#circuitpython-dev
manic glacierBOT
solar whale
@tulip sleet My mother always said "check the wiring" - bothte BMP280 and MCP9808 work fine on the M4 - sorry for the confusion. Great job on this - nice to have sensors!
solar whale
@tulip sleet also checked out an mpr121 touch sensor no problems so far.
hollow tartan
@slender iron very interesting proposition RE usb descriptors! ... we hope to run the library on CircuitPython itself! That way we can change the descriptor dynamically on boot! We’ll be able to choose between HID, MIDI and other things dynamically.
I am no expert in the USB standards arena and therefore I am unable to compare and contrast AdaFruit's approach verses Arm Keil's approach. https://www.keil.com/pack/doc/mw/USB/html/_u_s_b__descriptors.html
solar whale
@tulip sleet @pastel panther OK this must be close to a tri-fecta - I was able to write text to an ssd1306 on the M4 under 3.0 with the ssd1306 mounted to my metrowing!!! woohoo! So this tests i2c, framebuf and the metrowing!
and with the ssd1306 module loaded look at the free memory!! ```
from board import *
import busioImport the SSD1306 module.
import adafruit_ssd1306
i2c = busio.I2C(SCL, SDA)
display = adafruit_ssd1306.SSD1306_I2C(128, 32, i2c)
display.text("yowza",0,0)
display.show("yowza",0,0)
display.show()
display.text("yowza",10,0)
display.text("yowza",10,10)
display.show()
import gc
gc.mem_free()
114784
tidal kiln
usage question about @ property's:
i'm using a class that has several properties, so you can do:
reading1 = some_class.foo1
reading2 = some_class.foo2
but how can i create a sequence object of properties i'm interested in so i can iterate over them and get their values?
solar whale
The metro_m4 is under the metrowing! having way too much fun now that i2c is working!
idle owl
@solar whale So cool!
solar whale
Wow! running the color wheel demo on a "jewel " (7 neopixels) on the M4 - the update response is so much faster than on the M0!
slender iron
@hollow tartan yup yup, your comment made me want to look at the HID stuff 😃
solar whale
It really is great - I switched to a 16 pixel ring and it loops around about 3 times faster than on the M0
slender iron
we need a reaction emoticon for speed
drowsy geyser
Good Lord! There's some massive github activity tonight! 😃
idle owl
@solar whale Are any of the externals you have hooked up RGBW? I couldn't get mine to work.
solar whale
no - sorry - only RGB's here
idle owl
Ok. No worries 😃
tulip sleet
@solar whale thanks for all the I2C testing! Glad it's working out. SPI is probably next.
solar whale
@tulip sleet Thank you for getting it going!
idle owl
Hm.
Set bpp=4, nothing. Change it to bpp=3 and it does the weird three-quarter-light-wrong-color thing it does when you send RGB data to RGBW.
solar whale
@idle owl looks like I managed to sorrupt my FS - How to erase the FS on your M4?
idle owl
I got it to light up a single color! The issue lies in the animation code.
@solar whale JLinkExe, connect, erase, but then you have to reload the bootloader too. loadbin foo.bin, 0 I believe.
That's what I had to do yesterday because reflashing just the uf2 failed.
solar whale
ok - thanks
tulip sleet
@solar whale if you replace the bootloader and then use the update-bootloader.uf2 to overwrite itself, it should set the BOOTPROT fuse to protect it in the future. Unless JLink erase turns those fuses off. You could also use bossac -e with the M4-compatible bossac.
idle owl
@solar whale What code did you use for the color wheel on your NeoPixel jewel?
@tulip sleet I don't think erase turns it off. Scott posted a link yesterday with the command to run to turn them off if you need to.
solar whale
its been too long since I did this - can I upload the update-bootloader.bin with Jlink, or should I jsut use bootloader.bin
idle owl
@solar whale Thank you!
