#circuitpython-dev
manic glacierBOT
timber mango
Hello guys, I drive LED strip via the spi interface on FT232H from laptap. Led strip is lighted but it performs not as expected. I used the sample of Neopixel ring, where shows the basic usage of neopixelspi lib. Where can I get more information to use neopixelspi lib?
gilded cradle
@timber mango, assuming you've already seen https://learn.adafruit.com/circuitpython-neopixels-using-spi and https://learn.adafruit.com/circuitpython-on-any-computer-with-ft232h, you could look at the library code at https://github.com/adafruit/Adafruit_CircuitPython_NeoPixel_SPI.
tulip sleet
@onyx hinge something weird is going on; preprocess_frozen_modules is removing example dirs on 5.x but not 4.x
onyx hinge
@tulip sleet it's about subdirectories of examples. I'll double check what in seeing here. The background is that @idle owl 's adafruit_circuitplayground was freezing in advanced examples when I built with her work in progress version, making flash really not fit
tulip sleet
oh, it's the 'advanced examples', ok I see what you're doing
idle owl
Yeah, the extra directory that's new since the last freeze.
onyx hinge
(not at my computer right now)
tulip sleet
ok, I was confused: i thought it was all example dirs, which were working right. We didn't have subdirs before, ok, got it.
I misinterpreted "subdirs of examples..." as meaning they were the subdirs
since the variable is called subdirs π
tulip sleet
@onyx hinge are you off for the rest of the day?
onyx hinge
@tulip sleet I'm back around my computer now
ionic elk
I'm finding that displayio.release_displays() doesn't release the SCK pin of my screen. Is there another step I need to take?
onyx hinge
hmph the audio+display dma lockup went into hiding as soon as I wrote a program just to try to provoke it π
onyx hinge
I must be missing an ingredient. DMA from the SD card too, maybe?
maybe a rawsample doesn't provoke it at all, for some reason?
tulip sleet
@onyx hinge did you get the error originally with the latest SD card library? There were a bunch of fixes for larger reads and writes and support of certain cards. You could try varying the file size (should be at least several kB)
idle owl
@onyx hinge Ran into something that might be audio related. There's an example that plays tones on the Circuit Playground higher and lower frequency as you tilt the board. It runs fine on CPX, but on CPB, if I tilt it too far one way, I get this: Traceback (most recent call last): File "code.py", line 132, in <module> File "code.py", line 96, in run File "code.py", line 129, in change_tone_if_needed File "adafruit_circuitplayground/circuit_playground_base.py", line 922, in start_tone ValueError: Sample rate too high. It must be less than 62500
Is this an nRF issue? Or something I'm unaware of?
onyx hinge
@tulip sleet yes I have the latest as of a week or so ago. the one earlier than that had many more problems.
@idle owl hm that requirement can potentially be changed, I didn't know about that demo...! (there are lots of demos I know nothing of)
idle owl
@tulip sleet Testing on Bluefruit and finding issues with the examples, unrelated to the changes I made. I'm thinking I'll file issues for the bugs and deal with them in a different PR.
@onyx hinge It's in advanced examples.
onyx hinge
@idle owl can you please enter an issue about it? There's not a quick-quick fix but there might be a slightly less quick fix that I'll look into tomorrow.
idle owl
Yes, I already have it started, I will tag you on it.
onyx hinge
thanks, you can assign it to me, I find it helpful
any idea how high sample rate will go in that example?
tulip sleet
the library could keep the sample rate below the max
idle owl
@onyx hinge No idea.
onyx hinge
at the time I wrote it, I was unaware of how this is used in CP as a way to get sine waves of various frequencies. I figured "62500 is higher than any sensible wave file would go, because of the human auditory system" and so I made some other assumptions related to that.
not sure that clipping the playback at 62.5kHz will be satisfactory, in the context of this demo
tulip sleet
https://github.com/adafruit/Adafruit_CircuitPython_CircuitPlayground/blob/f042c596105d1cfd3a1a2ca2c6035b188fc5f6e6/adafruit_circuitplayground/circuit_playground_base.py#L913 already does some limiting, so we could adjust this in the calling code.
I think it makes sense to change the library, not CPy. A sampling rate that high is not necessary.
onyx hinge
yes, you could cap frequency at 22kHz or so, and if frequency > 5000Hz, set length to 20 instead of 100 or so
CPB audio will get worse faster than CPX audio when you request a high sample rate, because of the limitations of CPB PWM
idle owl
@tulip sleet I notice on CPB we don't have whatever we have in place on CPX to clear the NeoPixels when the code is not running, or whatever it is that it does that makes the LEDs turn off if you load new code that doesn't explicitly clear them. Worth implementing?
Ok found bugs, filed issues, PRing update.
tulip sleet
@idle owl yes, some other people noticed that. I'll file an issue.
idle owl
Ok
idle owl
@tulip sleet Hah I thought your response to that issue was a response to my PR on which I had literally just hit create. I was wondering how you had ninja'd it that fast. I was wrong π
tulip sleet
i type very fast
ionic elk
@slender iron So, uhh, just determined that the Meowbit shares a SPI bus between the TFT display and the SPI flash. How obnoxious would you figure that'll end up being?
Is that something I can hack around, or would it require major refactoring? The flash reserves the pins... does displayio rely on flash access for stuff?
tulip sleet
@ionic elk if it's easier, you could just use SPI instead of QSPI? It that the issue?
If you look at the Meowbit MicroPython impl, what do they do?
ionic elk
@tulip sleet no QSPI. The physical chips are on the same line
tulip sleet
as long as there are separate CS lines (there have to be, I assume), I don't see that it's a big problem, but the SPI filesystem code has to share the pins
ionic elk
Yeah that's just kind of my question. My current problem is just that there's no interface for having the SPI filesystem allow other modules to access the pins it uses. But I before I started hacking in that direction I figured I'd get a second opinion on whether there'll be even bigger, worse problems down the line
@tulip sleet on reconsideration: with the existing system, is there any actual way to have two modules use an SPI bus mediated by CS pins?
I guess the usual case would be that two modules would use a SPI instance you've defined, but that doesn't apply to SPI flash since it defines its own instance. So we need some way of fetching the SPI object being used by flash, so it can then be plugged into other things
slender iron
@ionic elk is there a designated SPI bus that it exposes for the user to use?
you could make this globally available: https://github.com/adafruit/circuitpython/blob/master/supervisor/shared/external_flash/spi_flash.c#L39
ionic elk
Yeah that's actually exactly what I was looking at
But this is where my experience starts to get shaky - how would I go about exposing that in the python interface?
I also need the CS pin, of course
something like spi = flash.getSPI()?
slender iron
why do you need the CS pin too?
to expose it to Python, add it to the board level pin table like SPI (board.SPI)
I'd call it INTERNAL_SPI or something
you can also init the display in the board def like pybadge so its automatically used
ionic elk
@slender iron you're right, I guess I don't use it since it'd be entirely internally managed. Not like I need to go in and disable it.
slender iron
π
ionic elk
I'll look at how pybadge does it. Will I need to add the internal SPI to the shared-bindings/board/__init__ and import it into the flash implementation or is there a simpler way I don't know of to get it into board/pins.c?
slender iron
I think you can just declare it in supervisor/shared/external_flash/spi_flash.h (add this file) and use it directly
ionic elk
So I'd include that file directly in my pins.c file, in addition to the shared bindings file, and then expose it in the global array? Sorry for all the questions, trying to make sure I'm doing this in the right way
ionic elk
good news is my control test with internal filesys is working, so that's noice
I'll pop up a PR now, and then see if I can get this extern SPI version in.
simple pulsar
@idle owl Is the issue you found different to https://github.com/adafruit/circuitpython/issues/2341 ?
tulip sleet
@idle owl I have the 4.x PR most ly donefor increased stack size, but will wait until I can update the frozen libs with the new release.
