It's has dual Arm Cortex-A7 and Cortex-M4 processors
#circuitpython-dev
gilded cradle
hybrid scarab
โค๏ธ
onyx hinge
ivory yew
600 whole bytes
slender iron
nope. damien was happy to hear about that savings too
thorny jay
Did you hear me?
lone axle
yep
slender iron
@idle owl you crash?
ionic elk
@idle owl dang it I should have thanked you for helping set up the PCBHelpers role
thank you for that
onyx hinge
I didn't quite get all the info in but I did my best
slender iron
onyx hinge
http://docs.micropython.org/en/latest/develop/cmodules.html describes how MicroPython thinks of things but it doesn't really cover the questions that you had @mental nexus
mental nexus
Cool. Iโll check those out to continue to educate myself. Iโll create a first draft and then see what corrections and clarifications are useful.
lone axle
Neat! I like the idea of Cartridges for PyGamer
slender iron
it may have been added before you could subclass group
lone axle
Interestingly it does extend Group also, but seems not to use the extended instance in favor of the property one.
slender iron
weird!
solar whale
onyx hinge
board.I2C() is preferred. Use busio.I2C if the pins or frequency need to be modified.
no idea which oldest version it works in
slender iron
I haven't gone through emails yet so spam away ๐
tidal kiln
there is a font guide here:
https://learn.adafruit.com/custom-fonts-for-pyportal-circuitpython-display
onyx hinge
board.I2C() is in docs for 4.0 so there's no excuse there ๐
@ionic elk if you have secret tips on how to help search engines prioritize one version, send links ?
hybrid scarab
Sphinx ๐ฌ
onyx hinge
Active versions that are hidden will be listed as Disallow: /path/to/version/ in the default robots.txt file created by Read the Docs. https://docs.readthedocs.io/en/stable/versions.html
however, builds cannot be triggered for inactive versions
slender iron
@solar whale mute?
onyx hinge
oh hey we can have custom 404 pages !
slender iron
thanks
mental nexus
Thanks!
modern wing
Thanks everyone! ๐
thorny jay
Thanks.
raven canopy
hehe. i'm double lurking (in a diff meeting) and was about to link RTD's SEO page. ๐
modern wing
@raven canopy I use a different laptop for work & personal -- one with external keyboard & mouse, and so often I reach for the wrong one and wonder... "why isn't that showing up? oops. posted in the wrong team chat."
crimson ferry
@vivid temple and others:
ESP32-S2 variants:
โข Espressif Saola (-R Wrover with PSRAM) (@ Digikey, etc.; module-only @ Adafruit)
โข Gravitech Cucumber (Saola near-clone) https://www.gravitech.us/curesdebo.html
โข @slender iron Saola-to-Feather adapter (https://oshpark.com/shared_projects/s0SDBOI1)
โข @atomic summit Feather https://unexpectedmaker.com/shop/feathers2-esp32-s2
โข Electronic Cats Feather https://electroniccats.com/store/bast-wifi/
โข Adafruit ...hopefully future #TopSecret :-)
Any others?
solar whale
I do a ls boards then copy/paste ....
tulip sleet
@ionic elk
thorny jay
bang
solar whale
must be a way to do it in emacs -- ๐
onyx hinge
Mon Aug 10 14:17:42 CDT 2020
jepler@eric:~$ !!
date
Mon Aug 10 14:17:42 CDT 2020
it's very old
before cursor up keys were invented
thorny jay
1978 ...
onyx hinge
see y'all later
onyx hinge
@idle owl i'll plan to run the meeting next week. we'll get into a regular cycle at some point soon
tulip sleet
!da
idle owl
@onyx hinge You running it next week will put us on that cycle. You'd run it every third week.
onyx hinge
thanks for hosting us
idle owl
Thanks for waiting through my network crash ๐
gilded cradle
Thanks
hybrid scarab
(๏พโใฎโ)๏พ*โฒ๏พ*๏ฝกโ ORGANISATION! ๐
thorny jay
@hybrid scarab Do you want to quick chat about what is left to do for Enviro+?
hybrid scarab
Potentially tomorrow -am chillin' now. I have a TODO checkbox list on my reply to the 5.x GitHub issue that can be expanded upon
thorny jay
Just let me know if you need help in testing or so.
I think trying to rewrite example using the new library and Adafruit PM25 would be something I can do.
Then it would be visible if there are still oddities or if everything is working.
I now have 3 or 4 Feather MCU that I can test with and I have a workflow to copy over a minimum set of files for testing the same thing on all of those board.
hybrid scarab
You've presently got 3-4 more feather MCUs than me then! I've been testing on a CLUE ๐ฌ
But- yes- broadly testing and feedback is super useful because I... I'm all alone ๐ข
thorny jay
I believe I have an OCD. I was buying a lot of Pimoroni HAT and pHAT. I solved that problem by buying Adafruit CircuitPython board...
hybrid scarab
CircuitPython will get ya'
thorny jay
The good news is that if there is ever a Global Pandemic, I would have for years of fun playing with why I have at home already...
Part of my plan is to drive Pimoroni HAT and pHAT from CircuitPython... either with Blinka or with an adaptor. I do that already with the CLUE and Bit:2:Pi ... but it will be good to try from a Feather. I ordered a PCB that is supposed to help with that.
solar whale
@ionic elk @slender iron Thanks -- very informative! Sorry - i have to leave.
slender iron
gets lunch before tackling the email pile
mental nexus
Here's my one-page image of explaining C<->Python:
idle owl
Fancy!
mental nexus
To do: 1) Understand how the input arguments are defined/type checked, set as default values, handling fixed vs variable # of parameters. 2) Understand data type conversion requirements between the C++ side and Python side.
I suspect that this library displayio.Shape may not be the best example to use, but it was one that I found that had a function with more than two parameters.
onyx hinge
Thanks @ionic elk
idle owl
@onyx hinge You still around?
Looking at your Adabot PR. There's another one in that, if merged first, is going to cause a (simple, I think) merge conflict. That said, if I merge yours, the newer PR will have a conflict, so it's going to pop up somewhere regardless. Anyway, I had a question about a change. I can ask on the PR.
onyx hinge
@idle owl I can take care of merge conflicts, see you in the comments on the PR
idle owl
@onyx hinge that works. Thanks.
ionic elk
@slender iron pushed the changes, sorry I forgot to after our conversation.
slender iron
np
idle owl
@gilded cradle When you get a chance, can you check out this PR? https://github.com/adafruit/Adafruit_CircuitPython_BitmapSaver/pull/12
gilded cradle
Sure @idle owl
idle owl
Thanks!
@ivory yew @tulip sleet This PR has a merge conflict, simple one as far as I can tell. But you both reviewed it, is it otherwise ready to go now? https://github.com/adafruit/Adafruit_CircuitPython_BusDevice/pull/51
tulip sleet
@idle owl @ivory yew I may be misunderstanding, but I believe the BusDevice PR for IC timeouts is not the correct place to fix the problem. The I2C devices hang at a low level, and this problem should be fixed in the C code, and can't really be fixed here. That's why I mentioned https://github.com/adafruit/circuitpython/issues/2253. So I don't think this PR should be merged.
idle owl
@tulip sleet Understood. I'll reference your comment and close it this week.
solar whale
sure -- just a few minutes
slender iron
np, I'm confident it's more correct
solar whale
how should I test -- storage.erase_filesystem() ?
slender iron
it's related to this: https://github.com/adafruit/circuitpython/issues/3133
I saw it if you erase, write a code.py, disconnect and then connect again
may only be a linux thing
solar whale
ok -- will try on linux
slender iron
basically if you write the first sector first then it'll be incorrectly written
because it'll think it's cached when it isn't
solar whale
Looking back at 3133 -- I was not able to reproduce the issue -- but I'll try the PR and see if helps my other linux issues
slender iron
kk
solar whale
@slender iron after storage.erase_filesystem() I get lots of errors in dmesg
solar whale
and my linux box hung -- will need to reboot -- cant get to them now ๐ฆ
solar whale
it worked once -- though I had to d a hard reset after the erase to get anywhere.
slender iron
ya, I have to do a reset after erase too
solar whale
but on second try it was not happy. -- shoukld be able to get to logs in a a minute
slender iron
kk
solar whale
this may have been my old hang on disconnect issue....
slender iron
ya
solar whale
will try it on my RPi
FS looks OK on it.
seems fine on RPi
hmm -- on Linux I retried storage.erase_filesystem() and it seem OK -- no CIRCUTPY but I had not done the hard reset yet. -- dmesg has many i/o errors - was trying to capture when the box froze again. this tie I had not unplugged it ..... rebooting again.
I think I'm done with Linux testing for now....
slender iron
haha, ya
do you know what usb controller you have?
did the rpi crash or just the other box?
solar whale
just sec -- got it
Bus 002 Device 008: ID 05ac:828a Apple, Inc.
Bus 002 Device 004: ID 0a5c:4500 Broadcom Corp. BCM2046B1 USB 2.0 Hub (part of BCM2046 Bluetooth)
Bus 002 Device 003: ID 0424:2512 Microchip Technology, Inc. (formerly SMSC) USB 2.0 Hub
Bus 002 Device 002: ID 8087:0024 Intel Corp. Integrated Rate Matching Hub
Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 001 Device 002: ID 8087:0024 Intel Corp. Integrated Rate Matching Hub
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 004 Device 002: ID 1058:2621 Western Digital Technologies, Inc. Elements 2621
Bus 004 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
Bus 003 Device 006: ID 046d:08b2 Logitech, Inc. QuickCam Pro 4000
Bus 003 Device 003: ID 0409:005a NEC Corp. HighSpeed Hub
Bus 003 Device 013: ID 0461:4d22 Primax Electronics, Ltd
Bus 003 Device 012: ID 413c:2106 Dell Computer Corp. Dell QuietKey Keyboard
Bus 003 Device 011: ID 046d:c52b Logitech, Inc. Unifying Receiver
Bus 003 Device 004: ID 1a40:0101 Terminus Technology Inc. Hub
Bus 003 Device 002: ID 1a40:0101 Terminus Technology Inc. Hub
Bus 003 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
jerryneedell@jerryneedell-ubuntu-macmini:~$
maybe thats not what you wanted
If I just plug it in -- mounts OK
solar whale
-[0000:00]-+-00.0 Intel Corporation 3rd Gen Core processor DRAM Controller
+-01.0-[04-9a]----00.0-[05-6a]--+-00.0-[06]----00.0 Intel Corporation DSL3510 Thunderbolt Controller [Cactus Ridge 4C 2012]
| +-03.0-[07-37]--
| +-04.0-[38-68]--
| +-05.0-[69]--
| \-06.0-[6a]--
+-02.0 Intel Corporation 3rd Gen Core processor Graphics Controller
+-14.0 Intel Corporation 7 Series/C210 Series Chipset Family USB xHCI Host Controller
+-16.0 Intel Corporation 7 Series/C216 Chipset Family MEI Controller #1
+-1a.0 Intel Corporation 7 Series/C216 Chipset Family USB Enhanced Host Controller #2
+-1b.0 Intel Corporation 7 Series/C216 Chipset Family High Definition Audio Controller
+-1c.0-[01]--+-00.0 Broadcom Inc. and subsidiaries NetXtreme BCM57766 Gigabit Ethernet PCIe
| \-00.1 Broadcom Inc. and subsidiaries BCM57765/57785 SDXC/MMC Card Reader
+-1c.1-[02]----00.0 Broadcom Inc. and subsidiaries BCM4331 802.11a/b/g/n
+-1c.2-[03]----00.0 LSI Corporation FW643 [TrueFire] PCIe 1394b Controller
+-1d.0 Intel Corporation 7 Series/C216 Chipset Family USB Enhanced Host Controller #1
+-1f.0 Intel Corporation HM77 Express Chipset LPC Controller
+-1f.2 Intel Corporation 7 Series Chipset Family 6-port SATA Controller [AHCI mode]
\-1f.3 Intel Corporation 7 Series/C216 Chipset Family SMBus Controller
this is an old Mac-Mini with Ubuntu 20.04
solar whale
yes -- on the Pi there is only one and its a VIA Technologies VL805 USB 3.0
slender iron
yup, makes sense
solar whale
sorry -- not much help
I have to go offline for the night -- let me know if you need anything tested tomorrow. Good luck.