Both will work with JLink I thought, the update-bootloader is the one that will set protect.
pastel panther
@solar whale Awesome! Glad it's working, at least in part! I should have mkii done later this week
hollow tartan
@idle owl Oh, I think you mean the changes/updates are applied very quickly. Can you do the math: Actual verses theoretical maximum for the NeoPixel Strip?
idle owl
@hollow tartan I have no idea how to do that.
hollow tartan
scope measurement and data sheet I think
meager fog
@tulip sleet @slender iron im taking a break from some horrible linux stuff. is m4 i2c workin and ready for test?
meager fog
kool yay i also got the feather m4's made if ya feel like making a config for em
idle owl
@meager fog oh rad!
solar whale
I ran tests on several I2C devices - ssd1306, mcp9808,bmp280, mpr121 -- all good!
meager fog
oh yay thanks jerry! thats awesome of you 😃
slender iron
@meager fog I can whip it up now if you like. Will have to bail when becca gets home
meager fog
sure
its committed - use Rev B2
ill test later tonite - i am in some linux stuff i want to get done
solar whale
It is fantastic to have memory to play with - have 2 temp sensors running and dispolaying on OLED!
meager fog
or cna do tomorrow, no rush
slender iron
@meager fog which chip does it have? samd51j19?
meager fog
g19
slender iron
just loaded it in kicad 😉
pastel panther
I'm assuming it's on on the adafruit github yet?
pastel panther
makes sense.
solar whale
@idle owl @tulip sleet whew! reloade and running again. Thanks for the help. I need to do that more often 😉
idle owl
@solar whale You're welcome! I know the feeling.
slender iron
@meager fog want a feather m4 bin?
meager fog
uf2 is fine
slender iron
feather m4 express
pastel panther
swoon
slender iron
I also claimed 0x8026 for it in the usb doc @meager fog
meager fog
k
slender iron
code is here: https://github.com/tannewt/circuitpython/tree/feather_m4
solar whale
How much space should teh be in CIRCUITPY on the metro-m4?
idle owl
I'm pretty consistently having eventual USB issues. Which is to say, if I leave it sit for a while and then try to interact with it in Finder or by editing a file in Atom, both freeze and I have to physically reset the board.
@solar whale Finder says 244kb.
solar whale
@idle owl and how much is free?
idle owl
I closed the info window before I could check and now it's not mounting. I want to say 190kb.
I have a couple of code files on there, my current main.py has quite a bit in it because I commented out a lot to try different code, and neopixel.py lib is on there.
solar whale
nevermind - in finder I see 244kb and on;y 7 in use -- confused on linux - I tried to copy entire /lib to it - not good!
idle owl
Ah, yeah
solar whale
I'm seeing some funny behavior too, but trying to figure out if it is just me doing silly tings.
idle owl
219kb free
solar whale
Thanks - sounds right - and my linux box is happier now after I reconnected it.
idle owl
I had to plug it into USB2 on Windows again, still reset it about 10 times before it would show up. Changed main.py to something else. Mounted immediately on Mac with USB3.
But it's still acting oddly. I don't have the brain to deal with it tonight though.
solar whale
I know that feeing - I am about to quit as well.
idle owl
@solar whale I deleted all the commented out code from main.py, put it in a different file. Seems to be handling being ejected and reset better now.
hollow tartan
@solar whale is the flash drive file system on M4 visible and usable? so you can just copy a .py in and run it from REPL?
idle owl
@hollow tartan Yes. It works like all the M0 CircuitPython boards.
solar whale
@hollow tartan yes
solar whale
@idle owl sounds better - I finally think I have thisgs restored. the M4 does not handle running out of space very well
@hollow tartan I doubt it will happen soon if at all. The USB interface to the esp8266 is very different and my understanding is that it cannot be made to work as a flash drive like the m0 and m4.
idle owl
@solar whale Good to know about running out of space.
solar whale
I need to get to bed - early moning ahead - Goodnight all!
idle owl
@solar whale Goodnight!
hollow tartan
thanks. nite.
pastel panther
'night @solar whale
tulip sleet
@idle owl are you doing a sync immediately after writing the file? That may help.