@simple pulsar yes, that's it, oops, removed my dupe issue
onyx hinge
okay, audio + display crashed out at about 2900s runtime, oops, I didn't have the debugger plugged in argfh
idle owl
@tulip sleet Ok releasing now if I can get it done before dinner is ready.
slender iron
@onyx hinge I know how that feels. I crashed my mac again yesterday by leaving a CIRCUITPY drive plugged in so I gave up and finally started debugging it with a beagle
idle owl
@tulip sleet Ok released.
onyx hinge
yeah I have lockups I'm blaming on inadequate USB robustness in Linux as well
idle owl
bbl. Dinner!
onyx hinge
not quite sure how to separate the debugging, building, and serial-terminaling though I have a pi4 sitting right here that could help
and the next lockup happened real quick. that's something
onyx hinge
OK my stack is the same as in https://github.com/adafruit/circuitpython/issues/2005
slender iron
I generally use a second computer as a usb sacrifice in that case
and build locally and load over jlink
tame rover
Anyone know if the Adafruit Edge Badge is the Same as the Adafruit PyBadge?
tulip sleet
EdgeBadge has a PDM microphone on board; PyBadge does not
slender iron
@gilded cradle are you working on hidapi ci issues? It's preventing me from fixing up new repos without hacking the deps
gilded cradle
@slender iron, I'm waiting on this PR to get approved: https://github.com/adafruit/Adafruit_Blinka/pull/204
gilded cradle
Awesome
slender iron
π thanks!
gilded cradle
yw, thanks for looking at it.
slender iron
np
gilded cradle
Ok, it's still doing it. I'm going to remove that requirement altogether.
@slender iron, can you take a look at https://github.com/adafruit/Adafruit_Blinka/pull/205 once it passes? I just removed it altogether so we can get some work done.
slender iron
@gilded cradle approved
gilded cradle
Thanks
slender iron
np, happy to give it the stamp of approval
gilded cradle
It should probably take around 5 minutes to get on PyPI and for travis to realize it's there.
slender iron
np, I'm multitasking
gilded cradle
I'm watching Travis
slender iron
π
slender iron
got past it. thanks @gilded cradle
gilded cradle
You're welcome.
slender iron
raven canopy
sees multiple favorite keywords: sphinx. auto. api. π±οΈ π±οΈ π±οΈ π
slender iron
π
what is the right way to add travis api tokens for new libraries? Or is there a way to convert to actions? cookiecutter is still travis
raven canopy
Settings->Webhooks, right?
oh wait...
i haven't done it in sooo long. lol
@slender iron https://github.com/adafruit/Adafruit_CircuitPython_TestRepo/tree/master/.github/workflows has the Actions siddacious and i worked out.
onyx hinge
@slender iron I'm feeling stuck about #2005, a hang in SPI DMA while audio is playing. It's impeding progress on JEplayer which is supposed to be my priority. can you help or point me at someone who can? I may be in and out but I'll read back.
slender iron
@onyx hinge we can chat about it in 15 minutes or so. I'm not sure what the best approach to debugging it is
onyx hinge
okie, I'd appreciate that. I'll be here.
slender iron
kk, just finishing up breakfast π
slender iron
@onyx hinge video or audio? ready now
prime flower
@solar whale I'm rebuilding nina-fw on 3.3.1 today, should I tag you in a PR for testing?
solar whale
@prime flower that would be fine with me
solar whale
@prime flower it may be this evening or possibly tomorrow before I can test the wifi-nina. Is that OK?
crimson ferry
@prime flower I noticed I put the wrong Code of Conduct links in the README.md (should be https://github.com/adafruit/nina-fw/blob/master/code-of-conduct.md. Is that something you could tweak while you're in there?
prime flower
@crimson ferry I could tweak it, which links are you referring to
@solar whale I'm not in a rush, IDF 3.3.0 is very stable, 3.3.1 mainly does some bugfixes/security patches
tulip sleet
@slender iron @idle owl oh groan, "4.1.1" is really a release of master π¦
because I didn't tag the commit before I released.
tulip sleet
@slender iron or @idle owl could you approve https://github.com/adafruit/circuitpython-org/pull/341 quickly so I can kill the bad release
ionic elk
@slender iron so, the internal SPI intercept "works"... but it seems super, probably unacceptably easy to brick the device. If my python code has any kind of bug with the display SPI, I lose my filesystem, and also my REPL access (maybe due to an MSC collapse?)
Can't just re-flash, either, since it won't erase internal flash - I actually have to roll back the firmware
orchid basinBOT
ionic elk
@slender iron I think it might be less apparent if the screen wasn't being used by the REPL when the chip is trying to do non-python operations like the soft reboot. Is there any option in DisplayIO to explicitly disable access to a screen by the REPL and keep it from being on the never-reset list?
idle owl
@tulip sleet Approved.
tulip sleet
tnxtnx
crimson ferry
@prime flower In the NINA https://github.com/adafruit/nina-fw README.md file, the two "Code of Conduct" links in the Contributing to nina-fw section look OK at first glance, but they are 404. The proper link is: https://github.com/adafruit/nina-fw/blob/master/code-of-conduct.md
I may not get to it soon, but I'm planning to work up an example for the ESP32SPI read / write features.
slender iron
@ionic elk we don't currently have a way to turn off the terminal separate from the display work
prime flower
@crimson ferry thanks for the location of the 404'd links, I've added the correct ones to the README in the active PR!
crimson ferry
Cool, thank you!
prime flower
np!
slender iron
@pastel panther does the test repo have the latest github actions files? I need to add them to my new repos
pastel panther
@slender iron I believe so however you should probably use the ones in the cookiecutter as a better reference
slender iron
pastel panther
I could have sworn @raven canopy added them π€
slender iron
he pointed me to the test repo
pastel panther
That works for now
slender iron
kk
raven canopy
@pastel panther your comment from the meeting makes more sense now. i only put in an issue as a reminder...not a PR. :D
https://github.com/adafruit/cookiecutter-adafruit-circuitpython/issues/56
pastel panther
whoopsie doodle
slender iron
I can do a PR to cookiecutter after I get display_notification going
pastel panther
sweet, thanks
slender iron
np
pastel panther
bleh, looks like I had @trim elm using a stale build.yml
pastel panther
@raven canopy can you look at these two and let me know which on is the most recent/update to your recollection?
https://github.com/adafruit/Adafruit_CircuitPython_RGBLED/blob/master/.github/workflows/build.yml
raven canopy
TestRepo. it has the release artifact upload.
pastel panther
thank you
ionic elk
@slender iron Meowbit is opening kind of a can of worms here. It seems like I would actually need to modify the DisplayIO API to allow an option to disable REPL access since some kind of concurrent display/flash write on restart bricks the USB and filesystem.
slender iron
ya, go ahead and move on @ionic elk
that is the first board with a shared spi bus for flash
ionic elk
Should we pick an option and keep the board? Or just toss the whole thing?
Since it also has all the bootloader junk
Display works fine with internal flash, and vice versa
So we could leave that up to the user.
ionic elk
Unfortunately internal flash makes it impossible to co-exist with the uf2 bootloader, since they require the same flash region...
can of worms π
slender iron
ok, then skip it for now
ionic elk
Internal flash, ST Link required, then?
slender iron
do we already have others that require ST Link?
using the ST Link will overwrite the bootloader right?
actions release worked: https://github.com/adafruit/Adafruit_CircuitPython_Display_Notification/releases/tag/0.8.1 π
I think it'd be worth spending a little time moving the internal fs. It'd be good to use uf2 on other boards too
ionic elk
@slender iron I'll look into our options for bootloader relocation (we can't move the filesystem since the ST flash sectors are too huge outside the first 4 sectors). ST-Link flashing will overwrite the default Meowbit bootloader, but maybe parts of it could be moved? It might take some magic.
orchid basinBOT
ionic elk
I'll see whether I can get the ST bootloader working on the Meowbit, since BOOT1 is actually exposed via PB02 on the right button
slender iron
can you make the internal FS smaller and only use the 2-4 sectors?
ionic elk
that'd match it up with others
slender iron
we should have a plan for UF2 on stm though
and it can't live in a big sector unless we leave empty space
ionic elk
Maybe. I can move onto internal filesystem improvements and UF2 uploading now, if that's best
sounds like that's a reasonable next priority.
slender iron
I don't know what you mean by "internal filesystem improvements" but UF2 support would be good
ionic elk
The internal filesystem needs work on the STM32 - it's very slow, and even smaller than it should be - only appears as 32 bits out of 48 available.
slender iron
is that the fat overhead making it smaller?
slow and small are fine since internal fs isn't used much
ionic elk
Possibly, I haven't had the chance to fully examine it. We moved on from it post feather
ionic elk
k, I've got plenty to workon. Will update later
For now though I'm going to clean up the Meowbit PR to just use the internal flash and use the STLink - it won't be far off from boards like the Blackpill in that use case.