hybrid scarab
Ruh roh, Sphinx strikes again
mental nexus
@onyx hinge some questions on my quest to decode CircuitPython core. 1) do you have a good example bindings file to point me to that has best-in-class decoding of KW_style input parameters? 2) whatโs the best way of decoding the *self pointer argument? 3) some functions are very explicit about using some kind of type changing functions to go between C and Python while others seem to throw care to the wind. Do you have a good function you can point to that shows the preferred method converting back and forth (preferably including all three: bools ints and obj)?
onyx hinge
I think stuff in displayio should reflect best practices, it's new-ish and scott's vision for how the shared-bindings / shared-modules / common-hal split should work should have been fully realized
stuck elbow
seconded
onyx hinge
the constructor of Display certainly has so many arguments it should hit all the relevant cases of what kwarg parsing can do
the only elaboration I can suggest is in a few cases I've introduced NUM_ARGS as the last item in the argument enum, so that I can write MP_STATIC_ASSERT( MP_ARRAY_SIZE(allowed_args) == NUM_ARGS ); and get a small guard against the enum and array items getting out of sync
stuck elbow
is that a compile-time thing?
onyx hinge
yeah
if you add/remove from the enum but don't add/remove from the array, you get a compile time error
@mental nexus ^
stuck elbow
neat
mental nexus
Thanks @onyx hinge I will focus on understanding Display as current best-practice.
onyx hinge
yay I got past the crashes plaguing me friday & monday .. it was not code I had written, it was code I had not yet written. Remaining: Enable the Sharp Memory Display framebuffer object to survive soft-reset
stuck elbow
@onyx hinge have you considered using displayio instead of the framebuffer approach?
onyx hinge
@stuck elbow "framebufferio", it keeps the full framebuffer in RAM but works with displayio primitives like groups and tilegrid
stuck elbow
the memory is what I'm worried about
onyx hinge
keeping the whole framebuffer in RAM in this case because the display can only do row-by-row updates and regenerating whole rows is slow. The framebuffer's about 12kB
stuck elbow
that's pretty much all the memory you have on a samd21
onyx hinge
this is not targeted at samd21
worn lark
Hi, I maybe have a bit of a weird question. I have a custom board with a samd21 where I'd like to use PA30 for SPI. Though this pin also is SWCLK. Does this make any problems? I currently get 'invalid pin' when starting spi with my configuration.
hybrid scarab
I can't be sure, but I strongly suspect the Circuit Python for the FT232H guide has lots of redundant steps
I built myself a driver installer package that targets the VID/PID of the FT232H, skipped the rest of the instructions, and just ran with set BLINKA_FT232H=1 with no problems on Windows 10
(Using the LibusbK driver wizard)
ivory yew
@worn lark that may not be a valid pin for the sercom peripheral
Let me check real quick
hybrid scarab
Now reading an LTR559 breakout through an FT232H, that's pretty sweet
Would be interested to know why Zadig was used in lieu of a driver package ๐ค
Curse you lot, you've made Blinka/CircuitPython very alluring
ivory yew
@worn lark you can use PA30 with SERCOM1 only.
If CircuitPython already give sercom1 to another set of pins, you might get that error.
But I'm not sure. I'm not all that familiar with our pin allocation stuff.
worn lark
I'll see if that is maybe the problem. Thanks so much for checking!
stuck elbow
cp uses sercom0 internally, all other is free to use
worn lark
do I need to set somewhere which sercom I want to use?
stuck elbow
no, it guesses based on the pins
worn lark
okidok! thanks ๐
stuck elbow
but several pins share the same sercom, so if you already use an i2c/uart/spi on some other pins, it may be in use already
worn lark
jep, that might be the problem in my case!
stuck elbow
swapping the order of initialization may help, because some pins have several sercoms, and it uses the first free
ivory yew
yeah, that
ionic elk
Isn't SWCLK a hard-protected pin, though, since it's required for debugging? I remember seeing some conversation about someone wanting to free it for another project... maybe that was another port though
idle owl
@hybrid scarab Sphinx always strikes. If you run into anything that you're giving more than x-amount of time (you choose what's reasonable), please ping me. I've been through all sorts of ridiculously obtuse Sphinx issues, and may have solutions for you in my head already.
hybrid scarab
@idle owl likewise (enough to never touch Sphinx again if I can help it), I think my issue might be a general is-this-the-way-CircuitPython-would-do-it issue as much as a Sphinx issue
idle owl
That is also distinctly possible.
hybrid scarab
Once upon a time I was naรฏve and full of gusto and thought "I'll do Sphinx docs for all our libraries!"
idle owl
I mean, we do. And we fight with it from time to time, but for the most part, the fighting is minimal. After we finally sorted out the kinks.
hybrid scarab
I raised an issue to myself for the particular case - https://github.com/pimoroni/Pimoroni_CircuitPython_MICS6814/issues/1
Yeah it's like anything else, learn the basics and stick to the same pattern and it'll all just kinda work as long as you don't get too fancy
A lot of my early problems stemmed from just being an inexperience Python developer coming in with some radical ideas about how user-facing libraries should work- and- oh boy- Sphinx didn't like those!
idle owl
Your issue might be solved by autodoc_mocking analogio. In conf.py in the docs folder (if you're following our standard anyway). Assuming analogio does work in your lib successfully?
hybrid scarab
It does, but I'm not 100% sure if I should have the analog setup in my lib, versus expect the user to pass in an analog pin explicitly
idle owl
Ah. Fair enough. So you're still working on the API.
hybrid scarab
Aye, yes! It can change if it's better
idle owl
Here is the magical bit you might need if it turns out you'd rather leave it that way: https://github.com/adafruit/Adafruit_CircuitPython_CircuitPlayground/blob/a0692dfbdd05ff0bfbcbad7294ac9c6f694b0404/docs/conf.py#L29
hybrid scarab
Oh wow, that's... a thing
idle owl
That one is a special one, needing a TON of autodoc_mock_imports, but it's ok to do only one thing there if necessary.
hybrid scarab
I spent a good 20 minutes searching for an example of my error ๐
idle owl
Basically it makes Sphinx not try to utilise that function.
hybrid scarab
Probably calls "mock.MagicMock()" or something behind the scenes to stub it all out?
idle owl
In CircuitPython, a lot of things aren't available outside of the board using them because the libs are separate from the core.
So we have to add them there to get Sphinx to stop trying to use them.
I don't know if it will solve your particular issue, but it might.
hybrid scarab
I have similar issues with testing all of my normal Python libraries- having to mock and stub a whole bunch of stuff out. Though when you're testing you want the mock objects to verify they've been called a certain way/etc
hybrid scarab
Yeah I think this is the right answer in terms for Sphinx anyway, thank you!
idle owl
You're welcome ๐
hybrid scarab
Not sure about the API though- I couldn't find another device library that uses a bunch of analog pins
hybrid scarab
Might still be a pattern to follow
idle owl
hybrid scarab
of course I'm almost sure I'd be likely to find both approaches in the wild
idle owl
The base for this assumes you're connecting a raw character LCD, which require a LOT of wires. versus the I2C/SPI versions which use breakout adapters to make it use less pins.
hybrid scarab
Oooh this link answers a question I never even thought to ask, that docstring is niice
idle owl
๐ Thanks.
hybrid scarab
But yes it looks like it expects an object of type digitalio.DigitalInOut, rather than a raw "pin"
idle owl
Check out the Circuit Playground Library if you want to see some serious documentation. It's one of the best documented libs we have.
Here's an example of the code: https://github.com/adafruit/Adafruit_CircuitPython_CharLCD/blob/master/examples/charlcd_rgb_simpletest.py
Or code that uses the RGB version, but still.
Same idea for the first set of pins.
I guess it depends on how interactive you want the user to be. The Circuit Playground library handles EVERYTHING in the background. https://github.com/adafruit/Adafruit_CircuitPython_CircuitPlayground/blob/a0692dfbdd05ff0bfbcbad7294ac9c6f694b0404/adafruit_circuitplayground/circuit_playground_base.py#L82
hybrid scarab
Okay some precedent for this approach is very helpful
idle owl
But that's not exactly a common thing. CP lib is a wrapper library.
hybrid scarab
I kinda feel this is the right approach- asking the user to supply a specific pin type
idle owl
Excellent.
hybrid scarab
On the off chance that they supply the wrong type of pin, I suppose
idle owl
You can throw an exception.
With a helpful error message
Supplied pin type must be digitalio.DigitalInOut() or some such.
hybrid scarab
Aye! thank you
idle owl
You're welcome!
hybrid scarab
Right- now to continue making a mockery of the FT232H setup ๐
idle owl
Good luck ๐
prime flower
@hybrid scarab you could do an if not isinstance(pin, TYPE): raise RuntimeError("Supplied pin type must be digitalio) with kattni's well-written error
worn lark
oh guys, sorry. I was not aware that I can not freely choose where miso/mosi/sck go on the sercom pads. I think it's working now ๐
hybrid scarab
I like that the runtime type checking actually means I still have to mock AnalogIO for Sphinx ๐
Yay! FT232H set up clean on a Windows 10 laptop that's never even had Python installed
Now I need to take it off my lap because it's a hot day and I'm on fire
prime flower
couldn't find another device library that uses a bunch of analog pins
@hybrid scarab Have you seen this lib >> https://github.com/adafruit/Adafruit_CircuitPython_MCP3xxx/
specifically the mocked analogin class here >> https://github.com/adafruit/Adafruit_CircuitPython_MCP3xxx/blob/master/adafruit_mcp3xxx/analog_in.py
hybrid scarab
That would appear to have the same pattern as the DigitalIO example, which is a two-for-two confirmation that I need to slightly refactor
Enough for me to assume that "pass in a pin of a specific type" is a rule of thumb
So I can replace:
def __init__(self, ox_pin, red_pin, nh3_pin, enable_pin=None):
self._enable_pin = None
if enable_pin is not None:
self._enable_pin = digitalio.DigitalInOut(enable_pin)
self._enable_pin.direction = digitalio.Direction.OUTPUT
self._enable_pin.value = True
self._ox_adc = analogio.AnalogIn(ox_pin)
self._red_adc = analogio.AnalogIn(red_pin)
self._nh3_adc = analogio.AnalogIn(nh3_pin)
with something like:
def __init__(self, ox_pin, red_pin, nh3_pin, enable_pin=None):
self._enable_pin = enable_pin
if enable_pin is not None:
self._enable_pin.direction = digitalio.Direction.OUTPUT
self._enable_pin.value = True
self._ox_adc = ox_pin
self._red_adc = red_pin
self._nh3_adc = nh3_pin
(docstrings and type checking notwithstanding)
prime flower
much cleaner!
hybrid scarab
Let's see if this build passes, whee!
manic glacierBOT
orchid basinBOT
ionic elk
@slender iron @tulip sleet Dahanzimin is sending me some CH579 chips, which is a 5 RMB/.72 cent BLE chip with equivalent ram and flash to the SAMD21. Wants to know if they'd be accepted into Circuitpython - I figure there's no reason they wouldn't?
manic glacierBOT
stuck elbow
so, I don't want to introduce panic, as I don't have a reproducer, but does anybody else get kernel panics on Linux due to usb stuff with the 6.x versions of CircuitPython?
