwing.py is
spi = board.SPI()
@lone axle when you construct the SPI bus for sdcardio, how do you construct it?
lone axle
sd_cs = board.D5
self._sdcard = None
try:
self._sdcard = sdcardio.SDCard(spi, sd_cs)
vfs = storage.VfsFat(self._sdcard)
storage.mount(vfs, "/sd")
except OSError as error:
print("No SD card found:", error)
displayio and sdcard (or adafruit_sdcard) on a single bus should work, but (A) the SD card must be initialized first before any other SPI device according to the SD card standard (even though they have distinct chip select lines), and (B) for CircuitPython's purposes, it had better be the same SPI bus object.
@lone axle is spi the same object as spi = board.SPI() before?
lone axle
Yes, that is correct.
I do think display is done first, I will swap them
oh, and the touchscreen is using SPI as well it looks like
onyx hinge
The effect of not doing (A) is that the SD card enters an undefined state, that would not give a "ValueError: SCK in use".
the fact that a display object can be kept over from a previous "run" of circuitpython could be relevant and maybe there's a wrinkle of that I have not thought through.
lone axle
moving the sdcard intialization (code from above) to happen before the touchscreen and displayio initialization's seems be a bit worse in this case. On the sdcard failing I notice the error is different No SD card found: [Errno 11] EAGAIN and now even when the sdcard is not plugged in I am getting the SCK in use error on resets
The initialization order I used originally was: Touchscreen, display, sdcard
onyx hinge
Interesting. At this point you may want to do more than soft-reset your board, it's clearly gotten to an undesirable state and SOMETHING undesirable is being retained across soft-reset
lone axle
with the original order the sdcard error looks like this: No SD card found: no SD card
Yep I have been using the on/off switch on the featherwing once I get to that state to get back out of it
I will see what else I can figure out and then try to put together a more concise example that reproduces it.
onyx hinge
OK. It would also be useful to know if at the initialization stage adafruit_sdcard behaves any differently. I know there are bugs when getting past the initialization stage, e.g., when you use the OnDiskBitmap
There is an important thing needs to be considered that the MMC/SDC is initially NOT the SPI device. Some bus activity to access another SPI device can cause a bus conflict due to an accidental response of the MMC/SDC. Therefore the MMC/SDC should be initialized to put it into the SPI mode prior to access any other device attached to the same SPI bus. http://elm-chan.org/docs/mmc/mmc_e.html#spibus
this may only be seen the first time after power-on because (successfully) putting the SD card into SPI mode is maintained until power-off, there's no sequence to go back to SDIO mode.
otherwise, it may be stuck in an unusable state until powered off (including being ejected and re-inserted)
turbid radish
?Serverinfo
digital shoreBOT
adafruit
turbid radish
?Serverinfo
digital shoreBOT
adafruit
analog bridge
Anyone online with an esp32s2 board and a multimeter handy ?
crimson ferry
@analog bridge yes
analog bridge
Hi @crimson ferry, I need to confirm an issue I am having with internal pullup resistors
I am currently working on implementing deep sleep support for CPY 
crimson ferry
sounds like I may need my magnifying goggles for this one
analog bridge
What board do you have ?
crimson ferry
any... Saola, Cucumber, FeatherS2, or Bast
analog bridge
great.... this requires re-flashing of CPY
crimson ferry
ok, np
analog bridge
I can provide you .bin for Saola with deep sleep enabled along with my code.py
Give me a minute....... 
crimson ferry
wroom or wrover? I'll start getting ready
analog bridge
wrover will work
thorny jay
?Serverinfo
@turbid radish I am sure that if you try again, you will get 25000. 🙂
turbid radish
?serverinfo
digital shoreBOT
adafruit
turbid radish
nope
thorny jay
Patience...
crimson ferry
ok, will take me a couple
turbid radish
?serverinfo
digital shoreBOT
adafruit
analog bridge
np, let me provide you some more info
crimson ferry
"writing..."
analog bridge
little disclaimer...... this code will put CPY in continuous loop and that is an expected behavior
crimson ferry
ok, ready, haven't loaded code.py yet (got a memory error with my prior code, that's new)
analog bridge
I need you to probe GPIO0 and check the pullup voltage you get.
analog bridge
With code.py running. After you obtain the results just press the boot button when rgb led turns yellow to enter safe mode
crimson ferry
and do I need to keep both USB active?
analog bridge
USB is not a issue....
analog bridge
Yup! it goes into sleep but keeps on resetting due to GPIO0 not being pulled up properly.
crimson ferry
ok, hang on,
analog bridge
On my board I get 1.36V on GPIO0 during sleep and I think it is below the digital threshold to be counted as a HIGH signal.
analog bridge
Does the board reset in a loop?
crimson ferry
yes, yellow, then green... voltage on GPIO0 rises during that time
actually it falls from 2.7 to 2.3, then rises to 3.3, then settles back at 2.7
I don't have a scope or LA
analog bridge
I get this issue on GPIO2,4 as well.
crimson ferry
I'll measure those across resets
analog bridge
First you need to update the pin in code.py by going into safe mode.
crimson ferry
how do I know it's in safe mode? it goes from yellow to back to green even when I RST during yellow
analog bridge
RST method doesn't work on the s2 you need to press boot button when yellow
crimson ferry
oh right, just found that one inadvertantly 😉
OK, in safe mode now
did I need to be in safe mode for the GPIO0 measurements, or just after?
analog bridge
safe mode is just for changing pin numbers
crimson ferry
ok...
I lost CIRCUITPY and haven't gotten it back, let me try a few things
I probably need to reflash then reload a changed code.py with the new pin
I can't get CIRCUITPY back, even with power cycle
analog bridge
no CIRCUITPY in safe mode ?
crimson ferry
no, it disappeared when I dropped code.py onto it the fist time and I haven't seen it since
give me a couple to reflash
analog bridge
that's odd...... I see CIRCUITPY when in safe mode (i.e. pulsing yellow)
crimson ferry
power cycle, safe mode, not safe mode, couldn't bring it back, reflashing now
what change should I make in code.py?
IO0?
to IO2
analog bridge
yes, in io_alarm.PinState() change IO0 to IO2
crimson ferry
one sec
turbid radish
?serverinfo
digital shoreBOT
adafruit
crimson ferry
@analog bridge sorry, fiddling with native USB to try to get that back
analog bridge
np.... 👍
crimson ferry
I've tried everything I can think of - can't get native USB to show up, I'm going to reflash the previous .bin, remove code.py, then reflash today's .bin, will take a few minutes
on the bright side... I never noticed before that the write location updates 🙂 Writing at 0x00050000... (42 %)
I only ever watched the %
analog bridge
what command are you using to flash the .bin ?
crimson ferry
esptool_3.0-dev.py --port /dev/tty.SLAB_USBtoUART --chip esp32s2 --baud 115200 write_flash 0 ~/pathto/firmware.bin is what I always use
I have CIRCUITPY back, with previous .bin
do you have diff params I should use, or continue with reflashing new firmware?
analog bridge
I see nothing strange
turbid radish
?serverinfo
digital shoreBOT
adafruit
solar whale
it's going the wrong way 😉
turbid radish
fix it
crimson ferry
I could make a troll army of new accounts
turbid radish
it'll likely tick over tomorrow
solar whale
Thats called server discovery
turbid radish
That's off, too much spam
solar whale
I know - just commenting on the"troll army" -- glad it's off.
crimson ferry
@analog bridge OK, CIRCUITPY disappeared again when I renamed code.py (with IO2 now)
green and yellow alternating... not sure what thsi is
analog bridge
what does the multimeter have to say.....
crimson ferry
basically 0V
on IO2
board is continually resetting from the looks of debug USB, that's expected?
analog bridge
yes, but 0V on IO2 is not
crimson ferry
but IO2 voltage sticking to ~0V
oops, sorry, my bad... 1.455V
stable through the board resets
(I had ground in the wrong place on my breadboard)
analog bridge
is the safe mode working now?
analog bridge
no need.... It probably has an external pullup, that is why the higher voltage reading
could test IO4 though....
crimson ferry
ok, one sec...