I'll put down an issue for returning to it.
onyx hinge
This is interesting, but probably circuitpython/micropython are too different than cpython/pypy to be able to get into one and the same mold.... https://morepypy.blogspot.com/2019/12/hpy-kick-off-sprint-report.html
slender iron
@tulip sleet it ran in my branch ok
tulip sleet
yah, it's just sitting there "There are no checks for this commit"
slender iron
tulip sleet
I could try approving and see if it runs after merging?
slender iron
fine with me
tulip sleet
tulip sleet
@slender iron tip of master is complaining nxp in tinyusb is not a git repo, even though it's in .gitmodules for `tinyusb
slender iron
I'm looking too
quick gale
Hi, I have a pyportal/circuitpython question. The file system on my board is corrupted and read only. I'm trying the directions in https://learn.adafruit.com/adafruit-pyportal/troubleshooting and trying to erase the file system from the serial console using MU, but nothing's happening.
tulip sleet
you did import storage; storage.erase_filesystem() ?
slender iron
@tulip sleet master builds for me, commit a3559f14e
tulip sleet
@slender iron i think it's broken for me because I did git submodule update --init --recursive. I know you are not recommending --recursive, but it shouldn't cause it to break completely.
quick gale
@tulip sleet, yes, but no response from the REPL or the device
slender iron
@quick gale did you unplug it and replug it in? or reopen the serial tab
tulip sleet
it causes a reset. Are you typing it at the >>> ? Could you copy/paste from the repl to here?
slender iron
@tulip sleet I think that can happen with folders being deleted
quick gale
I'm not getting the >>> prompt.
tulip sleet
can you type ctrl-c?
to interrupt the running program?
is there anything in the REPL subwindow?
quick gale
nothing in the subwindow.
quick gale
code.py output:
OSError: [Errno 5] Input/output error
Code done running. Waiting for reload.
Circuitpython 4.1.2 for pyportal
oh
I'm on Ubuntu
19.04
tulip sleet
@slender iron magic light was running, then it failed after a few minutes:
>>> import ble_magic_light_simpletest
Scanning for Magic Light
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "ble_magic_light_simpletest.py", line 50, in <module>
File "adafruit_ble_magic_light.py", line 73, in __setitem__
File "adafruit_ble/characteristics/__init__.py", line 147, in __set__
_bleio.BluetoothError: Unknown soft device error: 3002
never mind, I unscrewed the bulb, but maybe it should give a better error
quick gale
I can try recovering on a Mac when I get home.
tulip sleet
ubuntu is fine; i am using it right now, all the time
@quick gale are you surmising that the Input/outpu error is due to corrupted filesystem or is ubuntu reporting it's corrupted?
quick gale
Ubuntu's reporting it.
tulip sleet
is that the entire thing in the serial window?
quick gale
and when I try to load code.py in MU I get "could not load /media/user/CIRCUITPY/code.py"
tulip sleet
@quick gale Here is a special dangerous CircuitPython that will erase and reformat the filesystem on every restart. Load this .uf2, make sure things are working, and then load the regular one
on Linux, when you copy things to CIRCUITPY from the command line, it's a good idea to use the sync command after you're done copying everything. That will force everything to the CIRCUITPY filesystem immediately.
quick gale
Okay, when I dropped that on the drive, I got a "file system is read only" error message in Ubuntu
ah
tulip sleet
did you double-click first to get the BOOT drive?
you load .uf2 files onto the BOOT drive, not the CIRCUITPY drive
tulip sleet
the Mu editor is good writing immediately to CIRCUITPY, but copying via the command line, sync is a good idea. maybe you copied the 4.1.2 update to CIRCUITPY, and it was really big, and there was an issue??
anyway, great! you can squirrel that DANGER uf2 away in case you get stuck again. But usually you can just do the storage.erase_filessytem() from the REPL
pastel panther
@tulip sleet I got a crash into the HardFault_Handler while testing on a metro m4 with 5.0 Beta 1
@tulip sleet lmk if you want me to do some more poking to narrow down the issue or go ahead and file an issue
welp, I powercycled the metro and it seems to be back to normal
tulip sleet
@pastel panther go ahead and file an issue with the particulars. There have been various fixes since beta.1 but mostly on nrf.
pastel panther
@tulip sleet how many particulars and which ones especially? It's hard to say what exactly caused the issue, especially since it's not doing it anymore
tulip sleet
were you testing some I2C device, writing to the filesystem? etc.
pastel panther
Ok, will do
tulip sleet
were you handling the board at the time, or was it just sitting?
solar whale
@tulip sleet after breaking things with --recursive, I just deleted the lib/tinyusb folder then did git submodule sync;git submodule update --init and all was well
tulip sleet
yes, but, it shouldn't break with recursive, i think
solar whale
something in tinyusb made it very unhappy ....
I read, but forgot the error messages it reported.
tulip sleet
i recloned, and did git submodule update --init --recursive, and now it's happy. I'm not recommending it, necessarily, but I'd like to understand better why it broke. The stackoverflow page may help.
I need to change my alias, and we should change the Building CircuitPython guide, I guess
solar whale
here are the errors I saw ```Entering 'lib/tinyusb/hw/mcu/microchip/samd/asf4'
Entering 'lib/tinyusb/hw/mcu/nordic/nrfx'
Entering 'lib/tinyusb/hw/mcu/nxp'
fatal: not a git repository: /home/jerryneedell/circuitpython_master/lib/tinyusb/hw/mcu/nxp/../../../../../.git/modules/lib/tinyusb/modules/hw/mcu/nxp
Stopping at 'lib/tinyusb/hw/mcu/nxp'; script returned non-zero status.
Stopping at 'lib/tinyusb'; script returned non-zero status.
during the --recursive
tulip sleet
i got the same fatal: not a git repository. It may be that there was a directory that was not a git repo there, and it doesn't like that at all.
solar whale
more above that ```Synchronizing submodule url for 'tools/usb_descriptor'
Submodule 'frozen/Adafruit_CircuitPython_SD' (https://github.com/adafruit/Adafruit_CircuitPython_SD.git) registered for path 'frozen/Adafruit_CircuitPython_SD'
Cloning into '/home/jerryneedell/circuitpython_master/frozen/Adafruit_CircuitPython_SD'...
Submodule path 'frozen/Adafruit_CircuitPython_SD': checked out '5ad33e4ca219f0e216beab439cfa259cde32016c'
warning: unable to rmdir 'hw/mcu/nxp/lpc_driver': Directory not empty
Submodule path 'lib/tinyusb': checked out '7a05b177a43b368fea1d60ca344fe4ae9902a432'
Submodule 'hw/mcu/nxp' (https://github.com/hathach/nxp_driver.git) registered for path 'lib/tinyusb/hw/mcu/nxp'
fatal: not a git repository: /home/jerryneedell/circuitpython_master/lib/tinyusb/hw/mcu/nxp/../../../../../.git/modules/lib/tinyusb/modules/hw/mcu/nxp
Failed to clone 'hw/mcu/nxp'. Retry scheduled
BUG: submodule considered for cloning, doesn't need cloning any more?