idle owl
@tulip sleet No, how and where would I be doing that? I edit in Atom and save files, and sometimes I connect via serial, other times I don't.
manic glacierBOT
tulip sleet
Just type sync in another terminal window after you write in Atom. That flushes writes to all disk drives.
idle owl
Ah
tulip sleet
Or, edit the file locally, and then do cp /foo/bar/main.py /media/<username>/CIRCUITPY; sync
idle owl
And you think that will stop it from freezing everything? Or was this in response to a different issue.
tulip sleet
If the freezing is due to write delays, it will get rid of that. Scott has pushed several MSC changes recently. If you can pin it down to a specific commit, that would be helpful to him.
idle owl
I had significantly more code in main.py than I was using.... I'd say a full 2/3 of it was commented out as I was trying different things to get the rainbow animation to work on both an RGB strip and an RGBW ring. I removed all the commented out code, and it's been playing nice for at least the last little bit of time. But it seems like it just happens over time again, that it freezes and doesn't eject, requires physical removal, and then doesn't mount.
tulip sleet
for instance, removing all the comments makes the file smaller, and so it may write completely in one go instead of being delayed
idle owl
See, exactly 😃
tulip sleet
but it could just be length of time...
idle owl
I'll leave it running, connected to serial overnight and see what happens.
idle owl
I'm on MacOS at the moment. But I updated the Linux lapatop last night to 17.04.
tulip sleet
I'm on 16.04. But not 17.10?
idle owl
Yeah, I was wondering the same thing. I looked to see if it would let me do 16.04 through the GUI, but it offered 17.04. I just checked the updater to see if it would tell me what version I was running, it says 17.04 and offered to upgrade to 17.10.
So who knows why it wouldn't go straight to 17.10.
So I could run the upgrade to 17.10.
(It was on 16.10)
I hadn't done anything with it in a while.
tulip sleet
i usually just stick to the long-term support releases. I did drop unity for cinnamon (the mint UI), because it was kind of non-obstrusive
coarse barn
I'm having filesystem problems with a Gemma M0. According to https://circuitpython.readthedocs.io/en/2.x/docs/troubleshooting.html#valueerror-incompatible-mpy-file I should reflash with a uf2 to erase the CIRCUITPYTHON filesystem, but none of the ones linked seem to load.
idle owl
Gemma doesn't have SPI flash.
tulip sleet
is this on Windows?
tulip sleet
do you have Arduino installed?
coarse barn
I just flashed to CircuitPython 2.1.0 and am trying to update the libs for the Gemma 2.0
Yes, I have arduino installed as well.
tulip sleet
if you load any Arduino program on the Gemma it will erase the entire flash. Then load CPy again and it will create the filesystem from scratch (empty)
so just load blink or similar
what kinds of errors are you getting?
coarse barn
the filesystem appears to be full, can't save to the fs.
idle owl
Gemma will fill up if you try to load the entire library bundle.
Did you choose some specific ones and it's still giving you those errors?
coarse barn
I should be specifically running the lib directory from gemma_m0_2.0.0 which is a minimal set of hid, dotstar and neopixel
idle owl
Ah ok
tulip sleet
on the mac, many mac junk files are copied to the filesystem at the same time
I'm trying to find Tony DiCola's writeup of how to clean those up or avoid them in the first place.
idle owl
@coarse barn I would start by loading an Arduino program to your Gemma, reflash CircuitPython, load the new versions of the three libraries, and see if you're still running into issues. If you are, you can use the info in the guide to remove the Mac OSX hidden files.
@tulip sleet Good night
And I need to head out as well. It's late here.
coarse barn
left out the adafruit_hid and got the python code to run, once I removed the instances of Keyboard from all the demo code.
floral dagger
So, been working on trying to get cookiecutter to work, and am getting nowhere. Does it only work on linux?