IO4 is 1.46V, stable throughout the resets
I'm going to do IO0 again, since it's a critical one and I don't have confidence the board was in the right state before
analog bridge
no need....
analog bridge
let's run the test for an IO that I am sure is working
pick any one of the gpio till IO20
crimson ferry
I'll do IO7, it's pretty general-purpose
from ~150mV in safe mode to 3.27V when code is running, but no green-yellow, and no indication of board resetting in debug USB (lost CIRCUITPY again)
analog bridge
that's good
the tests confirm that io 0, 2, 4 have something strange going on with them
crimson ferry
always good to know 😉
analog bridge
Thanks @crimson ferry for all the testing
crimson ferry
sure, np, it's all sitting here so just ping me if there's anything else I can do
analog bridge
nothing for now.... will meet you tomorrow in the CPY meet
idle wharf
@turbid radish If a few more people retweeted your post it might help...
https://twitter.com/anne_engineer/status/1307750782981279755
turbid radish
yes, but I can't force them to retweet
idle wharf
people in channel 👀
turbid radish
?serverinfo
digital shoreBOT
adafruit
atomic summit
2 questions...
- Is there a way to enumerate board? I just need a way to go over every IO on a board, without having to know what the IO number or Board pin reference name is
- Is there a DigitalInOut all in 1 constructor like
p = DigitalInOut( pin, direction.INPUT, pull=DOWN)
Right now I'm building a list of DigitalInOut and then setting each to be an Input
azure iron
Hey I'm wondering if Circuitpython is currently working (with plug-and-play USB) with the ESP32-S2 yet. I know it was being worked on but I can't find if it works yet.
thorny jay
@atomic summit dir(board) ?
crimson ferry
@atomic summit I do pretty much the same with for pin in dir(microcontroller.pin): if isinstance(getattr(microcontroller.pin, pin), microcontroller.Pin): for alias in dir(board): if getattr(board, alias) is getattr(microcontroller.pin, pin): to filter out non-pins (and unexposed pins)
@azure iron the latest unstable builds on https://circuitpython.org have basic esp32-s2 features, still very much in development
azure iron
@azure iron the latest unstable builds on https://circuitpython.org have basic esp32-s2 features, still very much in development
@crimson ferry thank you.
does this mean specifically the tinyusb integration, or do you mean circuitpython in general has limited feature control over the S2?
crimson ferry
@azure iron re:tinyusb, I don't know. I meant the latter... https://circuitpython.readthedocs.io/en/latest/shared-bindings/support_matrix.html should be pretty close to the modules present
azure iron
oh. this lsit doesn't include the original eSP32...
crimson ferry
original ESP32 is not supported in CircuitPython due to lack of native USB
atomic summit
Oh... hmm, so I can't write to to the FS from inside CPY due to readonly - but if I switch the boot.py out with one that allows it, and reboot, then I can, but I can't get if off again as I cant write to the FS from USB anymore 😦
I can't use the "check IO" approach like in the learn guide
crimson ferry
safe mode?
atomic summit
Is there no way to do this from inside CPY?
crimson ferry
there's always nvm or sd 😉
atomic summit
and no rshell... no pyboard that I can get working
so no raw_rpl
so no way I can find to have my test jig "talk to the DUT" without having to setup software serial on both sides.... but that assumes working IO... than I cant check without having tested the board first.
Also, I now cant revert this boot.py change... only way to fix it is to re-flash CPY on it 😦
crimson ferry
is safe mode post-boot.py?
crimson ferry
right, I just meant to get out of writable-mode
atomic summit
how do we get into safe mode again ?
crimson ferry
on esp32-s2 i think just hit boot after reset, while yellow
pastel skiff
@turbid radish Hi, you helped me with the circuit python day newsletter post and the subsequent notification when I was impacted by the fires and power outages in Calfornia, FYI I've rescheduled our event to 9/24 https://hackerlab.org/en/blog/read/1376035664/hacker-lab-is-celebrating-circuit-python-day--how-you-can-join
atomic summit
My APA shows nothing on my board during boot
I re-flashed it
still, there needs to be a better way of either
- getting in and out of readonly at runtime
- being able to talk to CPY from a host via pyboard or similar
I understand the dangers of enabling write from CPY when the board is writable from USB, but being able to write to the FS is fundamental part of python.
Is #4 even possible?
or do I instead just delete the boot.py ?
crimson ferry
if not you could have a semaphore in a separate file or file/dir name, to indicate readonly or not
onyx hinge
atomic summit
@onyx hinge Yup, that show the bottleneck I'm referring to.
Plus I can't be constantly rebooting the DUT to get it in and out of readonly mode as that adds 5-6 seconds every reboot for the FS to come up so it can be re-mounted.
crimson ferry
darn, my next suggestion was going to be to boot with random readonly, code can detect mount state, and with a few reboots you can always get to one or the other 😉
onyx hinge
to my knowledge that's the best procedure right now.
atomic summit
Ok, I'm going to need to re-make all of this in c++ and do all test jig code that way (YUK) and then do a final flash of CPY for shipping.
raw_repl is the ideal solution here
I can't use os.remove() to delete the boot.py after I'm done writing my file 😦
crimson ferry
that's interesting... locked since it's executed presumably? (you could delete or change a semaphore file)
atomic summit
but that still requires a reboot to activate so I can then clear off all of the files and load shipping files
If I have to, then so be it... but it's really an ugly solution
idle wharf
@atomic summit your test process would be a great live stream
atomic summit
I've covered how I do it all on streams before...
idle wharf
I will have to check the back catalog then
atomic summit
I actually did one yesterday about it... but deleted it afterwards as it was full of dropped frames.
hearty forge
@atomic summit - is it possible to set up the flash with another partition that is not exposed to windows but can be written by cpy? (Apologies if this is not even vaguely possible!)
onyx hinge
I think it would be great to look into that for express-type boards with lots of storage
we can mount multiple FAT partitions inside circuitpython (that's how sd cards work) but only one can be exported over USB, and there's no code to subdivide/partition the built in storage yet. (instead, a dummy partition table is generated for the benefit of the PC it's plugged into, and the whole flash is treated as the single partition)
hearty forge
I can imagine that internal flash partitioning would be very useful. Can imagine situations where even a few 100k for data being extremely useful for test jigs/data loggers etc. Definitely for the future - lots of cool things becoming possible!
atomic summit
@hearty forge @onyx hinge I my case an extra small partition wont help if I cant get to it via USB. The purpose of me writing to the flash from inside CPY is to get the test results out - because I can't just talk to the board with raw_repl like I hoped.
Ok, I think I got it deleting the boot.py, so after a second reset it goes back to readonly again
slender iron
circuitpython does still have raw repl. I'm not sure why it doesn't work for you
tests on adafruit boards are usually done by flashing a bin and then flashing circuitpython
atomic summit
Hey @slender iron - I cant get pyboard.py or rshell to work... it connects, but then errors trying to initialise raw_repl
Sorry to bother you on a Sunday 😦
Hmm, do you have a modified pyboard.py ? Looks like the responses pyboard is looking for are MP ones.. not what CPY dumps out. That could be why I can't get it working
raven canopy
@atomic summit the symlink in circuitpython's repo is tests/pyboard.py -> tools/cpboard.py. micropython's is tests/pyboard.py -> tools/pyboard.py.
atomic summit
Ahh, right, so you have a modded version. Makes sense, thanks @raven canopy !
Ok, cpboard imports some extra stuff. Where do I fins sh and shutil? they just pip installable ?
raven canopy
also, i gather you're using an ESP32-S2. you'll need to add the VID/PID to the lookup dict.
yes, sh is pip'able. shutil is in stdlib, iirc.
atomic summit
Thanks. Now failing at disk.copy("skip_if.py")
FYI - I'm using this on a rPI.
So I don't have the circuitpython repo on here and I'd like to keep t as lean and empty as possible.
right, that's in tests.
raven canopy
i haven't run the existing test suite against a board, so i prob won't be any help there.
atomic summit
Does the VID/PID lookup want the exact board name from ports/boards?
raven canopy
i don't think so. its just the key value that gets passed in.
raven canopy
yep. you'll need to sudo apt install pmount and make sure to add your account to the plugdev group.
no docs that i know of. 😦
atomic summit
make sure to add your account to the plugdev group.