Submodule path 'lib/tinyusb/hw/mcu/nordic/nrfx': checked out '7a4c9d946cf1801771fc180acdbf7b878f270093'
fatal: not a git repository: /home/jerryneedell/circuitpython_master/lib/tinyusb/hw/mcu/nxp/../../../../../.git/modules/lib/tinyusb/modules/hw/mcu/nxp
Unable to fetch in submodule path 'lib/tinyusb/hw/mcu/nxp'
Submodule path 'ports/nrf/nrfx': checked out '3f55e49eb11e6db0da1da09e189bb094222702c9'
Failed to recurse into submodule path 'lib/tinyusb'
maybe if you just remove that directory and redo the git update, something might work better, or maybe the merge into your fork got messed up becuase of the presence of the directory
solar whale
it had been working for a long time -- something changed in the last few days
manic glacierBOT
solar whale
bed time here! good night!
tulip sleet
good night!
pastel panther
@tulip sleet I don't think I was handling the board at the time of the crash but I easily could have been
tulip sleet
it's good to have it on record, in case it happens to any one else. thanks!
pastel panther
sure thing
orchid basinBOT
onyx hinge
@slender iron noticed this https://blog.adafruit.com/2019/12/18/circuitpython-4-1-1-released-adafruit-circuitpython/
Translated messages are now compressed more effectively, preventing certain translated builds from being too large. Thanks @onyx hinge.
I don't think this is in 4.1.2
idle owl
Dan π
crimson ferry
@prime flower Will there be a .bin at https://github.com/adafruit/nina-fw/releases ?
prime flower
@crimson ferry I just ran into an issue with travis, re-deploying right now. There should be one in a bit. (~3min if travis doesn't error while deploying again)
I fixed it π οΈ , binary at https://github.com/adafruit/nina-fw/releases/tag/1.6.0
crimson ferry
Ah, nice. Is this the first time for an automated process to create the binary?
I can build my own, but my byte count is always different from the release, due to environment differences (path names, etc), so I like to use the released binary to be sure.
onyx hinge
it would be "fun" to work on repeatable builds for circuitpython ecosystem
prime flower
@crimson ferry nina-fw released with this build system on 1.5.0. The script is similar to IGRR's (Espressif dev), but I do need to manually change the script for each time we build with a new ESP-IDF version
onyx hinge
weird, the tilting arpeggios example doesn't appear in the copy of circuitplayground in frozen/ but it does in a clone / online. maybe circuitpython master branch submodule is behind ? Or maybe I missed a step. I did submodule update..
idle owl
@onyx hinge Does it strip them before or during freezing?
onyx hinge
it ... shouldn't? it should only do that within a copy
anyway I found it now so it's all good
also to my surprise just removing the 62500 Hz check altogether seems to fix it. I can't vouch for whether the notes are right, I might have a proverbial tin ear plus I'm in a coffee shop trying to test this π
bad choices
Submodule path 'frozen/Adafruit_CircuitPython_CircuitPlayground': checked out 'd87ea261c40ecbc6d893d72d337beefbea1cf932'
~/src/circuitpython$ cd frozen/Adafruit_CircuitPython_CircuitPlayground/
~/src/circuitpython/frozen/Adafruit_CircuitPython_CircuitPlayground$ git status
HEAD detached at d87ea26
nothing to commit, working tree clean
~/src/circuitpython/frozen/Adafruit_CircuitPython_CircuitPlayground$ ls examples/advanced_examples/
circuitplayground_acceleration_mapping_neopixels.py circuitplayground_gravity_pulls_pixel.py
this is all the way back at release 2.1.4 of the submodule
onyx hinge
the 62500Hz fix? I didn't PR it yet.
who should PR updating frozen/Adafruit_CircuitPython_CircuitPlayground for 5.x / master ?
nrf pwmaudioout is only in 5.x so I'll PR the fix for master.
idle owl
@onyx hinge I meant the fix to deal with subfolders in the examples folder
onyx hinge
4.x: https://github.com/adafruit/circuitpython/pull/2398 5.x: https://github.com/adafruit/circuitpython/pull/2397 both merged
onyx hinge
I think in 5.x/master the circuitplayground submodule has just not been updated for a long time
onyx hinge
that's not good, because doesn't it mean that if you get the latest build from circuitpython.org for cpx, 'import ....cb' won't work?
idle owl
@onyx hinge I think perhaps it does mean that. Worth pinging @tulip sleet about it.
onyx hinge
@idle owl Just PR'd it instead
idle owl
@onyx hinge Also a valid option.
onyx hinge
ok, besides the dma lockup I'm deliberately NOT looking at today .. what's next?
idle owl
Ooh. Means you're stuck on some other things. What else do you have on your plate?
onyx hinge
oh oh there's something that's been bugging me but I don't think it's filed. paste into repl on nrf seems to drop chars for me. I would like it not to.
idle owl
FIX IT.
orchid basinBOT
onyx hinge
I feel like it's nrf only but not 100% sure and I don't have a samd board to test. Does this sound familiar to anybody else?
(no samd boards here at the coffee shop, that is β )
slender iron
@onyx hinge I thought I fixed the buffering in. I think out isn't buffered right now
onyx hinge
>>> s= ("1234567890123456789012345678901234567890123456789012345678901234567890"
... "123456789012345678901234578901234567890123456789012345678901234567890"
... "12345678901245678901234567890123456789123456789012345678901234567890"
... "1234567890123456789012345678901234567890123456789012345678901234567890"
... "12345678901235678901234567890123456789012345678901245678901234567890"
... "12345678901234567890123456789023456789012345678901234567890123456789"
... "1234567890123467890123567890123456789012345678901234567891234567890"
... "1234567890123456789012345678901234567890123456780123456790123467890"
... "12345679012345678901234568901234567890123456789012345678901234567890")
>>> print(len(s))
615
``` it's sure not happy for me. Should have been 630 characters, all lines the same lengthis there code that should be shared but isn't? since this is tinyusb stuff?
tulip sleet
@onyx hinge I did pull it in : https://github.com/adafruit/circuitpython/commit/6265ee0e8ce0bf79f3cd73be4b15dc631a403a8b
onyx hinge
Thanks @tulip sleet
tulip sleet
it pulled in and worked with no changes
onyx hinge
I love it when that happens
ionic elk
@onyx hinge I was having a very similar issue with STM32 dropping characters a while back. If you figure it out please let me know.
onyx hinge
Will do
dawn rampart
Hi everyone. For those interested I've create a library for the very neat i2c encoder made by Simone Caron: http://www.duppa.net/i2c-encoder-v2-1/ The library works with version 2.1 of the board. The library can be found here: https://github.com/bwshockley/CircuitPython-i2cEncoderLibV21 and I've submitted a pull-request to get it into the CircuitPython Community Bundle.
tulip sleet
@gilded cradle can you say what is broken on circuitpython.org? I'd rather not make a new 4.x release just to fix something unrelated to the actual firmware.
maybe we can fix it by hand?
could we fix it manually, and if we make a new 4.x release later fix it before we do that?
orchid basinBOT
gilded cradle
@tulip sleet I submitted a PR to fix by hand, but it will break again if we do any more 4.1 releases without renaming.
tulip sleet
got it, I understand, so we just need to rename the board, that's all.
manic glacierBOT
orchid basinBOT
meager fog
@onyx hinge did you chat with paintyourdragon about the DMA lockup
onyx hinge
@meager fog not yet. Are they on discord or just GitHub?
onyx hinge
I have a workaround but it's simply to do displayio without dma πΏ
onyx hinge
Let's not, for now. I'll follow up on GitHub tomorrow.
onyx hinge
I also have some suggestions from Scott of things to investigate and try that I haven't done yet
And I saw the workaround used in the Nintendo emulator as well
So I'm sure something will work out
meager fog
yeah
ill ask philb to also suggest anything
but mostly we had workarounds
they were not super pleasant but they worked (somewhat)
onyx hinge
It doesn't match the specifics of that erratum but it's tempting to think it could be related
Do you know, does the one in nofrendo(?) Make the audio glitch or is it not noticible?