I just had a third one
lone axle
I think I recall seeing something about people having USB related kernel issues on Linux recently, but I'm not sure if it was specific to 6.X
I think maybe Danh looked into it possibly
stuck elbow
I have seen him working on the issue that happened when you reformatted the filesystem
but this is when you disconnect the board
(or reset it)
lone axle
Ah that might have been it then, and a different thing. I don't have any experience using Circuit Python USB devices on Linux yet personally. The time is coming though.
slender iron
@stuck elbow I know @solar whale has been seeing this. do you have intel usb host chips?
stuck elbow
this is most likely a Linux bug anyways, external devices shouldn't be able to cause crashes like this
slender iron
@ionic elk we'll accept it but not do the work to add it
stuck elbow
@slender iron not sure, but possible, this is X1 Carbon
slender iron
@hybrid scarab in general I like to pass objects in because then the pin can be provided by a different driver
stuck elbow
0a:00.0 USB controller: Fresco Logic FL1100 USB 3.0 Host Controller (rev 10)
stuck elbow
but also 07:00.0 System peripheral: Intel Corporation JHL6540 Thunderbolt 3 NHI (C step) [Alpine Ridge 4C 2016] (rev 02)
slender iron
I think jerry had an intel thunderbolt controller too
last time we had an issue on intel chips it was a timing issue in tinyusb
solar whale
My Linux box has become almost unusable for CP -- It often hangs when I disconnect. I've moved to my Mac and Raspberry Pi 4s -- no issues there.
stuck elbow
the year of Linux on the desktop \o/
slender iron
do we have a tinyusb issue? can we repro with a tinyusb example?
stuck elbow
@slender iron I don't think this is a tinyusb issue, since the moment you disconnect/reset it, it doesn't matter what it does
slender iron
I know dan reach out to kernel devs about it
solar whale
It is a relatively recent (last few months) problem. for me I seemed to really notice it about the time I started using esp32s2's -- I thought it was related...
slender iron
we have updated tinyusb to get s2 support
slender iron
I bet checking a beagle trace may shed some light
solar whale
"beagle trace" ??
slender iron
a usb trace using a beagle device
solar whale
IF you can give me some examples of how to test tinyusb, I'll be happy to experiment with it.
slender iron
this example is a dummy cdc and msc device
these are the supported BOARD= values: https://github.com/hathach/tinyusb/tree/master/hw/bsp
solar whale
Thanks -- I see what I can break....
slender iron
thanks as always!
solar whale
@slender iron FYI -- for the esp32s2, it looks like I have to use the example in cdc_msc_freertos. Should I try that on all or use the cdc_msc on the others?
slender iron
@solar whale non-freertos on the others
solar whale
ok -- first test on esp32 did not kill anything ....so far so good
solar whale
@slender iron have tried esp32s2 and feather_m0 express -- several connect/disconnects TinyUSB_MSC comes up with no issues so far. -- all on Linux box.
slender iron
hrm. interesting!
manic glacierBOT
solar whale
hmm -- gets a bit confused if I plug in both at the same time -- does not appear to recognize the second one
manic glacierBOT
solar whale
or -- it recognizes it, but does not show a new drive mounted
maybe expecting too much of the example...
solar whale
yes -- dmesg shows the second device as "sdd" first is "sdc" no errors just does not mount the "sdd" device. Don't think I need to pursue that.
hmm -- RPI -- mounts both -- maybe another Linux issue
Also tried on feather_m4_espress -- no issues connecting/disconnecting
silver tapir
@solar whale Some linux automounters (that would be software on the desktop side, not on the kernel) have a timeout in case a device is being read twice. As long as it shows up in dmesg and you can mount it manually, it should be ok.
solar whale
yes -- both seem to be working fine. I won't worry about this. So far I can't crash the linux box with this example.
manic glacierBOT
solar whale
@stuck elbow FYI --- when my Linux box hangs, I dont even see a kernel panic it just freezes immediately on disconnect of the USB -- I have to power cycle. Nothing reported in the logs.
stuck elbow
same here
manic glacierBOT
stuck elbow
I just assumed it hides the kernel panic message, since it's an ubuntu derivative and they don't like to confuse users with error messages
solar whale
@slender iron At this point, it looks like TinyUSB is well behaved on Linux on the M0,M4,esp32s2 --- I'll keep trying things and let you know if anything odd comes up, but it seems pretty solid so far.
slender iron
ok, so it may actually be circuitpython then
or the older version of tinyusb we're using
or some timing thing with msc interface
solar whale
or something else ๐
slender iron
yup
solar whale
Thats a nice test -- you can connect to /dev/ttyACM0 and it prints a message and echos keystrokes as well.
slender iron
yup! and the fs is held in ram
solar whale
@slender iron I put CP (tip of main) back on my feather_m4 express and it hung the linux box on the first disconnect ....
slender iron
the CP tinyusb version will be a little different
but I doubt that is it
may be timing related that changes with the slow writes to real flash
solar whale
do you want me to document these tests somewhere?
slender iron
ya, on the issue would be great
solar whale
your new issue -- I can't find another one related?
tulip sleet
@solar whale do you have crash dumps in /var/crash? You'll need to sudo to see them. If not, you can apt install linux-crashdump (I think) to get them to show up. there. The dmesg file is the most interesting, and will show a stack backtrace.
slender iron
a new one? is there not one yet?
solar whale
I'm not finding one.... @tulip sleet do you know of a related issue?
tulip sleet
i am late to this party, but do you mean, for example https://github.com/adafruit/circuitpython/pull/3223
solar whale
thats it. thanks
tulip sleet
i'm not sure there was an issue
solar whale
for linux-crashdump -- it is asking if I want it to boot to kexec on a reboot -- Do I?
tulip sleet
i don't remember; what is the default?
solar whale
oh - thats a PR -- Do you want me to add my notes on tinyusb testing to it or start a new issue for these linux crashes
looks like no -- I'll take it
took all defaults
tulip sleet
yes, a new issue would be good. I think there are multiple cases. The storage.erase_filesystem() was just the easiest way to get a crash, but I was beaten down in the linux-usb mailing list saying that I shouldn't expect it not to crash. But I also see crashes just on unplug, which is not necessarily the same thing. That should not cause crashes, because anyone can pull a USB drive suddenly. I am wondering if it's a FAT12-handling issue.
I have seen crashes on unplug only recently.
solar whale
AS I noted, I seem to recall them starting about the same time I started using the esp32s2. I initially thought it was only an esp32s2 issue. FYI -- what i just dis was build the tinyusb example/device/cdc_msc for feather_m0,feather_m4,esp32s2 and did several connect/ disconnects on each with no problem on the Linux box.. Then I restored latest CP to feather_m4 and it hung the lInux box on first disconnect.
tulip sleet
it's worth reading the bug report I submitted and the linux-usb correspondence. Could you add those to your new issue also?
(they are mnetioned in the PR)
raven canopy
@tulip sleet @solar whale a few months ago when i was going through this, i could reliably repro with microcontroller.reset into bootloader.
solar whale
@raven canopy is that the same isue dan was seeing associated with File Sytem reformat? For me it is on physical disconnect of the USB plug
raven canopy
ahh. nevermind then...
crimson ferry
@slender iron did random get explicitly removed from the latest esp32s2 builds?
tulip sleet
@raven canopy, no, it could be the same thing. I think there may be two problems: one is crash on disconnect; the other is crash on quick reconnect. The disconnect crash may happen if there is some ongoing activity on the USB drive.
Please do take a look at the issue after Jerry creates and add your two cents.
slender iron
@crimson ferry maybe. I think @ionic elk may have disabled it because we don't support the true random number generator
crimson ferry
oh ๐ฆ OK, thanks
solar whale
@tulip sleet once I have the linux-crashdump installed, how to I make use of it after I reboot from a crash
slender iron
@crimson ferry you could turn it back on. it'll use the clock to seed the random generator otherwise
tulip sleet
just look in /var/crash as root. You'll see some dated directories. e.g. I have this:
halbert@salmonx:~$ sudo ls -R /var/crash
/var/crash:
202008050000 202008071843 linux-image-5.4.0-42-generic-202008050000.crash
202008052303 202008072211 linux-image-5.4.0-42-generic-202008052303.crash
202008071839 kexec_cmd linux-image-5.4.0-42-generic-202008071839.crash
/var/crash/202008050000:
dmesg.202008050000 dump.202008050000
/var/crash/202008052303:
dmesg.202008052303 dump.202008052303
/var/crash/202008071839:
dmesg.202008071839 dump.202008071839
/var/crash/202008071843:
dmesg.202008071843 dump.202008071843
/var/crash/202008072211:
dmesg.202008072211 dump.202008072211
halbert@salmonx:~$
the dmesg files are what's readable. They'll have a traceback.
I cannot remember everything I did to set this up. I did something more elaborate and I'm not sure exactly what is generating these. But you don't have these now, is that right?
solar whale
correct
/var/crash/:
total 6204
drwxrwsrwt 2 root whoopsie 4096 Aug 11 15:33 .
drwxr-xr-x 14 root root 4096 Apr 23 03:42 ..
-rw-r--r-- 1 root whoopsie 0 Aug 11 15:33 kdump_lock
-rw-r--r-- 1 root whoopsie 312 Aug 11 15:33 kexec_cmd
-rw-r----- 1 jerryneedell whoopsie 6334327 Aug 10 20:54 _usr_share_discord_Discord.1000.crash
-rw-r--r-- 1 jerryneedell whoopsie 0 Aug 10 20:53 _usr_share_discord_Discord.1000.upload
-rw------- 1 whoopsie whoopsie 37 Aug 10 20:54 _usr_share_discord_Discord.1000.uploaded
ok -- just killed it -- but this time it is doing some sort of reboot on its own ....