The what to what? lolz
raven canopy
hehe. let me look up the command; i always forget it. basically, the user account needs permissions to plugdev group to access pmount without sudo.
atomic summit
oh, the user group on the rpi ?
raven canopy
yeah
atomic summit
it seems to be kinda working.. I'm able to call exec and I can see it has "run it" - but errors with a timeout
raven canopy
which line does the traceback point to? i think i know the cause of that...
atomic summit
well the exception is thrown on 442
but it's one of Python's During handling of the above exception, another exception occurred: gifts
last stacktrace before that was line 75
and I've not done anything with plugdev - but not sure this is related
raven canopy
@atomic summit what version of pyserial do you have installed? latest (3.4) should work. and are you calling exec inside a context block?
slender iron
sorry, I was just jumping on for a bit. will be around this week
raven canopy
try inside a context then:
board = pyboard.CPboard....
with board:
board.exec("import board")
without the context, it doesn't open the serial connection. rather, the context manager opens the serial connection. might be possible to also directly call board.open().
atomic summit
I don't think it's context-able"
AttributeError: __enter__``` pyb = cboard.Pyboard( device, baudrate=serial_speed )
with pyb:
output = pyb.execfile(filename)
return output```and yes I called my rPI version cboard not cpboard
by accident... 😉
raven canopy
ahh. sorry about that. you're using Pyboard. i've always used it one layer deeper with CPboard.
atomic summit
so I notice that a fresh flash of CPY firmware on a board has a code.py in the FS. how does one include code like that in in the firmware for inclusion in the FS?
Ok, I'll have a look at a level deeper.
if I can include my test stuff in the firmware, including the boot,.py that makes the FS writeable, then I only need to reboot once at the end.
raven canopy
this is the PR that introduced it, i think: https://github.com/adafruit/circuitpython/pull/2299
atomic summit
thanks!
turbid radish
?serverinfo
digital shoreBOT
adafruit
blissful palm
I just pulled from circuit python to update my local and tried rebuilding make BOARD=espressif_kaluga_1 and I get a bunch of esp-idf errors : starting with 'esp-idf/components/lwip/lwip/src/include/lwip/sockets.h:747:5: error: "LWIP_COMPAT_SOCKET_INET" is not defined, evaluates to 0 [-Werror=undef]'
and of course 'cc1: all warnings being treated as errors'
what am I missing, what changed..
lone axle
Just double checking, did you run these commands?
git submodule sync --quiet --recursive
git submodule update --init
yup
I just cloned a new CP. now I'm getting a different issue: -- Configuring done
-- Generating done
-- Build files have been written to: /home/awood/cp-alloy/circuitpython/ports/esp32s2/build-espressif_kaluga_1/esp-idf
QSTR updated
Traceback (most recent call last):
File "../../py/makeqstrdata.py", line 523, in <module>
encoding_table = compute_huffman_coding(translations, args.compression_filename)
File "../../py/makeqstrdata.py", line 241, in compute_huffman_coding
o = ord(atom)
TypeError: ord() expected a character, but string of length 0 found
../../py/py.mk:345: recipe for target 'build-espressif_kaluga_1/genhdr/qstrdefs.generated.h' failed
I cloned the new one in case i messed with something in the old one
It is 'main' branch I should be on right for esps2?
balmy stirrup
@blissful palm what ports dir are you in?
blissful palm
esp32s2
balmy stirrup
hmm
blissful palm
interesting if I go to CP top lever and issue : 'make -C mpy-cross' I get: File "../py/makeqstrdata.py", line 523, in <module>
encoding_table = compute_huffman_coding(translations, args.compression_filename)
File "../py/makeqstrdata.py", line 241, in compute_huffman_coding
o = ord(atom)
TypeError: ord() expected a character, but string of length 0 found
../py/py.mk:345: recipe for target 'build/genhdr/qstrdefs.generated.h' failed
make: *** [build/genhdr/qstrdefs.generated.h] Error 1
make: *** Deleting file 'build/genhdr/qstrdefs.generated.h'
make: Leaving directory '/home/awood/cp-alloy/circuitpython/mpy-cross'
so this isn't esp s2specific
I have some other issue
balmy stirrup
Worse case you can try to re-clone the repo.
blissful palm
I just did
balmy stirrup
oh ok in case you don't have the reference https://learn.adafruit.com/building-circuitpython/build-circuitpython
balmy stirrup
I am about to try that step myself.
I just pulled the repo and will let you know what happens on my side.
blissful palm
thanks
blissful palm
I am on windows but I am doing this in WSL
balmy stirrup
ok cool.
blissful palm
Everything was building OK last week
balmy stirrup
I was just able to successfully run.
make -C mpy-cross
blissful palm
clearly there have been some changes or I getting something mixed up
supple gale
I generally do the following when making for the ESP32S2
git submodule update --init
git submodule update --init --recursive ports/esp32s2/esp-idf/
git submodule update --init lib/tinyusb/
then a git status to make sure everything looks good
cd ports/esp32s2
blissful palm
Let me try that @supple gale
supple gale
and a make clean to get the esp-idf up to speed
also depending on changes to the esp-idf i do another install.fish (or install.sh) depending
its like an incantation,
blissful palm
make clean at what level?
supple gale
at the ports/esp32s2 with the usual BOARD=
i will try later. But usually works.
I'm using windows WSL2 with a Ubuntu20.04 instance,
blissful palm
OK tried all of that @supple gale no joy same error : -- Configuring done
-- Generating done
-- Build files have been written to: /home/awood/cp-alloy/circuitpython/ports/esp32s2/build-espressif_kaluga_1/esp-idf
QSTR updated
Traceback (most recent call last):
File "../../py/makeqstrdata.py", line 523, in <module>
encoding_table = compute_huffman_coding(translations, args.compression_filename)
File "../../py/makeqstrdata.py", line 241, in compute_huffman_coding
o = ord(atom)
TypeError: ord() expected a character, but string of length 0 found
../../py/py.mk:345: recipe for target 'build-espressif_kaluga_1/genhdr/qstrdefs.generated.h' failed
make: *** [build-espressif_kaluga_1/genhdr/qstrdefs.generated.h] Error 1
make: *** Deleting file 'build-espressif_kaluga_1/genhdr/qstrdefs.generated.h'
it's weird
balmy stirrup
What specific ESP32 device do you have @blissful palm ?
supple gale
seems its a Kaluga
blissful palm
I have my own board and also a Kaluga
onyx hinge
Good morning @blissful palm -- I have recently been working in that script makeqstrdata.py and perhaps there are some other problems remaining. Can you please tell me what running the command locale in your terminal says? Pasting the first two lines of output should suffice.
supple gale
i will make for a Kaluga. make -C mpy-cross worked
balmy stirrup
oh ok
blissful palm
@onyx hinge : LANG=C.UTF-8
LANGUAGE=
LC_CTYPE="C.UTF-8"
LC_NUMERIC="C.UTF-8"
LC_TIME="C.UTF-8"
LC_COLLATE="C.UTF-8"
LC_MONETARY="C.UTF-8"
LC_MESSAGES="C.UTF-8"
LC_PAPER="C.UTF-8"
LC_NAME="C.UTF-8"
LC_ADDRESS="C.UTF-8"
LC_TELEPHONE="C.UTF-8"
LC_MEASUREMENT="C.UTF-8"
LC_IDENTIFICATION="C.UTF-8"
LC_ALL=
onyx hinge
OK that looks good, not likely to be the problem. puzzled
balmy stirrup
It's weird as you re-cloned the repo you should be able to do the make -C mpy-cross without issue.
supple gale
yes
onyx hinge
Let's find out whether the problem is due to a recent change or not. In the top directory, please make -C mpy-cross clean, then git checkout 6.0.0-alpha.3, then make -C mpy-cross.
this will build 6.0.0-alpha.3 which is before my recent tinkering
balmy stirrup
Good call @onyx hinge
supple gale
i'm working off of tannewt esp_flash_sizes branch. so not really applicable here
blissful palm
ok that looks good : make: Entering directory '/home/awood/cp-alloy/circuitpython/mpy-cross'
Use make V=1, make V=2 or set BUILD_VERBOSE similarly in your environment to increase build verbosity.