If so, that really should be mentioned in the instructions
coarse barn
@idle owl thanks for the help!
slender iron
@floral dagger cookiecutter is written by others so I'm not sure of the windows state. its a step you could skip though
@coarse barn glad you got it going
timber mango
Hi all, I don't know if this is the right place to drop my question (Mike advice me to look here), but has someone managed a working connection with an ethernet module (for example the WIZ820io) in combination with MicroPython/CircuitPython (for example the Feather Huzzah/Feather M0 Express)? For my project I need to set up a telnet connection <> Uart Tx/Rx. The physical connection isn't the problemen, but it's my lack of driver experience (if any exists).
With kind regards, Jeroen
drowsy geyser
Hullo, CircuitPython friends. Is anyone else having trouble with tonight's firmware build on Windows? I built the firmware, flashed it, and now I can't talk to the COM port and I don't get CIRCUITPY mounting, regardless of reboots and resets. Just wondering. I'll head off to GitHub and open an issue.
hollow tartan
@timber mango i just looked on github and there seem to be some relevant projects there. go to the main page and search. Sorry but I have no knowledge of that chip.
stuck elbow
@timber mango I'm afraid we haven't gotten the drivers for all the radio and network chips written yet
solar whale
@drowsy geyser Just did a fresh pull/build on linux box - works ok ```Adafruit CircuitPython 3.0.0-alpha.0-1154-gb4fc464 on 2017-11-08; Metro M4 Express with samd51j19
Hmmmm, thanks @solar whale !
drowsy geyser
No worries! Yes, I did.
solar whale
good luck!
I see you havea 3.0.-beta tag - is that something local?
I don't see that tag on the repo.
solar whale
@tulip sleet @slender iron Just a heads up - not sure if this is somethign to watch. First, I was able to make a 3.0 Bundle - yay! - just made a local tag and then built to it.... Then,I forgot the M4 has only limited Flash space and I tried copying the whole lib/ to it. It did not give me any errors - inidicating it was out of space but ithe FS really got very messed up and eventually I just had to erase the whole thing and reload. It is working now with just the necesary divers loaded. I have not tried to reproduce it. I was just curious if you had any experiece with filling the FS on the M4 yet. Do you want me to explre this and file an issue on it or just chalk it up to alpha testig at this point?
timber mango
@stuck elbow no problemen, from Mike I hard the samen answer. @hollow tartan I will search on Github. I can't imagine ethernet and python isn't something that's unique ;)
stuck elbow
I guess we need someone who would be interested in using all this networks things and willing to work on them. Most of us so far focused on sensors, displays and actuators. I briefly dabbled with the nrf24l01, but I don't currently have a project that would use it.
Generally, the communication stuff is rather hard and complex, and requires domain-specific experience, so it takes longer than, say, neopixels.
tulip sleet
@solar whale (@slender iron) Do you have a non-Express M0 board you could try filling up the filesystem on with 2.1 loaded? See what kinds of errors it gives. It could be that the fs code is not as robust in 3.0. In any case, filing an issue would be great. Thanks!
solar whale
@tulip sleet Yes - I'll try it on a gemma or tinket and sse how it goes. I know I have recieved out of space erros on the trinket in the past. Hopefully I can get some testing on it later today. This day job gets in the way 😉
manic glacierBOT
tulip sleet
@drowsy geyser I can confirm that the USB drives are not acting right on Windows. On Win10 build 17025, I see two drives without names. On Win7 I see one device with messed-up properties, and no drives in Explorer. I'll do some bisecting to find the cause. I have some ideas but it could be one of several things.
drowsy geyser
Thanks, @tulip sleet ! Let me know if you need any data from my system.
tidal kiln
@idle owl how do i play silence with cpx.play_tone() ?
bronze geyser
I'm collecting sensor data on a Huzzah running Circuit Python. If I do a While True: .... sometimes I can't break in and stop main.py...it just won't. I end up having to erase/reinstall everything. Is there a recommendation for similar to loop() in arduino but with the ability to break? (note: sometimes I can get into repl and os.rename('main.py','foo.py') but not always...then i really don't seem to have any pins left on the Huzzah to insert a button...?