Honestly for JEplayer not dma'ing display is adequate, but in general it is not
meager fog
it 'glitches' the video because i set the screen to black
but its rare and...nobody complains
its only noticable if you're really looking
π€· i think the dma subsystem has bugs that nobody else has noticed cause we use the chip more than anyone
ironic cause theres' like 25 channels
but if you use more than 1 it hangs
:/
despite the unpleasantness....if folks are ok with it, could check every 1ms during systick
the check is fast
onyx hinge
@meager fog I hadn't considered checking and kicking in systick.. I'm not sure I know what "the check" is, though
lone sandalBOT
old smelt
Is there any documentation on what changes to make for doing a CP build with a different size flash memory part?
tulip sleet
@old smelt is the chip in supervisor/shared/external_flash/devices.h
old smelt
No, not currently.
I was looking at that file earlier today. I suspect that makes things a bit more complicated?
tulip sleet
ok, that is the hard part, you'll need to figure out the settings by reading the data sheet. If it's a new manufacturer not in there, it could be tricky
old smelt
It's a Giga Device part
tulip sleet
you can try imitating one of those, but compare the datasheet for the existing part carefully with the new one. Is this a bigger size than we've used?
old smelt
Smaller. We have a couple of boards that were built with 2Mbit parts instead of 2MB. I was wondering if there's anything I can still do with it.
tulip sleet
besides that change, you just need to name the chip in boards/<newboard>/mpconfigboard.mk
i am not sure this is worth your time, as opposed to just replacing the chips if you want to salvage the boards
considering the cost of the chips is minimal
old smelt
Whoa... Why was that text red?
tulip sleet
discord having trouble sending
old smelt
Hmm.. ANyway, I agree. And we will replace them. I was just curious if it was possible.
Ah. I've not seen that before.
tulip sleet
def possible
old smelt
Oh wait - you said to change the makefile. I was thinking of the mpconfigboard.h
Yes, saw the references in the .mk file, too.
tulip sleet
EXTERNAL_FLASH_DEVICES = "name of new chip"
old smelt
Ok. Yes, I see that line. Cool. And that references the definition in devices.h?
Won't be much room to do a lot, but I was thinking I could at least get a minimal CP build running.
That'd be something
tulip sleet
yup, that's all you need to change. you can list more than one device there if the board might be manufactured with more than one type. It looks up the ID at run-time
old smelt
Really? That's good to know.
tulip sleet
e.g.
EXTERNAL_FLASH_DEVICES = "W25Q64JV_IQ, GD25Q64C"
old smelt
Ha! You read my mind. I was just thinking, "I wonder if they are comma-separated."
tulip sleet
there are examples already
ivory yew
old smelt
@tulip sleet - the build worked:
Adafruit CircuitPython 5.0.0-alpha.5-26-g01bf93216-dirty on 2019-12-20; Alorium Technology Evo M51 with samd51p19
Thanks for your help!
tulip sleet
@onyx hinge you checked all the latest DMA errata for SAMD51, right?
onyx hinge
@tulip sleet in the current document, 2.10 is errata for DMA. It's not 2.10.1 because the channel being newly enabled is higher than the active channels. It's not 2.10.2 because we use equal priorities. It's not 2.10.3 because that's only to do with debug mode.
tulip sleet
onyx hinge
yes revision J
tulip sleet
do we use linked descriptors?
onyx hinge
I think audio uses linked descriptors
onyx hinge
OK so I'd said that I had what I thought was a quicker lockup that might also be DMA, but it wasn't .. it was this ^^
solar whale
@tulip sleet a question came up about putting CP on a Nano 33 sense -- There is a .uf2 for it, but do you first have to put our uf2 bootloader on it with a JLink?
tulip sleet
yes, and then it breaks using it in Arduino. You need our UF2 bootloader because it includes the SoftDevice too, which CPy needs. I think this is kind of a mess - you have to commit to using the board for one or the other unless you're prepared to keep swapping bootloaders
the Arduino BLE support does not use the SoftDevice
solar whale
OK -- that is what I thought -- I'll pass the info along
tulip sleet
tnx!
onyx hinge
okay the good news is, I captured a DMA hangup and as promised by resetting the stuck channel (manually in gdb) it unstuck
I have a new problem which is that occasionally I am getting "crunchy" audio, but restarting audio playback fixes it
idle owl
Hooray for a new problem!
onyx hinge
(i.e., next/previous track buttons)
idle owl
π software.
onyx hinge
this project keeps me guessing, that's for sure
idle owl
Stay on those toes, Jeff. Stay on those toes.
onyx hinge
it's better than having someone else on my toes
idle owl
valid.
onyx hinge
jeff@river quest I suppose? I don't use slack otherwise.
argh I forget that mentions you-know. hopefully he doesn't notice/mind
meager fog
lol is ok
alright ill send it now
oki done
anywho
so - with the dma nonsense
do you want to dig into it (not most fun - has to happen eventually)
or do you want to not draw while playing audio (workaround)
onyx hinge
if we're going to have icons and navigate them with the joystick, it pretty much needs draw-during-audio
(switching over there ->>>)
slender iron
<@&356864093652516868> any objections to me releasing beta.2 today?
stuck elbow
none
tidal kiln
None
manic glacierBOT
onyx hinge
@slender iron https://github.com/adafruit/circuitpython/pull/2411 maybe? https://github.com/adafruit/circuitpython/pull/2410 would also be nice to have
I don't know of anything newly and egregiously broken
gilded cradle
And possibly https://github.com/adafruit/circuitpython/pull/2415 (though it's on the 4.x branch, so I don't know if it would have an effect on the generated circuitpython.org files)
Otherwise, no objections.
idle owl
@slender iron Can we get the frozen Circuit Playground library updated as well please? Not sure if it was already and that's done, or if it needs to be done specifically.
slender iron
looking at the prs now
onyx hinge
@slender iron thanks for the merges!
meager fog
merrrrge party
onyx hinge
I'm about to go out for pizza, I'll consider it a celebration
pastel panther
@slender iron dooooo eeeeeeet
drowsy geyser
onyx hinge
@slender iron safe travels!
quasi fjord
clearly the correct answer is the schedule an automated deployment at 1am...when you're on a plane
meager fog
just click deploy and then immediately shut your laptop
live, love, package with no regrets
drowsy geyser
@meager fog Is that like, "In case of fire, git commit, git push, run"?
solar whale
What could go wrong? π
drowsy geyser
@solar whale It depends if you're Boeing?
(not referring to the 737Max, referring to the Starliner software bug that caused mission failure this morning - 'cause nobody died there.)
solar whale
I thought so -- another "software" issue though...
drowsy geyser
Yeah, they need to do some SDLC tweaking.
drowsy geyser
This might be the wrong channel, but when writing CircuitPython applications we should take into account accessibility issues, e.g., non-hearing users. Does Adafruit have a list of these considerations we should accomodate?
slender iron
@drowsy geyser definitely good to do. we don't have anything official though
drowsy geyser
Thanks, @slender iron . I'll update the spec with some considerations. I'll have to ponder it a bit to see if there's an industry recommended approach.
solar whale
Hooray!! after a 8 hour build, I now have a working build of gcc-arm-none-eabi-9-2019q4-major running on my Raspberry Pi 4 !! just successfully compiled a Circuitpython board -- WooHoo! It actually went quite smootly from the source tarball -- there was one package not installed that made the first attempt fail after about 7 hours!!
raven canopy
@solar whale oh dear. i haven't gotten back to that stage of my project. thanks for undertaking the first trial!
drowsy geyser
@solar whale I run mine on a very high-end System76 Linux laptop. Reminds me of when Seymour Cray visited Apple. Steve Jobs proudly told him "we're designing the next Mac with a Cray." Seymour replied, "oh, I'm designing the next Cray on a Mac."
solar whale
I now have tarball of the binaries if you want it.
old smelt
Where does the release name come from when doing a local build? 5.0.0-alpha.5-26-g01bf93216-dirty
raven canopy
iirc, its the tag+commit_sha+dirty (dirty meaning uncommited/merged changes)
old smelt
Ok.