QSTR updated
text data bss dec hex filename
210850 21048 976 232874 38daa mpy-cross
make: Leaving directory '/home/awood/cp-alloy/circuitpython/mpy-cross'
under build 6.0.0-alpha.3
onyx hinge
OK, at least we've established that it's a new problem.
what's your python3 --version?
blissful palm
Python 3.6.9
supple gale
It did build here off of tannewt branch, not sure where he is versus main
onyx hinge
I think most of us are using systems with Python 3.7 or newer, but we aren't intentionally incompatible with 3.6.
supple gale
3.8.2 here
onyx hinge
I don't use Windows so I don't have the ability to test with WSL/WSL2 systems
supple gale
it should generally be the same as Ubuntu or whatever instance the windows user is running
the only issue is flashing as that has to be done under windows
onyx hinge
tannewt's branch esp_flash_sizes is before the most recent changes to makeqstrdata.
blissful palm
Just checking for updates
onyx hinge
@blissful palm If there is some content in the file mpy-cross/build/genhdr/qstrdefs.generated.h after that error occurs, can you attach it for me to see?
blissful palm
Just apt-get updated etc.. no change on the python3 front
do you just want that file @narrow dirge E.?
onyx hinge
yeah if there's something useful I think that's where it'll be
blissful palm
How do you want me to get it to you?
onyx hinge
you can attach files on discord with the "+" button
blissful palm
Here is the generated file @narrow dirge E.
onyx hinge
OK, I think that's the one from when you checked out the old version and built successfully. Can you git checkout main, then go through the make -C mpy-cross clean and make -C mpy-cross step again, then finally attach the file after you get the build error
blissful palm
Ah yes sorry my bad hold on
supple gale
i just did a pull from adafruit main and doing a build for the kaluga. This is different repo from previous
Worked
onyx hinge
absolutely no problem!
blissful palm
no file
supple gale
git describe --tags --dirty
6.0.0-alpha.3-339-g8eda9174b
onyx hinge
OK, cd mpy-cross, and then run manually at the terminal (hope you can cut & paste this!): python3 ../py/makeqstrdata.py --compression_filename build/genhdr/compression.generated.h --translation build/genhdr/en_US.mo build/genhdr/qstrdefs.preprocessed.h > build/genhdr/qstrdefs.generated.h
I think make tried to be helpful by deleting the output file, since it encountered an error creating the file.
blissful palm
hdr/compression.generated.h --translation build/genhdr/en_US.mo build/genhdr/qstrdefs.preprocessed.h > build/genhdr/qstrdefs.generated.h
Traceback (most recent call last):
File "../py/makeqstrdata.py", line 523, in <module>
encoding_table = compute_huffman_coding(translations, args.compression_filename)
File "../py/makeqstrdata.py", line 241, in compute_huffman_coding
o = ord(atom)
TypeError: ord() expected a character, but string of length 0 found
onyx hinge
oh hey if I install old Python3.5 I can reproduce this problem locally!
that's super helpful
blissful palm
hah !
supple gale
I have a Mac I could test later on. Although I have a feeling this will be resolved quite soon 😉
onyx hinge
OK, I know what the problem is.
blissful palm
yay!
onyx hinge
do you want to hear about obscure things or just a change you can make to fix it?
onyx hinge
OK sure let's start with the fix
blissful palm
interested
onyx hinge
class TextSplitter:
def __init__(self, words):
words.sort(key=lambda x: len(x), reverse=True)
self.words = set(words)
- self.pat = re.compile("|".join(re.escape(w) for w in words) + "|.", flags=re.DOTALL)
+ if words:
+ pat = "|".join(re.escape(w) for w in words) + "|."
+ else:
+ pat = "."
+ self.pat = re.compile(pat, flags=re.DOTALL)
def iter_words(self, text):
s = []
words = self.words
for m in self.pat.finditer(text):
You can make the following change in the script py/makeqstrdata.py. I'll pull request it soon
TextSplitter is used to help find frequent substrings in error messages, which improves compression. The first time it is used, "words" is an empty list, so the regular expression it generates is |. which is a regular expression that matches either "" (nothing, the empty string) or "." (any single character).
This regular expression doesn't really make much sense when you ask the question "what are all the matches in some other string". But, in Python 3.7 and newer, it turns out that the answer is: ... wait for it .... ```>>> [m.group(0) for m in re.finditer("|.", "hello")]
['', 'h', '', 'e', '', 'l', '', 'l', '', 'o', '']
there is an empty match alternating with a match of one character. This isn't exacty right, but it was close enough that the problem went unnoticed.
Python 3.5 says: ```>>> [m.group(0) for m in re.finditer("|.", "hello")]
['', '', '', '', '', '']
so it matches the empty string at 6 positions (before each letter of "hello", plus one more at the very end of "hello" just for good measure)
because it never gives a "non-empty" answer, some things go wrong and finally we encounter an error because we try to take the ord() of an empty string which is the exception you actually saw
blissful palm
lol, that's a good one..
balmy stirrup
How did you find this @onyx hinge?
blissful palm
make -C mpy-cross
make: Entering directory '/home/awood/cp-alloy/circuitpython/mpy-cross'
Use make V=1, make V=2 or set BUILD_VERBOSE similarly in your environment to increase build verbosity.
text data bss dec hex filename
205653 21512 976 228141 37b2d mpy-cross
make: Leaving directory '/home/awood/cp-alloy/circuitpython/mpy-cross'
onyx hinge
The fix is to take into account that words can be empty and just use the reguar expression "." in that case to match single characters. Whenever words is non-empty we do the same as before so that for example we might get the regular expression an|the|. , so that an and the are treated as words and every other thing is just treated as a single letter
blissful palm
Trying build now
onyx hinge
@balmy stirrup Once I had the 'bad' program locally I could see in the qstrdefs.generated.h file that a variable called words was empty, this is a debugging statement that I'd put in in the past .. from there it became a question of "why" and I was quickly able to run multiple times with additional debugging prints and trace it down
balmy stirrup
@onyx hinge are you using the MP-LIB unittest or is there one in the CP repo? Just curious. I assume there is no PDB as well at this stage?
blissful palm
The build worked worked yehaaw, we are off again..
onyx hinge
Thank you all who helped diagnose this, and I'm glad you're back to being able to build, @blissful palm !
blissful palm
Thanks to everyone else also
ionic elk
@onyx hinge want to talk about reserved and never_reset in the in-the-weeds today?
onyx hinge
@ionic elk not necessarily, but it's something I have to wrap my head around someday
balmy stirrup
Another question if I am building from source for the Bluefruit express which ports dir and make BOARD= should I use?
onyx hinge
which form factor? There are all these: circuitplayground_bluefruit feather_nrf52840_express metro_nrf52840_express
balmy stirrup
circuitplayground_bluefruit
onyx hinge
and ports/nrf
balmy stirrup
make BOARD=circuitplayground_bluefruit_express?
onyx hinge
nope, just circuitplayground_bluefruit, no express on that name
you can look at the subdirectories of ports/nrf/boards/ as a reminder
balmy stirrup
Ahh ok thanks again @onyx hinge Did you catch my question about unittest and PDB?
onyx hinge
no I didn't
balmy stirrup
Does CP take advantage of unittest from the MP-LIB or do you use your own in the CP repo? 2nd question - I would assume you have the same issue was MP regarding PDB to effectively get that to work as well?
onyx hinge
we don't use unittest in circuitpython, to my knowledge. and no, you can't use the python "pdb" debugger with circuitpython.
onyx hinge
for my "udecimal" library I used unittest (which is really just running a doctest) simply by running it in desktop python; this was fine as it's a helper type library, not a hardware type library. I also used the "run-tests" infrastructure from micropython/circuitpython to do a subset of the tests in the unix port of circuitpython. I "picked" doctest beacuse that's what the original implementation of Decimal in Python used.
balmy stirrup
Yea I would assume we can use the MP-LIB version of unittest.
I now have to try to get perms for my mac for the GCC compiler.
-different issue-
onyx hinge
Yeah I've only heard about how hard macos is making it to run software, I haven't directly experienced it. Sounds frustrating.
onyx hinge
(there's not a way for brew to at least batch-prompt you to give the necessary permissions?)
balmy stirrup
Make sure you go into Developer Tools and check terminal.
Just in case anyone runs into it I wanted to share.
onyx hinge
maybe Microsoft will bring WSL2 to Mac
balmy stirrup
I think it will be necessary as the complications with the Catalina are extremely restricting.
Enabling terminal in Developer Tools seems to work (for now).
Its not a deal breaker can always create an Ubuntu VM and just use that so there are work arounds.