..hmmm...i guess since the sensor thing is publishing mqtt, i'll have it subscribe to a control mqtt message and tell the thing to break out of main.py...
opal elk
how are you trying to stop it?
stuck elbow
@bronze geyser I think adding a sleep() in the loop might help
bronze geyser
I was hooking up to serial, opening screen, then ctl-b. Sometimes this worked. But then another time I had a bug in my main.py code, so the code crashed..crashed...kept cycling crashing. I couldn't break in. I thought then it would be great to have a jumper to GND and one of the pins. Then if the pin goes low. ...
bronze geyser
thanks @stuck elbow I had a time.sleep()
@stuck elbow I tried ctl-c ...and every other ctl-combo just for giggles.
the thing is, the Huzzah refused to release the serial once main.py got upset.
screaming "get a grip" to main.py didn't work either...
solar whale
🔨
bronze geyser
stuck elbow
that might actually be the reason for it being non-responsive, what kind of bug?
solar whale
not sure if it matters for control-C , but other contol characted have to be the first char in al ine to be recognized. Make sure you have not sent anyhting else ahead of it.
bronze geyser
i am not sure what was ultimately wrong. It got late yesterday. So I stopped and started today by wiping the Huzzah and installing everything again. I realize this might have been selfish, but i wanted to feel better about all of this....
@solar whale thanks. I like the hammer idea too though...
solar whale
@stuck elbow do you know if control-c should be recognised any time or just as first char in a line?
stuck elbow
@solar whale it only works when your program is running
solar whale
good point!
stuck elbow
@bronze geyser no worries, but it does look like some kind of more low-level crash than a regular exception
bronze geyser
@stuck elbow i think you are right. In the future, is there an action i can take that would benefit circuit python ?
stuck elbow
@bronze geyser writing the minimal example program that reproduces the problem and reporting it as a bug
bronze geyser
@stuck elbow Thank you. will do.
slender iron
@bronze geyser another thing you can do is asking someone like @stuck elbow or @solar whale to reproduce it on their huzzah
bronze geyser
@slender iron thank you. Good advice.
solar whale
@bronze geyser I'll be happy to give it a try if you have an example.
bronze geyser
@solar whale thanks...unfortunately I erased before I put together debugging.
solar whale
@bronze geyser No problem - it'll come back 😉 anytime.
bronze geyser
Yes indeed @solar whale
tidal kiln
>>> cpx.play_tone(4000,1)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "libraries/drivers/circuitplayground/adafruit_circuitplayground/express.py", line 394, in play_tone
File "libraries/drivers/circuitplayground/adafruit_circuitplayground/express.py", line 421, in start_tone
ValueError: Unsupported playback frequency
anybody know what the supported range is? (having trouble finding it)
appears to be 3500 ?
idle owl
@coarse barn You're welcome! I'm glad you got it sorted.
@tidal kiln I was hoping entering a frequency of 0 would work, but it doesn't. I'm not sure how you would use it that way.
tidal kiln
@slender iron your link is to core code though, right?
slender iron
ya
tidal kiln
so how would the cpx lib affect the 350000 cut off?
slender iron
@tulip sleet I found one issue where the inquiry response isn't always aligned
cpx is play a sample of length n at frequency n * frequency I think
idle owl
@tulip sleet I left the M4 connected overnight, and at some point the serial connection dropped, and when I tried to interact with it in Finder, it froze until I physically removed tfhe M4. I ran into the same Windows 10 errors last night when trying to resolve my issue - 2 unnamed drives - but a bunch of resetting eventually got circuitpy to mount.
tulip sleet
@slender iron good detective work. do you know the place to fix yet?
slender iron
do COMPILER_ALIGNED(4) for inquiry_info and format_capa
I don't think this would cause things to half work like @idle owl is seeing
tidal kiln
@idle owl for simple melody playback where a 0 could allow for a pause:
melody = (400, 0, 323, 480, 0, 0, 340)
for n in melody:
cpx.play_tone(n, 0.1)
@tidal kiln It gave me an unsupported frequency error when I tried to do cpx.play_tone(0, 1)
tidal kiln
i'm currently doing something like this:
melody = (400, 0, 323, 480, 0, 0, 340)
for n in melody:
if n == 0:
time.sleep(0.1)
else:
cpx.play_tone(n, 0.1)
@idle owl yep, that's why i was asking how to play silence
slender iron
@tidal kiln I like the second form because its clear its not playing a tone
tidal kiln
but i'm also having an issue on the high end, can't play anything above 3500
idle owl
Is there a theoretical limit for the onboard speaker? Or is that a code issue.