How can I build a clean released version?
Say I wanted 5.0.0-beta.1 ?
But build it locally for our board.
slender iron
@old smelt you should be able to check out the tag
slender iron
it'll be dirty if you add a board def
old smelt
Which I did
slender iron
you may need to rebase your commits onto the newest tag. (about to tag beta.2 too)
orchid basinBOT
solar whale
@slender iron Congratulations! Have a great holiday break!
slender iron
thanks @solar whale
old smelt
@slender iron - sounds good. Maybe I should wait for beta.2 and then start fresh on adding our board?
slender iron
@old smelt just tagged it π
old smelt
Whoa! And you're not offline already? I though you had a tight tag and fly plan. π
slender iron
not sure when I'm actually leaving. mixing in packing (just for two days) and babysitting the release
old smelt
Nice.
Also - the Add a New Board guide recommends opening an issue for a USB PID. In the long run, I think we'll get our own VID, but we don't have it yet. I'd like to start the process of adding our board to CP in the not-too-distant future, but I'm not sure the best route how to handle the USB VID/PID stuff. I don't want to squander Adafruit PIDs if we are getting our own later.
This is new territory for me. @meager fog - any guidance here?
slender iron
@meager fog may have an opinion. We haven't had anyone at that point yet
slender iron
thanks!
meager fog
@old smelt go ahead and request one - its when requests get into 3 or more boards that we ask people get their own VID
old smelt
Sounds good. Thanks!
slender iron
ok, release is out and blogged
orchid basinBOT
plucky flint
Hi folks... I'm about to step away for holidays. Season's greetings to you all and a happy new year. Here's to CircuitPython's continued flourishing in 2020. β π π
slender iron
You too @plucky flint ! We'll do #circuitpython-dev2020 posts in January so start brainstorming. π
drowsy geyser
I guess I better get my new library done well before January, then.
idle owl
@plucky flint Have a wonderful holiday!
manic glacierBOT
pastel panther
heya @slender iron did you upgrade to Catalina yet? If so, any gotchas?
slender iron
ya, I did early on
and yes, but doable
be ready with the security settings box open to OK a bunch of unsigned binaries
pastel panther
ok, thanks for the tip
I think I had heard an option to turn off checking for unsigned bins but that seems like a bad idea.
teal bear
I'm looking for the python? code that links ports/atmel-samd/boards/circuitplayground_express/pins.c#L7 { MP_ROM_QSTR(MP_QSTR_A1), MP_ROM_PTR(&pin_PA05) }, and cpx.touch_A1 ... or what sets the .. io direction?
tulip sleet
@teal bear touchio is a native module, so you won't see a Python impl of if
is that what you're asking?
teal bear
maybe - leme find it...
tulip sleet
teal bear
touchio.TouchIn yes.. that. thanks
whoever decided to hid this from the happy user ... good call
tulip sleet
it is hacky to save space in the library. The straightforward way is dcouemnted here: https://circuitpython.readthedocs.io/en/latest/shared-bindings/touchio/TouchIn.html
teal bear
it's also a little complicated if all you want to do is turn on a light
tulip sleet
import board
import touchio
a1 = touchio.TouchIn(board.A1)
while True:
if a1.value:
print("touched")
i agree it's harder than cp.touch_1
teal bear
touchio.TouchIn(board.A1) <- more dragons
ruby atlas
Apparently I need BUILDING.md already (both for the git submodule commands and for the JLink/gdb stuff!)
crimson ferry
Can someone sense-check something for me on a Feather STM32F405?
5.0.0-beta.1 doesn't do this, but 5.0.0-beta.2 does: Can't get it to show up on USB (macOS). Program with just a single print statement seems to run to completion (pulsing green NeoPixel), but no sign of the device in USB or as a Volume. In bootloader mode, it shows up fine (via system_profiler SPUSBDataType or dfu-util). Tried redownloading beta.2, reloading into device, etc.
If I revert to beta.1, it's all good again.
onyx hinge
@crimson ferry if you have github, will you file an issue please? I can check it out but not for awhile. It would be better not to let stuff get lost while the holidays are going on!
crimson ferry
@onyx hinge OK, will do. I was planning to, but thought I'd find out if I was missing something first. But I'll file one.
onyx hinge
we appreciate it
drowsy geyser
I think I borked something. I got a new PyBadge, so I built the latesst CircuitPython. When I loaded it, it said "line 6: board no such module". Since board is supposed to be compiled in, I was suspicious. I blew away my source tree and rebuilt the whole thing. When I flashed the new firmware, same problem. I'm sure I'm just doing something stupid probably because it's late and I should be asleep. But, can anyone advise? The build was clean.
manic glacierBOT
manic glacierBOT
solar whale
@drowsy geyser sorry -- all I can offer is a "works for me" I built CP 5.0Beta2 and loaded it into aPyBadge ```Auto-reload is on. Simply save files over USB to run them or enter REPL to disable.
Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 5.0.0-beta.2 on 2019-12-20; Adafruit Pybadge with samd51j19
import board
dir(board)
['class', 'A0', 'A1', 'A2', 'A3', 'A4', 'A5', 'A6', 'A7', 'A8', 'A9', 'ACCELEROMETER_INTERRUPT', 'BUTTON_CLOCK', 'BUTTON_LATCH', 'BUTTON_OUT', 'D0', 'D1', 'D10', 'D11', 'D12', 'D13', 'D2', 'D3', 'D4', 'D5', 'D6', 'D7', 'D8', 'D9', 'DISPLAY', 'I2C', 'LIGHT', 'MISO', 'MOSI', 'NEOPIXEL', 'RX', 'SCK', 'SCL', 'SDA', 'SPEAKER', 'SPEAKER_ENABLE', 'SPI', 'TFT_CS', 'TFT_DC', 'TFT_LITE', 'TFT_MOSI', 'TFT_RST', 'TFT_SCK', 'TX', 'UART']
I plan to add an API to MP3File so that the opened file can be changed hopefully without doing any memory allocations. I wonder, maybe I can just make the ".file" property assignable? Otherwise, what would the function be called?
drowsy geyser
@solar whale Thanks! -beta.2 from that link has the same behavior.
I wondered if I had broken something because when I got a box of CPBs I built Beta 1 for them and they worked fine. I'm curious if this is PyBadge specific for some reason.
I'll try it on a Feather M4 Express after I've been awake more than 10 minutes. π
Kinda let the caffeine and ADHJD meds hit the bloodstream. 
solar whale
@drowsy geyser you may want to make sure there is nothing odd on the file system. Delete any old code.py or wipe the file system and reload.
drowsy geyser
Ok, I'll do that. BTW, I did ```
git submodule sync
git submodule update --init --recursive
@solar whale Well that's odd. I deleted code.py and lib and when I connect to the REPL I can import board just fine....
I'm using the library bundle for 5.x from the 16th, so that should be good, yes?
(Starting to wake up. Thank you for the help! And I was all ready to blame Scott's "release and leave" model.)
crimson ferry
Is the "Measuring Battery" section of the Adafruit STM32F405 Feather Express learn guide https://learn.adafruit.com/adafruit-stm32f405-feather-express?view=all#power-management right? dir(board) doesn't show an A6 or board.VOLTAGE_MONITOR, maybe just not exposed yet?
solar whale
@drowsy geyser BTW drop the βrecursive. Should not need it anymore and it might cause problems.