My next step is to try to get this to work with CP - https://github.com/micropython/micropython-lib/blob/master/unittest/unittest.py
ionic elk
@slender iron did you make an issue for the port-status documentation project we discussed last week? I'd be interested in taking that on.
solar whale
Building CP on a Mac is not a problem once you go through getting the one time pain of enabling each of the tools it uses.
balmy stirrup
@solar whale I had to enable terminal in Developer Tools for it to work for me as I posted a pic above other than that no issues with CP just other microcontrollers.
I just wanted to share in case other MAC users were struggling.
solar whale
I'll look at that -- I don't recall doing that, but it may make it simpler.
I don't have it enabled, but it probably would have saved a lot of time!!
balmy stirrup
Yea as a test I rebuilt my MAC Air and cloned down a fresh CP repo and literally followed the https://learn.adafruit.com/building-circuitpython/build-circuitpython and that was the only step I had to do.
solar whale
Good to know!
solar whale
IIRD I had to manually enable every tool, gcc ...
balmy stirrup
No need to do that now.
solar whale
you can enable them after that error by going to the "secuirty" settings and allowing them -- one at time. Your way is much better
balmy stirrup
Cool I hope it helps.
thorny jay
I am lurking (and did not even mention it in the doc)...
It seems that I have not done a lot this week.
I have watched some streaming, I had some conversation here, ...
I'll try to listen to it live, if there is an "in the weed" about Circuit Python natively compiled for Raspbian or Circuit Python bare metal on Pi hardware... I'll try to participate.
If someone discuss the gameduino 3X Dazzler I am also interested...
onyx hinge
<@&356864093652516868> Our weekly meeting is coming up in a bit less than 2 hours. If you'll be participating in the meeting, it's super helpful for you to add your notes to the document: https://docs.google.com/document/d/1lkj5xteD6rW1eqh9swbGeuidZTKsB3CtW2Lpgff1P3U/edit?usp=sharing -- if you don't have time to add your notes, please just add your name in alphabetical order in two spots so we know you're participating. If for some reason that's not possible, let me know (mention me) and I'll add you. If you're not in the document but in the chat, I'll assume you're lurking. Doing it this way seemed to work well on CircuitPython day, so let's try it again. See you in the voice channel at 2PM ET, I'll come earlier if possible if anyone wants to do mic checks. If you're NOT a circuitpythonista yet and want to participate, let me know and I'll add you so that you can speak in the voice channel.
balmy stirrup
I just confirmed that unitest.py does work with CP yay! If that was also not known we can take full advantage of it now from the MP-LIB repo -> https://github.com/micropython/micropython-lib/blob/master/unittest/unittest.py
code.py
def foo(param):
param = param + '!'
return param
test_code.py
import unittest
from code import foo
class TestCode(unittest.TestCase):
def test_foo(self):
# Calls
result = foo('test')
# Asserts
self.assertEqual(result, 'test!')
REPL
import unittest
unittest.main('test_code')
test_foo (TestCode) ... ok
Ran 1 tests
OK
Apologize in advance if this is already known.
onyx hinge
nice!
balmy stirrup
Yea this is a big deal when working with larger projects.
I just successfully tested this on the latest alpha build that I just compiled on a CP Bluefruit.
Ill put this up on my GitHub soon so people can grab but its very simple and straightforward. I will do a small tutorial in there.
onyx hinge
if you write it up in a blog or even just a tweet, that sounds like good material for the Newsletter as well
balmy stirrup
Cool will do Jeff! I will send you a link to my GitHub repo for it.
@onyx hinge let me know if you want anything added or amended.
slender iron
@ionic elk I didn't. I forgot. Go ahead and make one
ember mirage
@lone axle I see you also have the edgeBadge I was wondering if you ever had issues with it being recognized by your computer then it's no longer able to connect. I've tried it on a PC and a mac, rest button, and turning it on and off again. I know it's getting power because I see the light blinking and it increases when i toggle the on and off switch. If so what did you do to fix this? I'm a software developer so I'm new to the debugging hardware lifestyle lol
onyx hinge
@balmy stirrup would you like us to add you as a Circuitpythonista? It would let you talk in the weekly meeting if you wanted to, mostly.
balmy stirrup
Yes I would love that!
onyx hinge
Enjoy being purple
balmy stirrup
woohoo
lone axle
@ember mirage do recall the last time the device worked? was it in Circuit Python mode?
I will be on the road here shortly but I can check back in a little later in the afternoon. My top guess is that it may just need Circuit Python loaded on it. There are many non-Circuit Python demos available for that device (tensorflow etc.) if you used one of those then you'd need to re-install Circuit Python by double clicking reset and dragging the Edge Badge Circuit Python UF2 file onto it.
If that doesn't work for you or you are 100% certain it already has Circuit Python but it's still not showing up. Then it'll require a bit more troubleshooting. On windows 7 (not sure if this affects Win10 or other platforms or not) I've had the OS just decide that "Unknown Device" was the correct driver for a Circuit Python device before and refuse to let it connect as anything else useful.
When I ran into that someone here pointed me to the Device Cleanup Utility on this page: https://www.uwe-sieber.de/misc_tools_e.html and I was able to use it to delete all of my Circuit Python known driver associations and ten re-plug the device to finally get it to recognize as the correct type.
analog bridge
make translate doesn't seem to translate strings being referenced from header files
onyx hinge
@analog bridge you are right. So far there weren't any. @slender iron might be able to shed some light on whether it's intentional that only .c files are inspected or whether we should fix it. the fix would be to add an additional -iname "*.h" to the top level Makefile's find $(TRANSLATE_SOURCES) -iname "*.c" -print | (LC_ALL=C sort) | xgettext -f- -L C -s --add-location=file --keyword=translate -o circuitpython.pot -p locale
onyx hinge
@lone axle will you be able to drop links in the channel during the meeting today?
slender iron
I can do links
analog bridge
@onyx hinge I tried implementing -iname "*.h" but no luck, maybe my implementation is wrong.
onyx hinge
drat, wonder why that didn't make it work
slender iron
what does the use in .h look like?
onyx hinge
maybe this would be the right incantation? find $(TRANSLATE_SOURCES) -iname "*.[ch]" -print | (LC_ALL=C sort) | xgettext -f- -L C -s --add-location=file --keyword=translate -o circuitpython.pot -p locale
".c" -> ".[ch]" instead of introducing a second -iname .. or maybe it has to be -iname foo -or -iname bar
analog bridge
I came across this in safe_mode.c. Some safe mode messages are board specific and are being referenced from board specific config files.
onyx hinge
you can't exactly preprocess the file, because we need to see the translate() directives in all #if/#else branches
onyx hinge
supervisor/shared/safe_mode.c: serial_write_compressed(translate(BOARD_USER_SAFE_MODE_ACTION)); so it's basically this, right?
analog bridge
we could but that might create inconsistency
onyx hinge
xgettext "sees" translate() but the content inside the translate() is not a literal string so it doesn't add anything
analog bridge
@onyx hinge yes that's the one, also there is one before the function #define FILE_AN_ISSUE translate("\nPlease file an issue with............/issues\n")
onyx hinge
if we had serial_write_compressed(board_user_safe_mode_action()) and then the board.c had compressed_string_t board_user_safe_mode_action(void) { return translate("...."); }
so that the message DOES get moved to a regular source file as translate("string literal")
I'm hanging out in the voice chat if anybody would like to do a mic check
analog bridge
that could work.....
onyx hinge
The meeting will be getting started in about 5 minutes
slender iron
goes to refill on water
manic glacierBOT
turbid radish
Lurking
timber mango
lurking today
slender iron
@idle wharf @inland tusk lurking?
idle wharf
Yes, lurking
tidal kiln
lurking
balmy stirrup
lurking
idle wharf
I likely get pulled off into work stuff, but I enjoy listening in
turbid radish
@anne_engineer on Twitter
slender iron
turbid radish
anneb(at)adafruit(dot) com
gilded cradle
@onyx hinge it was a CircuitPython library, but included an example that only ran on CPython which is where you got your wires crossed.
balmy stirrup
Thanks!
slender iron
@analog bridge I added you to circuitpythonistas
analog bridge
text only today
slender iron
🐱
ionic elk
your cat sounds like a person pretending to be a cat
solar whale
She makes her presence known.
blissful pollen
the mustache magically appeared
slender iron
👍
ionic elk
Am I on deck or Foamyguy?
onyx hinge
ionic elk
He's got an update too
slender iron
@ionic elk maybe the analog apis?