oooh ```[sudo] password for jerryneedell:
/var/crash/:
202008111542
kdump_lock
kexec_cmd
linux-image-5.4.0-42-generic-202008111542.crash
_usr_share_discord_Discord.1000.crash
_usr_share_discord_Discord.1000.upload
_usr_share_discord_Discord.1000.uploaded
/var/crash/202008111542:
dmesg.202008111542 dump.202008111542
jerryneedell@jerryneedell-ubuntu-macmini:~$
it can take a while; that's a bit of a disadvantage
upload the dmesg and I can point out what to look for
if it's not obvious
solar whale
[ 554.275212] general protection fault: 0000 [#1] SMP PTI
[ 554.275222] CPU: 5 PID: 5454 Comm: kworker/5:1 Kdump: loaded Tainted: P OE 5.4.0-42-generic #46-Ubuntu
[ 554.275225] Hardware name: Apple Inc. Macmini6,2/Mac-F65AE981FFA204ED, BIOS MM61.88Z.0106.B0A.1509111654 09/11/2015
[ 554.275235] Workqueue: events acm_softint [cdc_acm]
[ 554.275242] RIP: 0010:acm_read_bulk_callback+0x11f/0x280 [cdc_acm]
[ 554.275247] Code: 44 24 10 48 c7 c1 70 a8 47 c1 48 c7 c2 5b a6 47 c1 48 c7 c7 48 d4 47 c1 48 8d 70 30 e8 9a ae 0d ee 4d 8b 34 24 e8 e1 14 cc e
d <49> 89 86 60 02 00 00 49 63 45 18 f0 49 0f ab 84 24 80 07 00 00 ba
[ 554.275250] RSP: 0018:ffffb303c01e8ca8 EFLAGS: 00010016
[ 554.275253] RAX: 000000811a85cab8 RBX: 00000000ffffffb9 RCX: 0000000000000018
[ 554.275256] RDX: 0000000000000000 RSI: 000000000062a4ce RDI: 0013ec2bf0fc0f53
[ 554.275258] RBP: ffffb303c01e8cc8 R08: 00000000ffffffb9 R09: 0000000000000080
[ 554.275260] R10: 0000000000000310 R11: 000000022ea4c310 R12: ffff93409dc8a000
[ 554.275263] R13: ffff93409dc8a668 R14: 6ace3cfc1dbe9fd2 R15: ffff9340a47ee000
[ 554.275266] FS: 0000000000000000(0000) GS:ffff9340a7140000(0000) knlGS:0000000000000000
[ 554.275269] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 554.275271] CR2: 00007f8e0d4fd000 CR3: 000000020900a001 CR4: 00000000001606e0
[ 554.275273] Call Trace:
[ 554.275289] __usb_hcd_giveback_urb+0x7a/0x120
[ 554.275294] usb_hcd_giveback_urb+0xca/0xe0
[ 554.275300] xhci_giveback_urb_in_irq.isra.0+0x7b/0xf0
[ 554.275304] xhci_td_cleanup+0xf6/0x140
[ 554.275308] finish_td+0xcc/0x190
[ 554.275313] handle_tx_event+0x55d/0x11c0
[ 554.275318] xhci_irq+0x27a/0x3d0
[ 554.275322] xhci_msi_irq+0x11/0x13
[ 554.275330] __handle_irq_event_percpu+0x42/0x180
[ 554.275335] handle_irq_event_percpu+0x33/0x80
[ 554.275339] handle_irq_event+0x3b/0x5a
[ 554.275343] handle_edge_irq+0x93/0x1c0
[ 554.275347] do_IRQ+0x55/0xf0
[ 554.275353] common_interrupt+0xf/0xf
[ 554.275356] </IRQ>
[ 554.275362] RIP: 0010:vsnprintf+0x34e/0x4e0
I think this is where it happened
onyx hinge
This feels like a bug, anybody have opinions? The lock on the board.SPI() object can survive soft resets! (nrf52840 particle xenon): ```>>> import board
board.SPI().try_lock()
True
soft reboot
...
Adafruit CircuitPython 6.0.0-alpha.2-190-g817a52228 on 2020-08-11; Particle Xenon with nRF52840
import board
board.SPI().try_lock()
False
yep, that's what they want to see. It would be interesting to know if this is this pretty much the same for all crashes or is different. If you (and anyone else, e.g. @raven canopy) can start collecting these in a new issue then I can bring it up again on the linux-usb mailing list (maybe only to be beaten down again).
solar whale
OK -- in this case _ was connected to the REPL when I unplugged. I don't think that is relevant but will do more tests.
tulip sleet
i think it might be quite relevant, if related to CDC
@onyx hinge it does look like a bug
raven canopy
i never could get crashdump to grab the panic on the Pi, and then it stopped happening. ๐ฆ
raven canopy
all of mine were on ubuntu server; didn't see anyone have issue on Raspbian.
tulip sleet
@onyx hinge is it just that board, just nrf, or is it across ports?
solar whale
@tulip sleet let me know what I left out of the issue... I'll keep adding reports and crash dumps -- but probably not until tomorrow....
onyx hinge
@tulip sleet I only checked nRF so far
simple pulsar
@tulip sleet Hello. Did you have any thoughts on achieving predictable ordering of the ManufacturerData fields in Advertisement subclasses? https://github.com/adafruit/Adafruit_Learning_System_Guides/pull/1185#issuecomment-670944106
tidal kiln
what's the best way to clear out a displayio.Group? i.e., remove all items that have been added via append()
simple pulsar
@tidal kiln I don't know if it's the best way but I use this for clean-up of Group objects ``` def emptyGroup(self, dio_group):
"""Recursive depth first removal of anything in a Group.
Intended to be used to clean-up a previous screen
which may have elements in the new screen
as elements cannot be in two Groups at once since this
will cause "ValueError: Layer already in a group".
This only deletes Groups, it does not del the non-Group content."""
if dio_group is None:
return
### Go through Group in reverse order
for idx in range(len(dio_group) - 1, -1, -1):
### Avoiding isinstance here as Label is a sub-class of Group!
### pylint: disable=unidiomatic-typecheck
if type(dio_group[idx]) == Group:
self.emptyGroup(dio_group[idx])
del dio_group[idx]
thanks. i'd want to remove tilegrids also.
there's remove and pop, but i'm seeing something odd trying to use those
simple pulsar
Actually the description is wrong
tidal kiln
it looks correct? you're checking type and only removing Groups?
simple pulsar
That's for the recusive destruction of any Group within a Group
I think I evolved the whole thing and didn't revisit the description
tidal kiln
oh. i see. recursive call.
stuck elbow
@tidal kiln I think you should be able to just do dio_group[:] = []
the garbage collector will take care of the recursive destruction
tidal kiln
>>> splash[:] = []
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
NotImplementedError: Slices not supported
@stuck elbow @tidal kiln Does group allow you to do that? I'm not sure if it has a full implementation of all the bits for sequences
Ah, indeed
stuck elbow
ouch
tidal kiln
@stuck elbow like that? looks like there's a slice support issue maybe?
stuck elbow
yeah, I can't understand why the groups are their own thing instead of being python collections
alternatively, you can just make a new group and let the garbage collector free the old one
tidal kiln
yah, i'm think that may be the way.
stuck elbow
just make sure no references to it are left anywhere
simple pulsar
Have you hit the ValueError: Layer already in a group ?
stuck elbow
I only ever have one group in my programs
tidal kiln
not sure if this is a bug?
Adafruit CircuitPython 5.3.1 on 2020-07-13; Adafruit PyPortal with samd51j20
>>> import displayio
>>> splash = displayio.Group()
>>> layer1 = displayio.Group()
>>> layer2 = displayio.Group()
>>> layer3 = displayio.Group()
>>> layer4 = displayio.Group()
>>> splash.append(layer1)
>>> splash.append(layer2)
>>> splash.append(layer3)
>>> splash.append(layer4)
>>> len(splash)
4
>>> for layer in splash:
... print("removing")
... splash.remove(layer)
...
removing
removing
>>>
You're modifying something you are iterating over
You need to know the per-language rules for doing that.
stuck elbow
it's hard to write that nicely without slicing
maybe something like:
try:
while True:
splash.pop()
except IndexError:
pass
could you just del splash and re-create?
simple pulsar
In some cases splash will be being displayed.
lone axle
maybe:
while len(group) > 0:
group.pop()
@lone axle maybe
Adafruit CircuitPython 5.3.1 on 2020-07-13; Adafruit PyPortal with samd51j20
>>> import displayio
>>> splash = displayio.Group()
>>> layer1 = displayio.Group()
>>> layer2 = displayio.Group()
>>> layer3 = displayio.Group()
>>> layer4 = displayio.Group()
>>> splash.append(layer1)
>>> splash.append(layer2)
>>> splash.append(layer3)
>>> splash.append(layer4)
>>> len(splash)
4
>>> while len(splash) > 0:
... print("removing")
... splash.pop()
...
removing
<Group>
removing
<Group>
removing
<Group>
removing
<Group>
>>> len(splash)
0
>>>
and that behavior with remove may be what @simple pulsar mentioned
tulip sleet
@simple pulsar Sorry, I missed your comment. I just answered now. I get about 500 emails a day and sometimes I miss one, especially because there's a lot of chaff in my GitHub emails.
simple pulsar
@tulip sleet NP, thanks, I'll go through it tomorrow,
@tidal kiln BTW, unrelated to last discussion, there's a guy in forums with a genuine bmp with negative height and OnDiskBitmap doesn't like it: https://forums.adafruit.com/viewtopic.php?f=65&t=168292
tidal kiln
@simple pulsar thanks. i'll take a look.
lone axle
@simple pulsar @tidal kiln if they are not worried about memory usage and want a way to show the image without converting it I believe the imageload library should support bmp images with negative height.
slender iron
yeah, I can't understand why the groups are their own thing instead of being python collections
@stuck elbow because I had enough to do when I implemented them and a static allocation up front doesn't secretly grow
lone sandalBOT
onyx hinge
and because a list couldn't enforce the "a tilegrid is in at most one list" invariant, which could lead to interestingness ๐
silver tapir
The site is live. I still need to fix a bit of spanglish and minor things.
https://diacircuitpython.org/
Hosted via IPFS and build with Hugo.
SVG source for Quetzalblinka is in the site's blog.
soft patio
This has probably been asked many times before. What toolchain is used when building Circuitpython 6? I'm assuming it has changed since 9-2019-q4-major for Circuitpython 5. Or do I assume too much?
stuck elbow
@soft patio still GCC
soft patio
@stuck elbow Still 9-2019-qr-major or 9-2020-q2-update or maybe even 10-2020-q2-preview?
soft patio
In the Adafruit documentation for building, the mention specific versions of the toolchain: "For CircuitPython 5 development we are using the 9-2019-q4-major version. If you want to build CircuitPython 4, use the 7-2018q2-update version." I'm guessing then that it isn't a significant issue which recent verions I use (but avoid 10)?
thorny jay
Reproducible build? Did we try and compare automatic build on GitHub and manual build at home to see if it is binary exactly the same result? It is always reassuring to have exactly the same result both from a security point of view and to know everything is properly setup.
solar whale
a few months ago when i was going through this, i could reliably repro with
microcontroller.resetinto bootloader.
@raven canopy When you had this issue -- what board were you using? With a feather_m4_express, I triedimport microcontroller microcontroller.on_next_reset(microcontroller.RunMode.BOOTLOADER) microcontroller.reset()and it reset into BOOTLOADER OK on a RPi 4
hybrid scarab
Connector or socket on the feather board? I just opened the M4 Feather and our Enviro+ FeatherWing and they both come with breakaway pin headers ๐
solar whale
I usually use these on the MCU board -- then I can add a featherwing or put it on a breadboard, https://www.adafruit.com/product/2830
solar whale
hybrid scarab
Pretty sure I've got some of those floating about, possibly, unless Nat stole them for his feather ๐
Though I might wait for midday to pass before I brave The Great Outdoors
stuck elbow
note that male headers should always point down
female up
this way you can plug it into a breadboard
hybrid scarab
You can still plug it into a breadboard if the headers point up ๐คฃ for a fun challenge transposing the pinout diagram.
stuck elbow
you can't see the leds and don't have access to the reset button then
and if you are not consistent, half your wings won't fit
hybrid scarab
Hard mode!
stuck elbow
I had to make some adapter boards with female headers on both sides
hybrid scarab
I was a little tempted to do that
stuck elbow
wasn't easy to solder
stuck elbow
I inserted the female headers in the same holes on both sides
and soldered them from the side
hybrid scarab
๐ฌ
hybrid scarab
yeah Iโve got a box of wonders in my office somewhere that I think has some of those left in from some arduino boards
raven canopy
@solar whale it was any board for me. Metro M4, Grand Central, and a Feather M0. Again though, this was running Ubuntu Server, and it stopped happening.
hybrid scarab
@timber mango did you give your Pinout.xyz for CircuitPython idea any more thought?