Thatβs new
drowsy geyser
Ah, ok. I'll drop a note to Kattni to edit the "Building CircuitPython" guide. π
solar whale
I think it has been updated
drowsy geyser
Wow. My brain probably just added the --recursive 'cause that's what I always did. Thanks!
Now that I have it working, is there a list of the libraries that PyBadge needs for basic functionality (i.e., before FeatherWings and sensors?)
I didn't see one in the guide....
solar whale
I have to go offline. Check for the pybadger guide.
drowsy geyser
Okie doke. Thank you again for the help!
idle owl
@drowsy geyser @solar whale The guide had not been updated. It has been now.
You're thinking of the conversation surrounding the new BUILDING.md. It was updated there.
tulip sleet
@onyx hinge an MP3File.file property sounds fine. But maybe then the class should be called MP3Stream or something?
onyx hinge
I named it like WaveFile, but you might be right
tulip sleet
well, if WaveFile could do the same thing, then we could just be consistent. I was just wondering how immutable these are meant to be (also things like OnDiskBitmap)
it's not MP3Player because the .play() is elsewhere, right?
it's a bit of a weird secret that mp3.file = would be more efficient than creating a new object
maybe it's a "leaky abstraction"
idle owl
@gilded cradle You around?
gilded cradle
Hi Kattni
idle owl
Hey. Quick question: How many boards are on circuitpython.org/downloads?
If you know, figured you might.
gilded cradle
yeah
idle owl
Or 90+
Trying to write a quick 2019 closing blurb in the notes doc, and wanted to mention it.
gilded cradle
Looks like 94
idle owl
Ok. Thank you!
idle owl
oi... if I had known, I would have done that! I figured you had some magic grep thing or something.
gilded cradle
Probably some easier way, but I haven't looked into it
idle owl
Sorry about that π
gilded cradle
No problem
idle owl
Thank you so much for doing that
idle owl
Not a bad idea. File an issue maybe?
gilded cradle
Sure
raven canopy
there is a report now that says how many boards are supported... π
raven canopy
though, that doesn't exactly == circuitpython.org.
idle owl
Well NYAH then. π
raven canopy
Number of supported boards: 97
I wonder what the discrepency is. Maybe some disabled ones.
raven canopy
could also be some of the newer ones that aren't added to the site yet?
Β―_(γ)_/Β―
onyx hinge
@tulip sleet I assume that it's memory fragmentation, but jeplayer eventually crashes with out of memory, typically when creating the MP3File object for the new song to be played next..
where "eventually" is "maybe << 20"
raven canopy
@gilded cradle robothamm1 (original) is still listed on the report, so that's 1 of 3.. π
gilded cradle
Yeah, that should be gone now.
raven canopy
hmm. can't find cp32 & datalore... they're in the data file and have a _board/x.md. but that would make 3.
idle owl
<@&356864093652516868> Here is the notes doc for the NOTES ONLY Edition of the CircuitPython Weekly, wrapping up on 27 December 2019. THERE WILL NOT BE A RECORDED MEETING THIS MONDAY OR THE NEXT. Everyone is welcome to add notes! Please feel free to add your Hug Reports and Status Updates throughout the week. We will be posting the notes doc to GitHub as usual. The next recorded meeting is 6 January 2020. I will post another reminder later in the week with the link and the meeting schedule. Thank you so much for everything you do. We'll see you in the new year! https://docs.google.com/document/d/1qcUDQqpMWpGJG-k72A39dbQ3a86JXMlnighWci6pS20/edit#
raven canopy
solved. both the cp32 and datalore boards have downloads_display = false. that was bugging me. π
idle owl
@raven canopy I knew you were still figuring that out.
raven canopy
πΏοΈ!! its a condition...
also, the curse of idle time while waiting on a deployment..
idle owl
The code is crazy, there's this aggregate pixels thing that maps out the whole tree vertically and horizontally I believe, for animations that go around the tree in different ways.
Roy really wants to make it so an animation goes up the tree to the star and explodes outwards. The pixels in the star are in series so they can all be sliced as needed.
The star happens to fit 5 NeoPixel sticks and a jewel. Didn't plan it that way, simply printed the first hollow decent looking one I could find.
crimson ferry
Has anyone built mpy-cross in Catalina? sudo to get past all the permissions issues (o_O), then I'm getting errors I don't grok: make: msgfmt: No such file or directory make: *** [build/genhdr/en_US.mo] Error 1
raven canopy
@crimson ferry iirc, that error is related to not having gettext installed.
crimson ferry
Hm, had an outdated version, but same error with upgrade
idle owl
@crimson ferry Scott ran into all kinds of issues upgrading to Catalina, almost all of them security related. I don't know how gettext works but it could be a security issue. I don't know the details well enough to assist further than informing you of the vague possibility. Scott is out, so he's not available to assist either.
crimson ferry
Thanks, sudo seemed to get past of tons of permissions issues (probably not the best way), I'll keep digging.
Found a binary https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/creating-a-library#mpy-2-11 I'll just use this for now since I just need it to test something for an issue. ...oh, this is from CP 3.x days.
solar whale
@crimson ferry I don't think mpy-cross has changed sinc 3.x so it sohuld be OK
crimson ferry
Looks like Scott already worked through the make problem too: https://forums.adafruit.com/viewtopic.php?f=60&t=139567&p=690861&hilit=make+msgfmt#p690861
solar whale
good -- I am not haveing any issues with Catalina -- once I dealt with the security issues
drowsy geyser
I use my Mac (Catalinia) for writing files from my NAS down onto boards. π
@idle owl Should I just have gone ahead and modified that guide (how to build circuitpython)? I thought it would be more polite to just let the author know....
idle owl
@drowsy geyser Either way is fine. I didn't write that one anyway.
ruby atlas
Sigh, found a new fun one with subclasses of native classes. Native methods don't get the instance handed to them. The protocol used for properties works great, so I think I probably need a custom attr handler.
tulip sleet
@ruby atlas there are several examples in displayio of Python subclasses of native classed, I think extracting the instance.
timber remnant
Is there a known way to add QSPI or SPI RAM to the Edge Badge? I'm willing to create hardware and software, but I don't need to re-invent it. Can't find any examples after searching. I need to record signals. Flash would get hot and wear out quickly. I am admiring the QSPI-capable RAM chips such as http://www.issi.com/WW/pdf/IS62-65WVS5128FALL-BLL.pdf ( ISSI IS62WVS5128FALL-20NLI ).
tulip sleet
@timber remnant the QSPI bus is already in use for the QSPI flash chip, and chip-select pin on the flash is tied high, so you'd have to do some unsoldering etc if you wanted to share the QSPI bus. But you could use regular SPI on the Feather header pins. Or you could use an SD card with the Adalogger FeatherWing (https://www.adafruit.com/product/2922) and write gigabytes of data. It depends on your application.
ruby atlas
@tulip sleet thanks.
tulip sleet
i might be repeating myself. Look for mentions of "native" in those C-implemented classes.
ruby atlas
Hmm... I'll poke around more in there when I have time (tomorrow probably!)
looks like native_group in Group.c might be a clue.
tulip sleet
yup, Group was one I remembered
ruby atlas
what's not clear is if overridden native methods in Group behave the way I need (a native method call being able to call a subclass method -- specifically for native .fill() to call .show() ).
tulip sleet
i think that is the idea, but I'm not sure
ruby atlas
hmm, native_pixelbuf is pretty much the same as native_group . The only thing I don't do is assert that it must be subclassed, because PixelBuf is usable natively. Anyhow, it gives me something else to test with!
timber remnant
@tulip sleet My application is to keep taking data with different sensors, process, and display it as quickly as possible with CircuitPython. The code is on github with an MIT license. Recently it has been refactored into a Model-View-Controller structure. Check it out, the sensors channels all inherit from the same base class.
https://github.com/tomacorp/edgemicscope
cursive condor
can someone confirm that accessing accelerometer via lis3dh is faster than using cp.shake() ? (on a circuit playground bluefruit)
meager fog
yes, shake does a lot of waiting around - its meant for ease not speed π
cursive condor
ok. I noticed the snowglobe animation with tft gizmo was slower with cp.shake(). thanks a lot π
astral dagger
Hey, all β brand new to circuit python, and have got an oddness with two brand new Feather M0 boards
I can run the bootloader, install the current version of CP, and then ... no CIRCUITPY
I can get to the REPL on the serial console, no problem
Since I've done nothing with the boards thus far, I figure a therapeutic reset may be in order
but I can't seem to erase the filesystem on either board:
>>> import storage
>>> storage.erase_filesystem()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: Invalid pins```without the filesystem mount, I'm pretty much stymied. (OSX, 10.14, for what that's worth)
raven canopy
@astral dagger that...is def a confusing exception. i would try loading 4.1.1. also, could you file an issue on what you're seeing? we're in a holiday lull right now..
astral dagger
Can do!