ionic elk
sure
crimson ferry
@solar whale re: weeds... NINA... and available in NINA Releases https://github.com/adafruit/nina-fw/releases
marble hornet
oh, I forgot to post the link to the watch motherboard. if anyone is interested: https://github.com/TG-Techie/TG-Watch02-PCB
manic glacierBOT
mental nexus
Thank all!
balmy stirrup
Thanks!
gilded cradle
Thanks
ionic elk
👍
crimson ferry
(sorry @solar whale, didn't mean to mess you up, just meant... to also make the build available on the NINA releases page, as well as the guide)
solar whale
just FYI -- the "working" version still reports as 1.6.1 -- will be 1.7.0 when buildable.
analog bridge
@slender iron need to have a chat about deep sleep api..... maybe tomorrow 🙂
slender iron
kk
marble hornet
Great job @onyx hinge
crimson ferry
@solar whale yeah, there are a number of places where the version number has to get changed unfortunately
it's in the README
solar whale
1.7.0 builds OK and works for WiFi, just not BLE
gilded cradle
I'm heading out. My connection has degraded where I can't even understand anything at the moment.
analog bridge
@ionic elk @slender iron TouchIO needs to configured for the s2 before I can implement touchpad wakeup
inland tusk
@onyx hinge I am interested in the CANio for cars is there a circuitpython board for the SAME 51 that will be coming up soon?
onyx hinge
@inland tusk Limor talked about it last night on Desk of LadyAda. She showed a prototype board in Feather form factor that has CAN support and a CAN transceiver with screw terminals.
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/11JULkBnDzsJVIfGiiqWPszEd7mb001P7qg-8LyucmcE/edit
@slender iron don't forget to set the previous document to read-only
slender iron
done
how does this sound for port status:
CircuitPython has a number of "ports" that are the core implementation for a variety of microcontroller families. Stability varies on a per-port basis. As of this release, atmel-samd and nrf are stable. cxd56, esp32s2, and stm are being actively improved but may be missing functionality and have bugs. litex and mimxrt10xx are in an alpha state and will have bugs and missing functionality.
@ionic elk for release notes ☝️
slender iron
@onyx hinge I'm in the amelia voice channel
onyx hinge
@slender iron OK just joining now
atomic summit
I got my FeatherS2 test-jig working - using the flash, remount with write access, test, then remount without sequence. It's stable, but really slows down the test cycle.
Since I took this pic, I've managed to have a few more seconds off, but until I can get cpboard.py working for raw_repl, I'm going to keep using this.
idle wharf
That is very cool @atomic summit !
teal bear
when I try to out 17 bits into a 16 bit thing, it errors, good. struct.pack("H", 65536) OverflowError: value must fit in 2 byte(s)
where is the chack for that? like I can't figure it out looking at https://github.com/micropython/micropython/blob/master/py/modstruct.c#L215
gah.. not MP's it doesn't check...
lone axle
@onyx hinge sorry to ghost on you there. I got pulled away into some day job things. Will be catching up on the meeting later on tonight.
slender iron
ionic elk
@slender iron sorry I went out for a run and missed your message. Sounds good to me. What steps would you like to see for STM32 to enter the stable category?
slender iron
good question. I'd feel better about it based on how many people we see running it
I keep forgetting about the stm feather
ionic elk
it's missing rotaryio, rtc, and audioio compared to nrf
slender iron
we could make it stable. I don't really care about where it sits
ionic elk
How do STM32 numbers compare to the other boards on nrf and samd
slender iron
I haven't looked but I imagine its much less that the other two because its only on one board
ionic elk
I could try and do a big testing run before we mark it as stable. Get every featherwing and run it through the gamut
Make extra sure we don't have anything like the SD card thing still bumming around somewhere
how do you tell, as far as board stuff? Do you get any info from that from adafruit?
slender iron
I can look it up. but if you think its stable I can just edit the release notes
ionic elk
Hmm I'm waffling as to whether I'd say that now, it was a genuine question as to what you'd like to see done. But I don't have any more major bugs logged on github
only error I see that should get investigated is https://github.com/adafruit/circuitpython/issues/3376
slender iron
I think you are reading more into it than I gave thought
ionic elk
Lol
slender iron
my main goal was to highlight that stability varies and that esp is beta and imx/teensy is alpha
ionic elk
Yeah go ahead and mark that STM32 is stable then.
slender iron
should we only say stm32f4?
ionic elk
Yes. F7, H7 and F1 are all at different levels of completion
As of this release, atmel-samd, nrf, and stm32 for the F4 family are stable. cxd56, esp32s2, and other ST families are being actively improved but may be missing functionality and have bugs
Also, by the way, I have a friend who's currently trying to make a linux-capable featherwing using the H7
slender iron
CircuitPython has a number of "ports" that are the core implementation for a variety of microcontroller families. Stability varies on a per-port basis. As of this release,
atmel-samd,nrf,stmfor the F4 family are stable.cxd56,esp32s2, andstmfor other chip families are being actively improved but may be missing functionality and have bugs.litexandmimxrt10xxare in an alpha state and will have bugs and missing functionality.
ionic elk
we'll see if he can do it, it's a LOT of routing and he has to squish both external flash and ram on there
slender iron
🙂 sounds like the groboard (iirc)
ionic elk
oh nice, I'll send that to him
ionic elk
I also still want to try that circuitpython-without-the-python idea... all these stable APIs but no raw C interface 🤔
ionic elk
right, I'd have to run some kind of project-wide replacement script
Or figure out something more clever
orchid basin
slender iron
ideally the C apis wouldn't raise exceptions
slender iron
goes for a walk while logs download
ionic elk
@slender iron for whenever you get back - microcontroller.cpu.temperature support was actually added by MicroDev
atomic summit
So what happens with IDF upstream changes now?
slender iron
@atomic summit my plan is to pull in upstream changes as needed
atomic summit
When are they brought in and how will they be tested etc?
So you'll just stay on top of them and see of any are needed or appropriate?
slender iron
no? if nothing is broken then there isn't a reason to update
atomic summit
Ahh, right, but there are often "improvements" in IDF updates... things that don't fix things, but are still worth including.
It's one of the things that often holds MP up, as if there isn't a "solution to a fix" to find... no one is looking at the IDF very often. And with the S2 so new, and still no official IDF release looming...
I guess for now, getting S2 support out is more important than getting it perfect.
slender iron
nothing is ever perfect
atomic summit
haha, you know what I mean 🙂
slender iron
interacting with the IDF is still new for us
so if they add something we'd benefit from then we'd update
similar to micropython
atomic summit
Yeah sure, that's why I'm bringing it up. IDF also often gets regressions as well so it's a tough balancing act.
slender iron
I'd like to change the way we do includes so our warnings don't apply. that'll make it easier to update
slender iron
ok, beta 0 is completely released
manic glacierBOT
slender iron
@grim dagger how are you trying to test the display?
grim dagger
@slender iron I have my own fork of CircuitPython with the SH1107 quirk coded into it: https://github.com/mdroberts1243/circuitpython
I put my new argument "column_and_page_addressing" everywhere it seemed to be needed.
Got a clean build
slender iron
Nice!
grim dagger
Built a new displayio driver... in another github repo. But when I try to set my quirk argument to true python is complaining that I'm passing too many arguments into super().init
super().init
Anyway... add a few _ in the above, can't convice discord to print them out.