I don't know if you've seen it- but there's a micro:bit one now too- which Josh (EduBlocks) mostly cleaned up and finished for us - https://microbit.pinout.xyz/
Though I think the variety of host boards in the Feather ecosystem might change up the format quite considerably
tidal kiln
is something changing in 6.x that's going to allow long ints on smaller boards, like trinket m0?
Adafruit CircuitPython 6.0.0-alpha.2 on 2020-07-23; Adafruit Trinket M0 with samd21e18
>>> import time
>>> dir(time)
['__name__', 'localtime', 'mktime', 'monotonic', 'monotonic_ns', 'sleep', 'struct_time', 'time']
>>> time.monotonic()
90.376
>>> time.monotonic_ns()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
NotImplementedError: No long integer support
>>>
or are those extra time methods just slipping through for some reason?
onyx hinge
@tidal kiln no, long int support for those boards is not likely to happen. @slender iron do we prefer NotImplementedError in these cases, vs not having the member at all? It is a trade off since you can't say if hasattr ....
tidal kiln
for ref, here's 5.3.1:
Adafruit CircuitPython 5.3.1 on 2020-07-13; Adafruit Trinket M0 with samd21e18
>>> import time
>>> dir(time)
['__name__', 'monotonic', 'sleep', 'struct_time']
>>>
I prefer the exception because it actually explains the issue
tidal kiln
is that the change then for 6.x? putting the methods back in and then throwing an exception?
slender iron
at least in that case, check the commit log
onyx hinge
Fix for Issue #2812. Instead of reporting a missing attribute for functions such as time.time() and time.mktime(); platforms that do not have long integer support will raise a NotImplementedError
tidal kiln
@onyx hinge ^^ thanks!
solar whale
@onyx hinge have we ever considered not offering all translations for the smaller boards?
onyx hinge
@solar whale yeah @slender iron prefers not to approach it that way
solar whale
ok
slender iron
languages are important
onyx hinge
The most advanced user can get the source, reenable what we had to disable for languages, and get their customized build in US english. A new user who doesn't have the knowledge to do any of those things benefits most from havning the translated core. All that said, I keep feeling like each time we hit size limits on M0 that "maybe this time it's the time we have to make the undesirable decision" and compromise on translations... I will work hard each time to find other ways, though.
slender iron
there are other things we can optimize instead
the uart pin finding logic is large and could be simplified for boards with few pins
fringe trench
This may be a completely off-base idea.
Why do the translated languages take up more space in the binary?
Is it a function of the character sets we're using for the translations?
slender iron
ya, the unicode strings tend to be longer (ja) and the words can be longer too (de)
fringe trench
I think UTF-8 isn't always the most efficient encoding, but I do think it's the standard.
๐
fringe trench
I guess I just need to get my dev environment set up and start hammering on this.
slender iron
help is always welcome ๐
fringe trench
Well, before I lost power I ordered 9 Espressif boards to poke at S2 support, because I guess that's a thing that needs to happen?
solar whale
@slender iron I agree that languages are important. I was just wondering if a compromise on the most constrained boards would be better for all. Would the users prefer more functionality over more languages. Just a thought. Sounds like it has been considered and dismissed. That's fine. As noted, people can make custom builds ...
fringe trench
Still have to try to get everything updated on this machine.
tulip sleet
a translation could be put in a file that lives in CIRCUITPY, leaving more room for firmware, but that would still make some languages diffrent
onyx hinge
@fringe trench fwiw the translated strings are huffman coded UTF-16, with an optimization applied when none of the characters in the translation are above unicode 255.
manic glacierBOT
thorny jay
I don't know if you've seen it- but there's a micro:bit one now too- which Josh (EduBlocks) mostly cleaned up and finished for us - https://microbit.pinout.xyz/
@hybrid scarab Sorry, no I did nothing, except suggest it should exist and Adafruit could spy on your code. Yes I noticed the micro:bit version and was unsure if it was really Josh. Also as far as I remember his EduBlocks also work at least with the CPX. The reality today is that even for some Adafruit Feather Wing, it is unclear what pin are in use. Sometime the best hint are parenthesis (.) (.) . around the used pin. I don't worry about the host boards, I just worry about what is physically in use and connected. One difficulty is to handle board where you can select the pins you enable or not with some soldering bridge or cut.
I am "good" at having idea and making suggestion (like a zip file always up to date with the library needed for a project or a board). But I don't have the time nor skill to implement that.
thorny jay
note that male headers should always point down
@stuck elbow Except if you use double sided breadboard such as this https://www.kickstarter.com/projects/252587878/cake-board-new-lego-friendly-solderless-breadboard then you can have the Feather on top and the MCU on the other side. Pimoroni was selling something like that, but I could not find the product anymore (but I have one or two at home). Cc: @hybrid scarab
onyx hinge
@tulip sleet cpx_displayio: how do you feel about dropping rotaryio instead of audiobusio? Otherwise, I can add code to only enable pdmin and not i2sout and it'll probably start fitting maybe
probably nothing applicable here but I'll read it in case https://www.unicode.org/notes/tn14/UnicodeCompression.pdf
actually SCSU seems kinda maybe relevant, as a compression step before huffman... though they don't really discuss how well these techniques work on CJK scripts
the script that does the compress step could really use some organization to facilitate testing alternatives like this .. though the size of the decoder in "C" / bytes of ARM code is pretty critical too, you can't judge "success" from just decreasing the compressed translation message size
onyx hinge
I had read that some proprietary toolchains can compress the part of the flash memory that holds data that is stored in RAM but is initialized other than with all zeros .. but then I checked, and the savings in CircuitPython would probably be negative, since the decompressor costs memory size. trinket_m0 only has 276 bytes in its data section, for instance
SCSU appears to need a LOT of memory by CircuitPython standards
(it has either 128 or 256 "windows", not sure yet. 256 of anything is a lot)
onyx hinge
Without fully elaborating it, it's no sure thing that I got the code right.. but I made a "SCSU-like" encoding. Characters from 32 - 127 encode themselves, 1 is a special "page change" byte (followed by page number) and characters from 128 to 256 encode a non-ASCII character within a page. It encodes the Japanese translation more densely than UTF-8, but huffman-compressing it is worse than huffman-compressing the UTF-16.
there are 104 distinct "pages" used in the Japanese translation
encoding the "page shift" operation as code values starting at 256 is better than encoding them as a shift byte followed by the page number, but not by enough
only about 50% of the time are two non-ASCII symbols from the same page
orchid basinBOT
manic glacierBOT
onyx hinge
Because some of the texts found by preprocessing are not actually in the final binary, any scheme which compresses everything together will include unused strings and make things bigger. In particular, these
if (MICROPY_ERROR_REPORTING == MICROPY_ERROR_REPORTING_TERSE) {
terse_str_format_value_error();
} else {
mp_raise_ValueError(translate("single '}' encountered in format string"));
}
So saying "oh let's use a better compressor on the whole corpus (and e.g., separate messages with the \0 byte)" is unlikely to pay off. You've got to pay for the better decompressor's code size AND for the unused messages AND the runtime cost of going through up to half of the message catalog every time to find what you're looking for.
these also throw off the huffman weights a bit but probably not much.
stuck elbow
@onyx hinge I'm curious, how is the problem from your example solved?
how do you determine if a string is actually in the binary from a run-time check like the one you quoted?
onyx hinge
LTO is able to see the message is never produced so it's translated form is omitted from the binary. The test is actually a compile time constant
stuck elbow
but we don't use LTO? I remember there were problems with it
onyx hinge
We do for samd and nrf, not for others I think
There are always problems initially turning it on, so for the ports that seem to have a lot of space it's off
stuck elbow
so how many strings are like that? what order of magnitude are we speaking about here?
onyx hinge
I think the "terse" is the most common cause for this, so you could count them easily. Dozens I think.
But I'm almost certain they disappear from our most constrained m0 builds
stuck elbow
if it's mostly one variable, wouldn't it be easy to work around it?
for example, by putting them in an additional translation file
that is also conditionally included
onyx hinge
You could change a few to #If and see if the binary got smaller. If it did, we'd want to do it across the board
stuck elbow
does the code that collects strings for translation understand preprocessor directives?
because most tools I've seen don't
onyx hinge
Yes it does, makeqstrdefs
It determines which strings are still in the source after preprocessing
stuck elbow
nice
onyx hinge
Indeed
onyx hinge
If you felt like doing some mindless work .. there might be some reason for it I don't understand though
I'd check with tannewt before getting started though
unless you determined it did improve binary size, then we'd be for it for that reason
afk for awhile
tulip sleet
@onyx hinge I tried meowbit_v121 without removing _pypixelbuf, to see how much space it overflow by.But it still fits nicely:
m -rf build-meowbit_v121
halbert@salmonx:~/repos/jepler/circuitpython/ports/stm$ make BOARD=meowbit_v121 LOCALE=ja
Use make V=1, make V=2 or set BUILD_VERBOSE similarly in your environment to increase build verbosity.
QSTR updated
384664 bytes used, 8552 bytes free in flash firmware space out of 393216 bytes (384.0kB).
53464 bytes used, 44436 bytes free in ram for stack and heap out of 97900 bytes (95.60546875kB).
so I do not understand what is going on that you would see an overflow
onyx hinge
CI did see it though
CI is bulding with main branch merged in, did something change in main? (I didn't merge main in again)
afk for real this time
tulip sleet
8k bytes seems like a lot; ttyl
stuck elbow
@onyx hinge I don't mind mindless work, I'm a python programmer after all
tulip sleet
@onyx hinge never mind, my fault, i did LOCALE=ja, not TRANSLATION=ja ๐คฆ
solar whale
Nice -- you found a way to save 8K bytes ๐
onyx hinge
@tulip sleet ah the -Os solution is better than mine. Can you review and merge it now?
tulip sleet
@onyx hinge The stm Makefile doesn't have CFLAGS_INLINE_LIMIT or -flto; wasn't necessary so far, but there's plenty more to add for squashing STM builds
@onyx hinge did you try anything like converting ja to UTF-16 and then doing huffman 16 bits at a time? Id don't know the density of UTF8 vs UTF16 for characters that are mostly 16-bit.
manic glacierBOT
tulip sleet
@long forum I looked for info about software support for the RTL8720DN, and found absolutely nothing. Do you have any leads on that? If the RTL8720DN provides the HCI serial protocol to control it to use BLE, then it would be possible in the long run. I am implementing _bleio for HCI, targeting the ESP32, but another HCI-based BLE module should be easy to port to, with maybe essentially no changes. We would not do that work, but it would be possible.
mental nexus
Here's the next extension of my graphical understanding of the connections between CircuitPython and C, this time with a cursory look at how parameters are passed and decoded. With a few mentions about type conversions. Corrections and clarifications are welcome!
tulip sleet
@mental nexus that looks great! I have not proofed it in detail. We have something roughly similar here: https://learn.adafruit.com/extending-circuitpython, but it's out of date. Do you have any interest in turning what you wrote into a (major) revision of this guide or a new guide?
ionic elk
@onyx hinge have you talked to @fiery silo about font storage?
mental nexus
@tulip sleet I actually referred to that guide when I got started, but it is a bit terse for my current ability. I think the linear scrolling visuals of this guide made it hard for me to make all the connections. I scoured examples like this and tried making these "2D" connections (minus the yarn and push-pins on a CSI corkboard). I'd be glad to contribute back what I am learning from y'all. Let me know what your feedback and I can see how to convert this over to something usable by others. I've yet to make a learn guide but am willing to figure it out with a few examples and some guidance.
tulip sleet
Thanks, I will bring it up internally and we'll get back to you!
manic glacierBOT
ionic elk
@slender iron do you want me to leave that NEOPIXEL wroom label issue alone as something a beginner could contribute or can I fix it real quick
onyx hinge
@tulip sleet yes, the huffman codes are built off the UTF-16 code points and have been for some time
@ionic elk no, and realistically I won't have time to work on that anytime soon ๐ kmatch98 and foamyguy have been very interested and active with font and text stuff in displayio though, they might be resources
ionic elk
@onyx hinge I just mentioned it because it seemed like you were really diving into the storage
onyx hinge
ah that's the connection .. but we don't even have most of the ja characters, I don't think that's where the additional storage requirement is coming from. I say that because there are a large number of messages about characters not being in the font ..
ionic elk
Babel got me up and running with unicode for Japanese really fast which is why I brought it up.
onyx hinge
yasssssss we have a japanese translation now ๐ thank you danh
and thank you ciscorn whereever you are
ionic elk
So what is the Japanese translation in? Where is CJK stored?
onyx hinge
The messages in the core marked translate()
ionic elk
I guess you don't need any font support for the REPL but if you run DisplayIO what happens? Does it just appear as unicode binary?