Or actual limit, I guess. Theoretical is probably not the right word for my question.
tidal kiln
probably code, like @slender iron pointed out, the resolution of the sine wave in cpx
idle owl
Makes sense
tidal kiln
more resolution = can go higher, but requires more storage
tidal kiln
the arduino lib works much differently, it just uses timer interrupts
hardwired to 100:
https://github.com/adafruit/Adafruit_CircuitPython_CircuitPlayground/blob/master/adafruit_circuitplayground/express.py#L385
350000 / 100 = 3500
yeah math
opaque patrol
the first parameter for play_tone is frequency, I don't think 4000 hz will work
tidal kiln
@slender iron i tried to drill in to the c code starting from here:
http://circuitpython.readthedocs.io/en/2.x/shared-bindings/audioio/AudioOut.html
but github link was stale
@opaque patrol correct. it won't. for software reasons. apparently.
slender iron
@tidal kiln I linked you to the core code earlier
tidal kiln
@slender iron yep. thanks. was pointing out other in case docs are out of date or something.
or maybe that was the wrong place to start looking?
(i was doing that hours ago)
opaque patrol
Looks like 4000 should work...
in ports\atmel-samd\common-hal\audioio\AudioOut.c
line 546: common_hal_audioio_audioout_set_frequency
if (frequency == 0 || frequency > 350000) {
mp_raise_ValueError("Unsupported playback frequency");
}
@opaque patrol that's lower level
slender iron
@tidal kiln I don't know how read the docs creates code links. I usually just search the code for the error message to find where its coming from
tidal kiln
@slender iron me too. it wasn't obvious in this case though.
see stack trace above
also confusing as line 421 is a comment line:
https://github.com/adafruit/Adafruit_CircuitPython_CircuitPlayground/blob/master/adafruit_circuitplayground/express.py#L421
(or is that due to mpy?)
slender iron
I don't know
tidal kiln
then i really don't know 😀
tulip sleet
@tidal kiln your .mpy might just be out of date with the source
tidal kiln
thought of that too, not sure how to confirm though, it's whatever is in 2.1 bundle
@tulip sleet that's it. just looked at files from previous commit.
tulip sleet
the bundle release has links to the version that's included. So look at: https://github.com/adafruit/Adafruit_CircuitPython_Bundle/releases/tag/20171019. Then you'll see https://github.com/adafruit/Adafruit_CircuitPython_CircuitPlayground/releases/tag/0.8.1. Click on that, then on the page that comes up, click on the little 0.8.1 to the left of the vertical line. Then you can look at the source for the 0.8.1 tag.
tidal kiln
@tulip sleet thanks. that's a better way to nav there.
next question would be how to get to the c code from there?
slender iron
@tulip sleet try a critical section around https://github.com/adafruit/circuitpython/blob/master/ports/atmel-samd/usb_mass_storage.c#L309
@tidal kiln find the code in shared-bindings that matches the module you are using and then find the common-hal code it calls
opaque patrol
I think I see the problem...
sine_wave is initialized to None self.sample.frequency = int(len(self.sine_wave) * frequency)
so it might be possible this is multiplying frequency by 0 which equals 0 and would result in the error message "Unsupported playback frequency"
stuck elbow
TypeError: unsupported operand type(s) for *: 'NoneType' and 'int'
TypeError: object of type 'NoneType' has no len()
tulip sleet
@slender iron critical section fixed it! seems to be ok on windows now
(I was doing some other paperwork.)
tulip sleet
does the other mscdf_xer_blocks at line 312 also need guarding. it does a final mscdf_send_csw()
slender iron
yeah maybe
on the beagle I'm seeing it respond with extra read blocks I think
and that extra read block is sent as a response to something thats not a read
opaque patrol
@tidal kiln I think the max value for frequency in start_tone is 3499 , I don't have an express yet to test myself.
split ocean
NeoPixel on CPX question:
do we have an equivalent of setting pixel colors and then "showing" them all at once like in Arduino when using the express cpx.pixels method?