I have been looking at the code and I can't see what I've missed in trying to pass the new argument
the circuitpython branch is 'origin/New_quirk_for_SH1107'
@slender iron Another question I have is how to make set_column_command and set_row_command to be equal to NO_COMMAND in the C code... I assumed I would just not pass them in the driver and commented them out.
slender iron
@grim dagger I’m on my phone so my ability to check things out is limited
grim dagger
End of the day for me, so no rush... I'm going to bed!
slender iron
You can us backticks to get things printed literally. Make a draft or and I’ll check it out tomorrow
Draft PR
grim dagger
How do you make a PR for the new driver? Or will it be sufficient to do it for my version of circuitpython?
manic glacierBOT
grim dagger
@slender iron The Draft PR throws an error in pre-commit: ##[error]The process '/opt/hostedtoolcache/Python/3.8.5/x64/bin/pre-commit' failed with exit code 1
slender iron
In the ci or with your commit? It’s ok if the ci is broken on your draft
grim dagger
@slender iron pre-commit / pre-commit (pull_request) Failing after 18s — pre-commit . The builds are still in process. Many have passed. Anyway, I'm probably not familiar enough with the Adafruit setup to really know what it's complaing about 😳
slender iron
It’s usually trailing lines or spaces. Nothing major
Easily fixed right before we merge
grim dagger
@slender iron Ok. Thx. Hopefully you just need to scan my code changes to suggest my next step. I'm testing with the STM32F405 Feather and SH1107 FeatherWing. GN.
slender iron
Yup sounds like something in shared bindings. Thanks! Good night!
manic glacierBOT
manic glacierBOT
manic glacierBOT
silver tapir
I hope this won't distract @slender iron... but this is so cool. It's running without an fpga, only with software on their processors (esp+k210) in their clone of micropython.
https://twitter.com/SipeedIO/status/1308353905034629120
onyx hinge
there are a LOT of messages on my car's canbus
slender iron
nice!
onyx hinge
https://gist.github.com/73572f0928e00be26c8b5e6f87076a9e -- I don't know how to interpret anything, this was simply snooping the bus
looking for a reference now..
slender iron
I think that's where a library would really help
@timber bridge has hacked on cars I think
simple pulsar
Hello. There's a post in the forums about what looks like an i2c problem on a straightforward, reasonable hardware combinations, a CLUE and an i2c plantower PM2.5 thing connected via stemma-qt. Why would that throw a OSError: [Errno 19] Unsupported operation ? https://forums.adafruit.com/viewtopic.php?f=65&t=169662
slender iron
I assume getting a logic analyzer on it is the next step
looks like you dug down to where the exception would originate already
simple pulsar
Should someone test that combination too? I have a CLUE and a PMSsomething but it's the serial version not i2c.
slender iron
ya, I only have the uart version too
onyx hinge
@slender iron had a quick chat with Limor, sounds like I should switch to doing the low level for STM32F405 rather than diving into automobile diagnostics.
slender iron
kk, up to her
I can put a call out on twitter to find a reference if you haven't found one yet
onyx hinge
I have found some stuff, bookmarked for when I pick that part of things back up
onyx hinge
see ya
simple pulsar
@tidal kiln Do you have the hardware combination mentioned in https://forums.adafruit.com/viewtopic.php?f=65&t=169662 ? As mentioned earlier, it seems odd that this wouldn't work.
analog bridge
@slender iron Up for a deep sleep api talk ?
tidal kiln
@simple pulsar i don't. and all the various AQI sensors are currently super out of stock.
solar whale
Has anyone else noticed the esp32s2 being a bit "flaky" at the REPL -- I get occasional "syntax errors" on inout lines that if I just hit up arrow and enter, they work fine. Also some occasional issues for cut/paste. I can file issues, but just wanted to see if anyone else was seeing this. I am using "screen" on a Linux box to access the REPL.
solar whale
the "flakyness" may only be after a cut/paste ....
analog bridge
@slender iron Up for a deep sleep api talk ?
@slender iron In case you missed my message ^^
slender iron
@analog bridge I'm around now
and yup, I missed it because I was eating breakfast 🙂
analog bridge
sorry to disturb your morning routine........ voice ? 🙂
slender iron
sure 🙂
slender iron
timber bridge
@onyx hinge what kind of car is that from?
onyx hinge
@timber bridge 2013 prius
slender iron
analog bridge
Nice talk @slender iron 👍
timber bridge
@onyx hinge I don't have any data for that, but these people might: http://illmatics.com/carhacking.html
grim dagger
@slender iron Commenting out the new kwarg stops the error message. None of my added debug statements are printing out. But the uf2 firmware file is copying over fine and I can see it on the CIRCUITPY disk with the correct date and time. Is there a special consideration with updating the firmware on the STM32F405 Feather? I got it running the first time this way.
slender iron
the uf2 file needs to go on a *BOOT drive, not CIRCUITPY
did the feather come with circuitpython on it?
you may need to use the DFU loader to load circuitpython. I don't know if UF2 is on there by default
grim dagger
@slender iron It might be because I switched over to my linux laptop for the circuitpython build... I originally used my windows laptop for the installation of the standard build on the Feather.
slender iron
@grim dagger did you figure it out?
onyx hinge
@grim dagger the stm32f405 feather is not set up by adafruit to use uf2, you need to use dfu-util to upload new firmware
grim dagger
@onyx hinge @slender iron Sorry... interrupted by day job 😉 Yeah... just now realized. I had to go back to STM Cube Programmer on windows to get the .bin file transferred over. So I can see my debug print messages now, but I'm still having the same problem, too many kwargs, so I need to figure out where I've missed something in the declarations.
slender iron
hrm, everything looked ok to me
onyx hinge
is there another level of library in pure-python? is an option mis-spelled somewhere?
onyx hinge
{ MP_QSTR_column_and_page_addressing, MP_ARG_BOOL | MP_ARG_KW_ONLY, {.u_bool = false} }, "extra" arguments also covers the "is spelled wrong" case. Make sure "column_and_page_addressing" is spelled consistently everywhere? Otherwise, I haven't spotted the problem yet.
grim dagger
I have checked and rechecked. Is there any weird limit like the length of an argument name?
onyx hinge
I am not aware of one but I couldn't rule it out
grim dagger
Where does mp_arg_parse_all live?
onyx hinge
py/argcheck.c, says a full text search (grep)
@grim dagger can you share your python code with me?
grim dagger
Basically the example python code is in the SH1107 driver I attached to the Draft PR, with minor changes now as. https://github.com/mdroberts1243/mdroberts1243_CircuitPython_DisplayIO_SH1107
onyx hinge
I had picked up one of these boards but sadly I seem to have damaged it. It does not appear on the I2C bus when I i2c.scan() it.
I pressed on the edge of the display when inserting it and it fractured, I guess it's cutting a vital signal. boo.
slender iron
any idea if wifi nina actually works with multiple sockets?
onyx hinge
@grim dagger OK figured it out
grim dagger
@slender iron @onyx hinge O.k. guys I can proceed now. Found an extra comma in the definition of static const mp_arg_t allowed_args[] = {...} 😩
slender iron
ah, that was it!? weird
onyx hinge
I don't think that was all
$36 = 0x8054b4a "column∧page_addressing"
There's magic in qstrs to represent some non-identifier characters. _and_ turns to the symbol ∧.. I doubt we deliberately do this anywhere in CircuitPython.
grim dagger
I also changed the argument name to be cp_addressing
slender iron
haha, oh no
grim dagger
That was a hail mary, but left it that way.
onyx hinge
I'm not sure how to "escape" the _and_, do you know @slender iron ?
grim dagger
I can leave it as cp_addressing... shorter to type anyway.
But maybe open an issue for future frustrated programmers?
onyx hinge
we like things to have GOOD names though
crimson ferry
@slender iron As I read this https://github.com/arduino/nina-fw/commit/519cb190ccfc223d5124cb1e0f20ee021fd1c8fd yes, haven't done it, but we know it's <= 10
slender iron
no, I don't know how to escape it
@crimson ferry I'm trying to understand why nina is unhappy with new requests but not the old one
I find the lack of errors from nina frustrating
crimson ferry
DEBUG can be a firehose
slender iron
I tried to change requests so that if the second connect fails it will close opened sockets
and then try again
but it still doesn't work
crimson ferry
this is a newb question, but in the old days (last week) when we do import adafruit_esp32spi.adafruit_esp32spi_socket as socket and use socket as-is in requests.set_socket(socket... that doesn't allow for multiple instances?
crimson ferry
ah, ok
onyx hinge
@grim dagger congratulations for having the most interesting problem come up on your first contribution to the core! It's not always like this.
grim dagger
😳 Some silly mistakes too
slender iron
😄
grim dagger
@onyx hinge @slender iron The display isn't rendering correctly so there's still work ahead...
onyx hinge
@grim dagger did you see my note about maybe a stray write to the panel in your changes?
afk, I will get another of these displays ordered so I can help out if the problem goes on for another couple of days
slender iron
@atomic summit what are you writing to the internal flash from circuitpython?
is watching your test jig recap
atomic summit
@slender iron Yes, and it's a painful process 😦
slender iron
I'm wondering why you need to do it
atomic summit
because I cant get cpboard.py to work to execute raw_repl
so I cant get my test results off my board
in MP, I execute my tests with pyboard and return the results...