Sorry for dumb question. I originally thought you fit the CJK fonts onto Circuitpython somehow but I guess the actual size reduction for the translations was just required because the actual translations ended up being bigger in terms of actual bytes?
I wonder if we could get Babel or just regular unicode fonts onto an SD card, and have them available for DisplayIO that way... or a CPY board with Babel built in as "Feather International".
ionic elk
yeah I should have arrived at that conclusion sooner, fitting CJK on CPY makes no sense
@tulip sleet what does happen with DisplayIO right now though?
in terms of displaying repl errors
tulip sleet
i think you just see junk; it can't display what it doesn't have
or boxes or whatever
ionic elk
Hmm, definitely seems like a devboard with a 4MB flash with 2MB used for Babel Unicode might be an opportunity there
Especially for something with a built in screen
tulip sleet
i think there's no one using just the screen for debugging, since they always have a host computer. It might be more an issue of printing native language strings, from the user program, as opposed to the canned messages. e.g., can I print "Hello, World", in Japanese with the terminal font? probably not ??
ionic elk
^ yeah no you def can't
the Babel thing I'm mentioning is the font package @fiery silo made for the Open Book, it's all the unicode characters squeezed into under 2MB, so it fits on a nor flash
But I'm just spitballing. Might be something to wrap into a future board aimed at international education or something like that, so it has fast native font support.
fiery silo
so i have displayio labels working in babel (at least for left-to-right languages, so far) but i definitely optimized more for speed and less for space... it works well on a board like the pygamer where it's a 2MB font file on an 8MB flash chip, but it would take up all the space on a board like the feather express
ionic elk
@fiery silo you beat any font I could find online for both speed and space at readable sizes. But honestly I'd say speed is the more important thing anyway, I'm still not sure if it was a bug but the last time I tried unmodified unicode on Circuitpython it took about 2s per character
Anyway I don't mean to put pressure on you lol, I'm just thinking that if I or anyone else were making a CPy board for international use, like for education or something, a 4MB flash chip with half used for Babel might be a really nice touch. Since we're already seeing small CPy communities spring up in Japan, China, etc.
Either that or you figure out something with compressed fonts via SDIO and ship the board with a font SD included, which would probably be slower
fiery silo
i dunno about speed (tho i did tests, and seeking in the font file on the QSPI flash was at least as fast as reading font data from the dedicated SPI flash chip), but my worry there is that if SD the card got removed, you'd no longer have any glyphs to print an error message in the user's language
and no pressure felt, no worries! ๐ in fact i thought about revising the open book to do away with the second flash chip and just have one 4MB QSPI flash for code and font data, once i saw speed wasn't a concern
summer belfry
Quick question if anybody is generous enough to answer. I just flashed my Seeeduino XAIO (Samd21) with circuitpython. The storage went from a little over 7 megabytes to 47 kilobytes. Why does this happen, and is it possible to unlock the rest of that storage?
stuck elbow
@summer belfry my guess is that it's now only using the internal flash for storage
@summer belfry where did you see the 7MB storage size?
slender iron
the 7mb was probably the fake uf2 partition
stuck elbow
it doesn't look like it has any external flash memory, looking at the schematic
you could probably make it a haxpress, but that's some effort
summer belfry
Yeah the datasheet said it only had 256kb of flash memory, but I dragged a 7mb file onto it before flashing and it worked fine. Honestly confused
stuck elbow
the uf2 filesystem is simulated, it doesn't actually store the files
it's just a way to flash the device
summer belfry
ahh. So you think there's any way to utilize the full 256kb of flash, or not without advanced knowledge? I'm a normie
stuck elbow
you are already utilizing it โ the circuitpython firmware itself takes up the rest of space
summer belfry
I see. Thanks!
mental nexus
How do you measure the amount flash used for a given CIrcuitPython build?
slender iron
it's output at the end of the build
ionic elk
we don't have that for ESP32-S2 yet, though, right? Or is there an option I should be turning on?
slender iron
correct, it isn't setup for -s2
you'll know when you go over though, it won't boot when you do
mental nexus
Wow how did I miss that? Thanks @slender iron
slender iron
the maximum size is fixed by the linker script usually
long forum
@tulip sleet no, no leads other than the mail I got from Seeed asking about potential support and them saying no.
idle owl
I've got this PR that uses finally in it, and I'm unclear on how it works. There is a while True: loop under a try: and then finally with a single command under it. Traceback implies that on ctrl+C in the serial console, it runs the code under finally. This does not clear up the why or how, simply what it's appearing to do.
@tulip sleet or @onyx hinge Do either of you happen to know and can explain the above? I don't want to merge something I have no grasp on.
onyx hinge
typing Ctrl-C creates a KeyboardInterrupt exception
tulip sleet
the code in a finally: runs no matter what, even if an exception happens (or not)
idle owl
The python docs imply that finally is meant to run the code in it regardless of whether try results in an exception.
Oh.....
onyx hinge
that causes all of the normal things that occur when an exception is raised to happen
onyx hinge
you could also katch the KeyboardInterrupt exception if you wanted
idle owl
In my mind there was no exception, so I was confused how it was doing anything.
onyx hinge
I am not sure whether there's a way to change how the finally processes in the case of a KeyboardException
tulip sleet
this is a good answer: https://stackoverflow.com/a/11552007/142996
idle owl
@tulip sleet That's what I read, but I don't understand how it's running that code no matter what since that stops the rest of the code from working. It only runs it on ctrl+C
Reading.
onyx hinge
the thing kjw is talking about here should probably be treated as a bug in the core and fixed. I encountered something similar recently with SPI. https://github.com/adafruit/circuitpython/issues/3266
tulip sleet
even if there is an exception (nothing is caught) in the try block, it will do i2c.unlock()
and if there's no exception, it will also unlock
idle owl
Then how does the while True loop work at all?
tulip sleet
the ctrl-c exception is not caught, but before it happens, we want to unlock
onyx hinge
using busdevice also takes care of this for you, but this example doesn't use it
idle owl
Yeah the I2C scan is super basic, it's supposed to be.
onyx hinge
(in a slightly different way)
tulip sleet
the while True scans the bus over and over, every two seconds. You can control-C out of it, and the program will stop, but before the program stops, we want to unlock. So this guarantees that happens
onyx hinge
but this only matters because of a bug -- if you lock you i2c and then re-run code.py in any way, it should not remain locked
(does reloading because of writing to CIRCUITPY also happen via KeyboardInterrupt?)
idle owl
Apparently it hangs pretty bad sometimes. With I2C scan. It totally did that with another sensor on my desk, I figured something was up with the sensor. Not sure.
Thank you so much @tulip sleet and @onyx hinge
tulip sleet
@onyx hinge There is an mp_reload_exception, but it doesn't have a name (you can't raise it from Python).
no wait, I am wrong, it does have a name, ReloadException. (bad grepping on my part)
onyx hinge
OK, but it happens via exception so you would get all the unlocks and finallys that are there. good.
tulip sleet
halbert@salmonx:~/repos/circuitpython$ ag reloadException
halbert@salmonx:~/repos/circuitpython$ ag ReloadException
are not the same thing
right, it delays until it would be safe to do a regular exception
that's Scott's idea
onyx hinge
makes sense to me
mental nexus
git question: I've have circuitpython fork called the old "master" that is way behind on commits. I made some updates on my local branch "bitmap_v2". I need to get my master up to date and merge it into my updates on my local branch. I think I need to reconnect somehow to the recently renamed "main" branch and then do some git commands to merge. I see some notes about "rebase" but I've heard that is dangerous! I'd be glad to have y'all git gurus to give me some incantations to use to reach git-nirvana.
raven canopy
@mental nexus this is a good article on how to do it: https://www.hanselman.com/blog/EasilyRenameYourGitDefaultBranchFromMasterToMain.aspx
mental nexus
Thanks sommersoft, will study that and see if I can get it figured out.
ionic elk
@slender iron I have a part of PulseIO that must be regularly serviced - FreeRTOS's ring buffer will crash the entire RMT if it fills up, so I need to periodically empty it. However, I can't access a callback based on the RMT interrupts (they're reserved by the idf). I tried to put these in background.c, but I'm finding that doesn't seem to consistently run. Do you have any suggestions?
Should background tasks be running as code.py does?
onyx hinge
you can "enable ticks" while your thing is running, that is intended to make the 1ms background task happen reliably.
supervisor_enable_tick when your peripheral is created (or starts capturing) and supervisor_disable_tick when it is deinitialized (or stops capturing)
@ionic elk
ionic elk
hmm. Interesting - ticks seem to be enabled the first time it runs off reset, but then maybe they get turned off.
A combination of that and a fix in the reset code fixed it, thanks @onyx hinge
onyx hinge
yeah there are probably bugs in that area
ionic elk
I'll check out the ticks stuff for ESP32 - I know there was an issue with STM32 where we accidentally had Systick on until a specific delay in Microcontroller was called that happened to turn it off. So it was disguising other errors until you called certain functions and then everything would break
Might be something similar here.
manic glacierBOT
mental nexus
@raven canopy thanks for the git pointer yesterday. I got my local branch renamed to main, and then I proceeded to pull my hair out thinking that it wasnโt updated when it actually was. My local only had a main now, but didnโt realize that github now had both a master AND a main branch on it. I spent an hour looking at the github status of master wondering why it wasnโt updated. Bleh... ๐คท
raven canopy
@mental nexus yeah, it can be a little daunting to keep local and GitHub in sync sometimes. did you update your GitHub to have main as the default branch? if you do that, you can go ahead and delete master afterwards to remove further confustion.
onyx hinge
fwiw I like to keep two remotes in my github -- "origin" (adafruit/circuitpython) and "jepler" (jepler/circuitpython). When I want to create a new branch, I "git fetch origin; git checkout -b new-branch-name origin/main^{}". I don't have to keep "my" main/master branch up to date at all.