(I was actually getting two boards together to do some holiday tinkering during my own lull, but I'll downrev and see if the behavior changes)
raven canopy
off the top of my head, 4.1.2 likely saw some USB and build toolchain changes at minimum. could be the cause... but i'll let the experts talk to that.
astral dagger
oh dang it
well, this is what rubber duck debugging does for one
short form: Feather M0 Express != Feather M0 Basic
I'll just be slinking off into the "can't read, shouldn't ever go on the internet" corner over here
raven canopy
ahh. hehe. i've got that t-shirt too! (i think most of us do.)
astral dagger
and lo! When I actually install the right !@#$ firmware, CIRCUITPY shows up.
raven canopy
invalid pins is still an interesting error message. but likely due to the VM/board reset functions.
astral dagger
thanks for your patient understanding. π
stuck elbow
I wonder if it would make sense to not even compile the erase_filesystem machinery on the non-express boards, could save some flash space
raven canopy
imo, erase_filesystem falls in the "better to have it, and not need it. then to not have it, and need it." category. π
granted, if we can keep board_DANGER_erase_filesystem.uf2 readily available, then its definitely an option.
stuck elbow
well, on non-express boards you can erase the filesystem by simply re-flashing the original uf2
no need for a special uf2, since the filesystem is on the chip's flash
raven canopy
huh. i thought it still left the fs intact, even on internal...
stuck elbow
only if the uf2 is smaller than your flash
fierce girder
@idle owl @slender iron is there CircuitPython meeting this monday (Dec 23)?
solar whale
@fierce girder this weeks meeting is "notes only" https://docs.google.com/document/d/1qcUDQqpMWpGJG-k72A39dbQ3a86JXMlnighWci6pS20/edit#
fierce girder
Ok thank you
stuck elbow
Anybody had this problem?
circuitpython/lib/tinyusb$ git submodule update
fatal: not a git repository: /home/sheep/dev/circuitpython/circuitpython/lib/tinyusb/hw/mcu/nxp/../../../../../.git/modules/lib/tinyusb/modules/hw/mcu/nxp
Unable to find current revision in submodule path 'hw/mcu/nxp'
@stuck elbow yes, jerryn and I ended up recloning the repos. it seems to be because the nxp dir was initially not a repo, and it can't recover (but I haven't looked at it in detail). You might try deleting the directory completely and git submodule update again, but for me it was easier just to trash and re-clone.
We thought it was due to git submodule update --recursive, but I see you are NOT including the --recursive, and it still fails
stuck elbow
I just re-cloned as well, thanks
onyx hinge
@slender iron jeplayer is driving a lot of changes in the core. do you want multiple PRs when they're separate identifiable topics, or can I just go ahead and accrete stuff? It's pretty much combined to audio DMA + MP3File changes at this point, but might get into displayio if I'm not careful
https://sourceware.org/bugzilla/show_bug.cgi?id=22647 I think this bug is affecting me π¦ π¦
slender iron
@onyx hinge one big PR is fine. I'm only lurking the next week or so
and I don't want to slow you down
onyx hinge
okie dokey. if you want me to split it later, that's fine. (I try to group within commits so they're different topics)
wooooo! I got a DMA hang detected + kicked
stuck elbow
what was it?
onyx hinge
well, it's whatever this samd51 DMA bug is
i'm detecting it a little differently than nofrendo did, by checking whether the new DMA started happening or not within a small time window
stuck elbow
and congratulations
onyx hinge
// CHSTATUS.reg = 0x3 (BUSY | PENDING). On the other hand, this is a
// legitimate state for a DMA channel to be in (apparently), so we can't use that alone as a check.
// Instead, let's look at the ACTIVE flag. When DMA is hung, everything in ACTIVE is zeros.
bool is_okay = false;
for (int i=0; i<10 && !is_okay; i++) {
bool complete = true;
if (rx_active) {
if (DMAC->Channel[SHARED_RX_CHANNEL].CHSTATUS.reg & 0x3)
complete = false;
}
if (tx_active) {
if (DMAC->Channel[SHARED_TX_CHANNEL].CHSTATUS.reg & 0x3)
complete = false;
}
is_okay = is_okay || (DMAC->ACTIVE.bit.ABUSY || complete);
}
if (!is_okay) {
dma_kick_count ++;
for (int i=0; i<AUDIO_DMA_CHANNEL_COUNT; i++) {
if(DMAC->Channel[i].CHCTRLA.bit.ENABLE) {
DMAC->Channel[i].CHCTRLA.bit.ENABLE = 0;
DMAC->Channel[i].CHCTRLA.bit.ENABLE = 1;
}
}
}
``` is what I've got right now, dunno if it's what the final form should be or not.and I wasn't in the room when the code activated so I don't know if it produced an audible glitch or not. but in any case, I'll take it for now...
stuck elbow
a glitch is definitely better than a hang
modern wing
I'm sorry I've been away for a bit, but I'm assuming no call today?
drowsy geyser
Google Doc only, @modern wing ...
modern wing
Gotcha, thanks @drowsy geyser
drowsy geyser
I'm just a relay. Message is from @idle owl . π
modern wing
You've passed the message baton with grace and strength π
The last few weeks, I've had....life...happen on Mondays, and I just haven't gotten to the recordings. So today I'm available!.....and we're doc-only haha.
main meteor
Would that be a bΓ’ton violet?
idle owl
<@&356864093652516868> No meeting today! We are doing a notes-only edition that will be ending on Friday 27 December 2019. Notes doc is here: https://docs.google.com/document/d/1qcUDQqpMWpGJG-k72A39dbQ3a86JXMlnighWci6pS20/edit#
onyx hinge
Group hug into the void!
stuck elbow
if you hug the void long enough, the void hugs back
drowsy geyser
If you hug the void long enough you reach the heat death of the universe.
stuck elbow
for some value of "you"
old smelt
"Hug the Void" would make an excellent name for a rock band.
stuck elbow
or "Void Huggers"
stuck elbow
@drowsy geyser I mean, long before the entropy reaches maximum, the particles will be too far apart to interact β hard to sustain life in such conditions
drowsy geyser
This is true.
modern wing
@old smelt Your line about "...good name for a rock band" reminds me so much of the humor columnist Dave Barry. π
drowsy geyser
Which reminds me, I need to say Happy Holidays to a few people, Matt O'Dowd and Diana Cowern included.
turbid radish
Happy holidays to you all. We'll have some great CircuitPython in 2020 which is saying something as 2019 was spectacular.
old smelt
@modern wing - that's exactly where it came from!
I was a big fan.
I have an Evernote note filled with names.
Another name recently inspired by this Discord channel is "Tarball of Binaries" which I think I owe to @solar whale .
ionic elk
@stuck elbow also having the submodule update issue. Making an issue on Tusb now.
neat folio
@old smelt "the Void Hugs Back" was their best album IMHO π