I'd love to get that working in cpy.. and @raven canopy was trying to help me the other day, but so far no success.
I can see code is executing, but times out before I get results back.
slender iron
could you get the results off by connecting to serial as you run it?
connect to serial and then copy test code.py over
you could also add the IO## pin names into pins.c
balmy stirrup
There should be at minimum something coming off the serial port if the firmware compiled.
atomic summit
how do I get them off via serial? you mean via something like screen and capture the output?
balmy stirrup
Yes when you connect screen or Putty and you reboot the device what are you seeing?
slender iron
yup, pyserial is what I was thinking
atomic summit
ok, I'll look at pyserial - never used it directly before
slender iron
uf2 only has neopixel support 😕
atomic summit
What do you mean? no dotstar support ?
atomic summit
@balmy stirrup Yeah I have pyserial installed as esptool needs it
I'll check out pyparsing though
slender iron
ya, just for the bootloader status LED
atomic summit
ahh, bummer.
slender iron
shouldn't be too hard to add
balmy stirrup
@atomic summit are you getting any output in the serial monitor when booting?
atomic summit
Which serial monitor?
Ok, I'm running some code on my feathers2 that just prints "hello serial" 4 times a second
and running pyserial on my rpi doing a print(ser.readline()) forever
it's not getting any data
balmy stirrup
So I am assuming you do not get any active REPL?
balmy stirrup
If you CTRL-C?
atomic summit
ow do I send ctlr-c ?
grim dagger
@grim dagger did you see my note about maybe a stray write to the panel in your changes?
@onyx hinge Thanks Jeff, nice catch that I cleaned up earlier!
atomic summit
shouldn't pyserial just grab whatever is being printed ?
balmy stirrup
@atomic summit can you use ampy to send a new code.py? You should be able to CTRL-C to drop to a REPL as I am assuming you have not disabled the REPL in main.c so it should work.
balmy stirrup
ok
atomic summit
sure, I cant get into a repl if I am directly connected to the device
using screen etc.
i need to copy test code over to the DUT, and then read the results via serial
balmy stirrup
I understand now. Perhaps you can mod the code.py to something you are expecting and check the results.
atomic summit
check the results how, that's what this pyserial talk is all about.
I am already doing that, but I cant find any way to "read the results".... so right now I am writing them to the filesystem, so I can copy them off
Perhaps you can mod the code.py to something you are expecting and check the results.that is exactly what pyboard.py does... but also allows me to just inject code into the repl.
but I cant get that working either
balmy stirrup
The only thing I can guess is your are locked in some infinite loop. When I use pyboard.py if I am locked in some loop I will not get write access it is only when I am able to get a REPL and disconnect from screen or Putty am I able to send a new file over with pyboard ./pyboard.py --device <DEVICE> -f <file> : but it sounds like this is something I have not encountered sorry about that.
slender iron
it might be an issue with detecting that serial is connected
setting the dtr line or something
balmy stirrup
That is definitely a possibility @slender iron more than likely.
atomic summit
Ok, I got it working.
slender iron
nice!
atomic summit
so for some reason, copying a new code.py is not making the board restart with that new code. I have to reset it
balmy stirrup
That's what I thought.
Yes when you successfully copy then you reset.
./pyboard.py --device <DEVICE> -f <file> :
atomic summit
No, I'm copying from the rpi via the mounted drive..
Oh, yeah, sync.. that would prob help.
doh
slender iron
though no led works
and for some reason my hub keeps disconnecting the device
we might not ask for the right amount of current
anyway, I gotta go run before I get too hungry
atomic summit
haha, enjoy food!
slender iron
byeeee!
balmy stirrup
Glad it is working @atomic summit!
atomic summit
thanks for the help Kevin
balmy stirrup
Pleasure!
atomic summit
w00p! Got it implemented and working!
About to find out what the new cycle time is...
Current time is approx 67secs
flashing takes 35s of that unfortunately
balmy stirrup
WOOHOO!
atomic summit
I'm expecting to only save 6-7 seconds...
balmy stirrup
It's something.
atomic summit
56.70s
so saved 11 seconds!
better than a kick in the teeth! plus is just a nicer process
and I am sure i can save extra time by tightening a few things up
52.3secs
#winning
grim dagger
Is there a way to disable/turn-off the console output temporarily on displayio?
balmy stirrup
@grim dagger what specifically does your program log to console via print? You can always remove those statements. Have you had a chance to read https://circuitpython.readthedocs.io/en/latest/shared-bindings/displayio/ ?
grim dagger
Thanks Kevin. I'm trying to debug a new display so I want to simplify things as much as possible.
slender iron
@crimson ferry I think I'm running up against the fact that esp32spi socket isn't a real socket
crimson ferry
not real?
crimson ferry
ohhh
slender iron
I think the alternating is that the socket is closed
which we should handle in requests
not sure why it's closed though
crimson ferry
is there a way to get the status of the socket to know if it needs closing? (TCP)
slender iron
unfortunately I don't think the cpython object has an is_closed
crimson ferry
PR the IDF 😉
slender iron
well, I want to go based on cpython
crimson ferry
that does explain the alternation... fail, open new socket, success, rinse, repeat, run out after 20 iterations
slender iron
nina we can fix if need be
yup, for some reason it won't let us send a second time
when we try and make the second http request
and we don't close a socket after it fails
which we should
code like this bothers me: https://github.com/adafruit/nina-fw/blob/master/arduino/libraries/WiFi/src/WiFiSSLClient.cpp#L272-L283
because it hides error detail
crimson ferry
and esp debug doesn't even help at that level
crimson ferry
I don't think there was a check before
I need a bigger display... trying to read Requests, ESP32SPI, and NINA simultaneously
sleek swan
I just read about the circuitpython beta featuring wifi support for esp32-s2 is there like a guide to assemble them both ?
balmy stirrup
@sleek swan Off the top of my head there is a video where Scott works with the esp32-s2 which might help. https://www.youtube.com/watch?v=8WvLviAr8a0
sleek swan
cool thank you @balmy stirrup 😄
balmy stirrup
Welcome!
solar whale
@sleek swan there is also some information here https://github.com/adafruit/circuitpython/tree/main/ports/esp32s2 -- especially if you want to build CircuitPython yourself --- more on that here https://learn.adafruit.com/building-circuitpython
manic glacierBOT
lunar crown
Can someone point me to a reference mapping the CircuitPython names to the Teensy 4.0 pins?
balmy stirrup
onyx hinge
graham: in our source tree, you can find the file ports/mimxrt10xx/boards/teensy40/pins.c -- it'll show the correspondence between the "pins" module and the "microcontroller.pin" names. Usually microcontroller.pin names match the microcontroller datasheet., while "pins" matches silkscreen
we prepend "D" to a digital pin, so pin.D0 should be what is silkscreened "0" or "D0", and "A0" would be what is silkscreened "A0"
old smelt
@onyx hinge - I see in the notes from the Monday call that you are close to wrapping the CAN I/O for SAME5X. Do you have any documentation or more info on that?
onyx hinge
old smelt
Thanks!
onyx hinge
hah and then I notice a typo in the docs 😛
https://gist.github.com/jepler/3627fbe0659ae8e54dee27d49699ae4d and this is the test program I used to validate it in loopback mode
lunar crown
@onyx hinge Thanks, i can figure it out from there. @balmy stirrup I specifically need the pin numbers assigned to the Digital IO pin assignments as named in CircuitPython. I think it might be the "Gray" colored pin numbers on the diagram you sent. Thanks.
slender iron
@lunar crown I think I went by the silkscreen or schematic
crimson ferry
@lunar crown You probably already have what you need, but using this code https://gist.github.com/anecdata/1c345cb2d137776d76b97a5d5678dc97 I get these microcontroller-to-board pin mappings https://gist.github.com/anecdata/84b0165d42655c7381aa109d6d1c4145
lunar crown
@crimson ferry That is a sweet little utility. I might have to go back to the BGA ball then come forward on the schematic, but it should all be there.
blissful palm
@slender iron just a gentle poke, I did a PR on you Picodes fork at the weekend in case you hadn't noticed it (I know your busy)..
granite juniper
prime flower
@slender iron Unburying myself - do we still want a sep. BLE_Socket library for this PR? I have time this week to take a look at separating it out. (https://github.com/adafruit/Adafruit_CircuitPython_MiniMQTT/pull/15).