I know that guides have suggested a different workflow but I have not grasped the benefit of doing it differently than I already learned how to do
raven canopy
for the core, i have adopted a similar system. i have a local with my fork as origin, a local with adafruit as origin, and then a local for each tagged version (result of docs work).
mental nexus
Just changed my default to main, thanks for the suggestion. Also it was making commits as my wifeโs name. I think I figured that out too.
onyx hinge
@ionic elk had another suggestion about makefiles and tab-completion. If you're using the lowercase-m makefile trick, you can add targets like board-name: ; $(MAKE) BOARD=board-name. Then make board-name becomes tab-completable with the standard make tab completion in bash-completion.
seems like you have to repeat this for each board but you only have to do it once
also if you want to build a lot of boards you can do that via make rules, Scott recently added something for this in the top-level Makefile.
well that's curious, but there's probably nothing to be done about it ... ```>>> framebuffer = sharpdisplay.SharpMemoryFramebuffer(bus, board.D2, 400, 240)
displayio.release_displays()
framebuffer
None
framebuffer is None
False
(it's an object of type NoneType, but it is not at the same memory address as NoneType)
onyx hinge
Does this seem like a familiar bug to anybody else? is my bundle just out of date? I have some labels that I keep changing, and the label position has moved left over time on the screen. it's using anchor_point / anchored_position. @lone axle ?
slender iron
@mental nexus @onyx hinge I also don't keep my master/main branch up to date. I just work off adafruit/main directly
mental nexus
@onyx hinge please get the latest version of label.py and report back.
slender iron
ESP32-S2 technical reference has USB and I2S sections now
onyx hinge
@mental nexus with the version in the latest bundle it also happens
the text seems to migrate left over time
idle wharf
ESP32-S2 technical reference has USB and I2S sections now
@slender iron Is that in the Espressif IDF docs or in an AdaFruit doc ?
onyx hinge
reestablishing the anchored_point and anchored_position each time the text is changed seems to help or fix it
@mental nexus I'll file a bug report about this after lunch, unless you think I should check with the absolute newest label, not just the bundle (august 11) version
slender iron
mental nexus
@onyx hinge I think the last PR merge was the same day so I assume you have the latest revision. Please make an issue and I will take a look.
lone axle
Interesting we definitely did have a bug like that at one point but I did think it was taken care of with one of the recent PRs. If you are seeing that in the latest one from the bundle then we will want to take another look and see if either we re-introduced it or missed one of the cases where it occurs back when we fixed it.
mental nexus
@slender iron Ouch I didnโt realize it added all those files to the commit. I was being careful to do git add file name but I think something must have gone way wrong during the merge from main. What are your suggestions on how to resolve?
slender iron
looks at the git commit history
I've never seen the 2 file thing
I'd probably do an interactive rebase. It should ignore the merge commit
I'd also suggest ignoring your own main/master
it just adds an extra step and no benefit
mental nexus
In order to create the circuitpython.pot you said
You'll need to merge the latest changes from main and probably also run make translate to get your new warning strings added to circuitpython.pot.
But maybe I also pulled in some old stuff from my master/main?
slender iron
ya, merging tends to be easier
I honestly have no idea where the 2 came from
the submodules I've seen before because they usually need to be updated separately
mental nexus
So I want to somehow grab the "adafruit/circuitpython/main" and merge in my three files and call it the same branch bitmap_v2
slender iron
mind if I walk you through it? start with git remote -v so I know what your remotes are named
mental nexus
iMac-86:circuitpython margaret$ git remote -v
origin https://github.com/kmatch98/circuitpython.git (fetch)
origin https://github.com/kmatch98/circuitpython.git (push)
source https://github.com/adafruit/circuitpython.git (fetch)
source https://github.com/adafruit/circuitpython.git (push)
upstream https://github.com/kmatch98/circuitpython.git (fetch)
upstream https://github.com/kmatch98/circuitpython.git (push)
Done.
mental nexus
iMac-86:circuitpython margaret$ git branch
bitmap_copy
bitmap_insert
bitmap_trial
* bitmap_v2
bitmap_write
main
restart
(or we can rename source to adafruit which I think is clearer)
mental nexus
iMac-86:circuitpython margaret$ git rebase -i source/main
Auto-merging shared-bindings/displayio/Bitmap.c
CONFLICT (content): Merge conflict in shared-bindings/displayio/Bitmap.c
error: could not apply 1dd4e51a7... Added bitmap.blit function for copying slices of bitmaps
Resolve all conflicts manually, mark them as resolved with
"git add/rm <conflicted_files>", then run "git rebase --continue".
You can instead skip this commit: run "git rebase --skip".
To abort and get back to the state before "git rebase", run "git rebase --abort".
Could not apply 1dd4e51a7... Added bitmap.blit function for copying slices of bitmaps
A lot of files listed. Here's at the top:
slender iron
so the idea with rebase is that it's basically redoing your edits based on the latest version
mental nexus
iMac-86:circuitpython margaret$ git status
interactive rebase in progress; onto 98469322b
Last command done (1 command done):
pick 1dd4e51a7 Added bitmap.blit function for copying slices of bitmaps
Next commands to do (7 remaining commands):
pick 14294bfbe Added bitmap.blit function for bitmap slice copy
pick 5581f25ec Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
(use "git rebase --edit-todo" to view and edit)
You are currently rebasing branch 'bitmap_v2' on '98469322b'.
(fix conflicts and then run "git rebase --continue")
(use "git rebase --skip" to skip this patch)
(use "git rebase --abort" to check out the original branch)
Changes to be committed:
(use "git restore --staged <file>..." to unstage)
modified: shared-bindings/displayio/Bitmap.h
modified: shared-module/displayio/Bitmap.c
Unmerged paths:
(use "git restore --staged <file>..." to unstage)
(use "git add <file>..." to mark resolution)
both modified: shared-bindings/displayio/Bitmap.c
Changes not staged for commit:
(use "git add <file>..." to update what will be committed)
(use "git restore <file>..." to discard changes in working directory)
modified: extmod/ulab (new commits)
modified: frozen/Adafruit_CircuitPython_BLE (new commits)
modified: frozen/Adafruit_CircuitPython_BLE_Apple_Notification_Center (new commits)
modified: frozen/Adafruit_CircuitPython_BusDevice (new commits)
modified: frozen/Adafruit_CircuitPython_CircuitPlayground (new commits)
modified: frozen/Adafruit_CircuitPython_Crickit (new commits)
modified: frozen/Adafruit_CircuitPython_DRV2605 (new commits)
modified: frozen/Adafruit_CircuitPython_DS3231 (new commits)
a long list of "modified" files
and some "untracked" files
mental nexus
The untracked files all have a " 2" at the end.
slender iron
we can leave those untracked
mental nexus
Ok, it did a lot of churning and is done.
slender iron
ok, how does git status look?
mental nexus
iMac-86:circuitpython margaret$ git status
interactive rebase in progress; onto 98469322b
Last command done (1 command done):
pick 1dd4e51a7 Added bitmap.blit function for copying slices of bitmaps
Next commands to do (7 remaining commands):
pick 14294bfbe Added bitmap.blit function for bitmap slice copy
pick 5581f25ec Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
(use "git rebase --edit-todo" to view and edit)
You are currently rebasing branch 'bitmap_v2' on '98469322b'.
(fix conflicts and then run "git rebase --continue")
(use "git rebase --skip" to skip this patch)
(use "git rebase --abort" to check out the original branch)
Changes to be committed:
(use "git restore --staged <file>..." to unstage)
modified: shared-bindings/displayio/Bitmap.h
modified: shared-module/displayio/Bitmap.c
Unmerged paths:
(use "git restore --staged <file>..." to unstage)
(use "git add <file>..." to mark resolution)
both modified: shared-bindings/displayio/Bitmap.c
Untracked files:
(use "git add <file>..." to include in what will be committed)
ports/atmel-samd/asf4_conf/same54/hpl_dac_config 2.h
ports/atmel-samd/asf4_conf/same54/hpl_gclk_config 2.h
ports/atmel-samd/asf4_conf/same54/hpl_mclk_config 2.h
ports/atmel-samd/asf4_conf/same54/hpl_nvmctrl_config 2.h
kk, so we can ignore untracked. open shared-bindings/displayio/Bitmap.c and find the >>>> marks
that's where you'll need to hand merge your changes with the newer code from main
mental nexus
Ok, I'll look into that file.
Hmm.. This looks like the really old version of the file.
slender iron
yeah, it'll be the old version of your changes because this is just the first commit
that's totally ok
mental nexus
Ok, so just resolve any of these >>> lines?
slender iron
Last command done (1 command done):
pick 1dd4e51a7 Added bitmap.blit function for copying slices of bitmaps
Next commands to do (7 remaining commands):
yup
once this is merged we'll continue to apply your commits to it
the first tends to be the toughest
mental nexus
Got it. I made the updates. Here is status:
iMac-86:circuitpython margaret$ git status
interactive rebase in progress; onto 98469322b
Last command done (1 command done):
pick 1dd4e51a7 Added bitmap.blit function for copying slices of bitmaps
Next commands to do (7 remaining commands):
pick 14294bfbe Added bitmap.blit function for bitmap slice copy
pick 5581f25ec Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
(use "git rebase --edit-todo" to view and edit)
You are currently rebasing branch 'bitmap_v2' on '98469322b'.
(fix conflicts and then run "git rebase --continue")
(use "git rebase --skip" to skip this patch)
(use "git rebase --abort" to check out the original branch)
Changes to be committed:
(use "git restore --staged <file>..." to unstage)
modified: shared-bindings/displayio/Bitmap.h
modified: shared-module/displayio/Bitmap.c
Unmerged paths:
(use "git restore --staged <file>..." to unstage)
(use "git add <file>..." to mark resolution)
both modified: shared-bindings/displayio/Bitmap.c
Untracked files:
(use "git add <file>..." to include in what will be committed)
k, now git add shared-bindings/displayio/Bitmap.c and then git rebase --continue
mental nexus
iMac-86:circuitpython margaret$ git rebase --continue
[detached HEAD 4ba9ff892] Added bitmap.blit function for copying slices of bitmaps
Author: Margaret Matocha <ksmatocha@gmail.com>
3 files changed, 94 insertions(+), 1 deletion(-)
error: Your local changes to the following files would be overwritten by merge:
shared-module/displayio/Bitmap.c
Please commit your changes or stash them before you merge.
Aborting
hint: Could not execute the todo command
hint:
hint: pick 5581f25ec4a9b03a19adcab88ecdec29d13dd94c Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
hint:
hint: It has been rescheduled; To edit the command before continuing, please
hint: edit the todo list first:
hint:
hint: git rebase --edit-todo
hint: git rebase --continue
Could not apply 5581f25ec... Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
what does git status show?
mental nexus
iMac-86:circuitpython margaret$ git status
interactive rebase in progress; onto 98469322b
Last commands done (4 commands done):
pick 5581f25ec Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
pick 14294bfbe Added bitmap.blit function for bitmap slice copy
(see more in file .git/rebase-merge/done)
Next commands to do (6 remaining commands):
pick 5581f25ec Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
pick 14f5d03b6 bringing up to date
(use "git rebase --edit-todo" to view and edit)
You are currently editing a commit while rebasing branch 'bitmap_v2' on '98469322b'.
(use "git commit --amend" to amend the current commit)
(use "git rebase --continue" once you are satisfied with your changes)
Untracked files:
(use "git add <file>..." to include in what will be committed)
ports/atmel-samd/asf4_conf/same54/hpl_dac_config 2.h
ports/atmel-samd/asf4_conf/same54/hpl_gclk_config 2.h
ports/atmel-samd/asf4_conf/same54/hpl_mclk_config 2.h
ports/atmel-samd/asf4_conf/same54/hpl_nvmctrl_config 2.h
ports/atmel-samd/asf4_conf/same54/hpl_osc32kctrl_config 2.h
weird that git status doesn't show shared-module/displayio/Bitmap.c as edited
mental nexus
Should I have done a commit after the add above?
slender iron
no, the rebase should do that for you
because it's updating redoing an existing commit
try git rebase --continue again
mental nexus
iMac-86:circuitpython margaret$ git status
interactive rebase in progress; onto 98469322b
Last commands done (5 commands done):
pick 14294bfbe Added bitmap.blit function for bitmap slice copy
pick 5581f25ec Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
(see more in file .git/rebase-merge/done)
Next commands to do (5 remaining commands):
pick 14f5d03b6 bringing up to date
pick 3d26ab150 Updated formatting of stubs for Python documentation
(use "git rebase --edit-todo" to view and edit)
You are currently rebasing branch 'bitmap_v2' on '98469322b'.
(fix conflicts and then run "git rebase --continue")
(use "git rebase --skip" to skip this patch)
(use "git rebase --abort" to check out the original branch)
Changes to be committed:
(use "git restore --staged <file>..." to unstage)
modified: shared-bindings/displayio/Bitmap.c
modified: shared-module/displayio/Bitmap.c
Unmerged paths:
(use "git restore --staged <file>..." to unstage)
(use "git add <file>..." to mark resolution)
both modified: locale/circuitpython.pot
Untracked files:
(use "git add <file>..." to include in what will be committed)
ports/atmel-samd/asf4_conf/same54/hpl_dac_config 2.h
ports/atmel-samd/asf4_conf/same54/hpl_gclk_config 2.h
ports/atmel-samd/asf4_conf/same54/hpl_mclk_config 2.h
ok, so locale/circuitpython.pot needs merging
mental nexus
Ok, do I "git add locale/circuitpython.pot"?
mental nexus
Oh, I'll look for >>>>'s.
There's 23 of them, does it matter how I resolve them?
I'm not sure what I'm lookint at.
slender iron
you can also do git checkout locale/circuitpython.pot and then make translate
the checkout will replace the merge conflict version with the one from main and then the make translate will re-include the strings you added
mental nexus
Ok, will do that.
slender iron
you'd only want to manually merge if you added translations
mental nexus
iMac-86:circuitpython margaret$ git checkout locale/circuitpython.pot
error: path 'locale/circuitpython.pot' is unmerged
Should I go ahead with make translate?
mental nexus
iMac-86:circuitpython margaret$ git checkout locale/circuitpython.pot
error: path 'locale/circuitpython.pot' is unmerged
iMac-86:circuitpython margaret$ make tranlate
make: *** No rule to make target `tranlate'. Stop.
iMac-86:circuitpython margaret$ make translate
find extmod lib main.c ports/atmel-samd ports/cxd56 ports/mimxrt10xx ports/nrf ports/stm py shared-bindings shared-module supervisor -iname "*.c" -print | (LC_ALL=C sort) | xgettext -f- -L C -s --add-location=file --keyword=translate -o circuitpython.pot -p locale
lib/tinyusb/lib/FreeRTOS/FreeRTOS/Demo/lwIP_MCF5235_GCC/lwip/src/core/ipv6/ip6_addr.c:79: warning: unterminated string literal
iMac-86:circuitpython margaret$ git status
interactive rebase in progress; onto 98469322b
Last commands done (5 commands done):
pick 14294bfbe Added bitmap.blit function for bitmap slice copy
pick 5581f25ec Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
(see more in file .git/rebase-merge/done)
Next commands to do (5 remaining commands):
pick 14f5d03b6 bringing up to date
pick 3d26ab150 Updated formatting of stubs for Python documentation
(use "git rebase --edit-todo" to view and edit)
You are currently rebasing branch 'bitmap_v2' on '98469322b'.
(fix conflicts and then run "git rebase --continue")
(use "git rebase --skip" to skip this patch)
(use "git rebase --abort" to check out the original branch)
Changes to be committed:
(use "git restore --staged <file>..." to unstage)
modified: shared-bindings/displayio/Bitmap.c
modified: shared-module/displayio/Bitmap.c
Unmerged paths:
(use "git restore --staged <file>..." to unstage)
(use "git add <file>..." to mark resolution)
both modified: locale/circuitpython.pot
ok, are the >>> out of locale/circuitpython.pot now?
if so, you can add it and continue the rebase
mental nexus
Yep >>>> are gone.
slender iron
k, add and continue away
mental nexus
iMac-86:circuitpython margaret$ git rebase --continue
locale/circuitpython.pot: needs merge
You must edit all merge conflicts and then
mark them as resolved using git add
Hm.. let me search again.
slender iron
did you add the file? that tells git you merged it
git add locale/circuitpython.pot
mental nexus
Here's my status.```
iMac-86:circuitpython margaret$ git branch
warning: ignoring ref with broken name refs/remotes/origin/bitmap_insert 2
- (no branch, rebasing bitmap_v2)
bitmap_copy
bitmap_insert
bitmap_trial
bitmap_v2
bitmap_write
main
restart
iMac-86:circuitpython margaret$
that branch status is ok since you are in the middle of a rebase
mental nexus
ok, did that and rebased again:
iMac-86:circuitpython margaret$ git rebase --continue
[detached HEAD a66ef32da] Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
3 files changed, 58 insertions(+), 31 deletions(-)
Successfully rebased and updated refs/heads/bitmap_v2.
great! now build circuitpython to make sure it's all ok
it should be up-to-date now
mental nexus
Ok, will build and check the code.
slender iron
๐
mental nexus
I think the code is correct. git status shows this:
iMac-86:nrf margaret$ git status
On branch bitmap_v2
Your branch and 'origin/bitmap_v2' have diverged,
and have 12 and 9 different commits each, respectively.
(use "git pull" to merge the remote branch into yours)
Changes not staged for commit:
(use "git add <file>..." to update what will be committed)
(use "git restore <file>..." to discard changes in working directory)
modified: ../../extmod/ulab (new commits)
modified: ../../frozen/Adafruit_CircuitPython_BLE (new commits)
ok great! now git push upstream bitmap_v2 --force-with-lease
mental nexus
iMac-86:nrf margaret$ git push upstream bitmap_v2 --force-with-lease
To https://github.com/kmatch98/circuitpython.git
! [rejected] bitmap_v2 -> bitmap_v2 (stale info)
error: failed to push some refs to 'https://github.com/kmatch98/circuitpython.git'
ok
mental nexus
Should I git add the bitmap files?
slender iron
no, everything looks added
the two changes not staged are submodules to update
do you push to upstream or origin usually?
mental nexus
Usually just say git push
slender iron
ok, then try git push --force-with-lease
mental nexus
iMac-86:nrf margaret$ git push --force-with-lease
Enumerating objects: 101, done.
Counting objects: 100% (101/101), done.
Delta compression using up to 4 threads
Compressing objects: 100% (74/74), done.
Writing objects: 100% (74/74), 12.28 KiB | 838.00 KiB/s, done.
Total 74 (delta 56), reused 0 (delta 0), pack-reused 0
remote: Resolving deltas: 100% (56/56), completed with 18 local objects.
To https://github.com/kmatch98/circuitpython.git
+ 9e73de381...b89005610 bitmap_v2 -> bitmap_v2 (forced update)
@mental nexus so commit 1f21 "bringing up to date" looks incorrect
mental nexus
That was my feedble attempt at merging in the adafruit/main
slender iron
so you'll need to do the rebase -i again but delete the pick 1f21 line in the editor it pops up
that will drop the commit from the new version
mental nexus
This is what it says:
noop
# Rebase b89005610..b89005610 onto b89005610 (1 command)
#
# Commands:
# p, pick <commit> = use commit
# r, reword <commit> = use commit, but edit the commit message
hrm, ok. that makes sense since you are still based on it
thorny jay
@idle owl Somebody made a PR on Pimoroni EnviroPlus Feather Wing that push sensor values to AdafruitIO. It involve an M4, the AirLift and the Enviro+. (It does not work with the PM25 because of an incompatibility on the Pin use... to have Enviro+ + Wifi we will need to wait for ESP32S2 Feather). I was wondering if that was interesting for a learn guide, as this might be a too complex example a library example. The idea would to finish the integration of Pimoroni Driver in the Community Bundle, then adapt that PR, the write a learn guide. Can I propose that to the author, or could I do the learn guide for the author (I wanted to learn about that). Here is the PR: https://github.com/pimoroni/EnviroPlus-FeatherWing/pull/13
mental nexus
@onyx hinge @lone axle I confirmed that label.py changing the text makes it move around strangely. Will look into it.
thorny jay
@idle owl The goal would be to celebrate the new drivers for that board. And since it is available from Adafruit and the example require two other Feather and an adaptor, it make a lot of sense.
mental nexus
How do I get out of the editor without messing somehting up?
slender iron
you can save, it won't do anything
onyx hinge
@mental nexus thanks, would you still like me to file an issue?
mental nexus
I'll talk with @lone axle and decide who will work on it.
Ok, then I'll delete the pick 1f21..... line
slender iron
yup
and the rebase should finish without interaction after
since you did the merges already
mental nexus
iMac-86:nrf margaret$ git rebase -i HEAD~8
Auto-merging shared-bindings/displayio/Bitmap.c
CONFLICT (content): Merge conflict in shared-bindings/displayio/Bitmap.c
error: could not apply dda9f2446... Deleted trailing whitespace
Resolve all conflicts manually, mark them as resolved with
"git add/rm <conflicted_files>", then run "git rebase --continue".
You can instead skip this commit: run "git rebase --skip".
To abort and get back to the state before "git rebase", run "git rebase --abort".
Could not apply dda9f2446... Deleted trailing whitespace
@idle owl also require Adafruit IO...
slender iron
I jinxed myself
mental nexus
iMac-86:nrf margaret$ git status
interactive rebase in progress; onto 98469322b
Last commands done (4 commands done):
pick a66ef32da Added inclusive indexing for x2,y2, fixed default value handling for x1,y1, added bitmap palette comparison
pick dda9f2446 Deleted trailing whitespace
(see more in file .git/rebase-merge/done)
Next commands to do (2 remaining commands):
pick 2b8f63da1 deleted whitespace
pick b89005610 deleted whitespace
(use "git rebase --edit-todo" to view and edit)
You are currently rebasing branch 'bitmap_v2' on '98469322b'.
(fix conflicts and then run "git rebase --continue")
(use "git rebase --skip" to skip this patch)
(use "git rebase --abort" to check out the original branch)
Changes to be committed:
(use "git restore --staged <file>..." to unstage)
modified: ../../shared-bindings/displayio/Bitmap.h
modified: ../../shared-module/displayio/Bitmap.c
Unmerged paths:
(use "git restore --staged <file>..." to unstage)
(use "git add <file>..." to mark resolution)
both modified: ../../shared-bindings/displayio/Bitmap.c
@onyx hinge if you have some basic code that illustrates it go ahead and make an issue. If it's deep into something and not easy to extract then I'd say don't worry about it.
I'll be around tonight and I can look into it for sure. If it doesn't end up getting resolved I can make an issue with what I've found.
slender iron
@mental nexus go ahead and edit that file to fix the >>> areas and then add and continue
mental nexus
Got it.