I'm trying to change all colors at once, instead of w a slight chase effect I'm getting iterating through them
stuck elbow
Yes, disable autoupdate when creating the NeoPixel object
lofty topaz
@stuck elbow Bingo!
Was just about to....
split ocean
since I'm doing this w express I'm not explicitly creating the NeoPixel object.
stuck elbow
I guess that forces you to create it yourself instead of using cpx
split ocean
crossed streams :)
I'm doing this for a Learn Guide, and gonna try to keep it as simple as I can
stuck elbow
sorry for slow typing, I'm on a screen keyboard
stuck elbow
you might be able to disable autoupdate on cpx.pixels
split ocean
I know i can do cpx.pixels.brightness = 0.2 to change the object
stuck elbow
Just look at the source to see how the attribute is called
opaque patrol
@split ocean you could try setting the neopixel auto_write to false
split ocean
@opaque patrol that's what I'd like to do, and then after setting all their colors do a write/show. just need to figure out the syntax!
slender iron
cpx.pixels.auto_write = False cpx.pixels.show()
split ocean
thanks @slender iron !!!
slender iron
pixels is just a NeoPixel object thats already made
split ocean
I dig it
solar whale
@tulip sleet thanks for the M4 eraser! It'll be a lot easirer to recover if I can recreate the issue I had last night!
manic glacierBOT
slender iron
@solar whale I think I found and fixed your issue
I was copying the bundle over and then deleting it to stress test it
slender iron
just testing it one last time
solar whale
no rush!
slender iron
I did get file corruption like you did
tulip sleet
@solar whale I tested the erase .uf2's on Gemma, but not on m4. Something not working on m4, so hold on.
solar whale
@tulip sleet thanks for the warning.
slender iron
pushed a second commit to https://github.com/adafruit/circuitpython/pull/425
tulip sleet
@slender iron it's still pretty flaky on Win7 and Win10. Sometimes doesn't show name, just "Removable Disk". Try it yourself. I had to reboot to clear some bad state. Very slow to appear if at all. Try to compare it with the reverted version from my pull request
btw, I cloned tannewt/circuitpython to build. Surprised that repl said "0.9.0-xxx". Your repo is missing a lot of tags. Maybe do "fetch --tags"?
slender iron
Can you post some traces? I’m finishing up lunch
tulip sleet
we have to go out in 5 mins or so, I'll be back around 8:30ish ET
slender iron
kk
pulsar bloom
Maybe a little off topic, trying to program CPE on arduino. Red led breathing, see cpe on com12, verbose turned on and arduino runs F:\Users\ritas\AppData\Local\Arduino15\packages\arduino\tools\bossac\1.7.0/bossac.exe -i -d --port=COM12 -U true -i -e -w -v F:\Users\ritas\AppData\Local\Temp\arduino_build_274068/cp-blink.ino.bin -R
com12 times out
Tried running bossac from command line with same result
slender iron
@solar whale ping me if you get a chance to test the pull request
tidal kiln
@pulsar bloom try again after double clicking the reset button to put the CPX into bootloader mode. may need to check the com ports in the arduino IDE and re-pick it.
https://learn.adafruit.com/adafruit-circuit-playground-express/set-up-arduino-ide#manually-bootloading
solar whale
@slender iron I should be available in an hour or so.
slender iron
ok thanks @solar whale
idle owl
@slender iron What do you need tested
slender iron
pulsar bloom
@tidal kiln tried all possible interations of button clicking, multiple usb cables, multiple usb ports, multiple cpe it all seems to point to bossac not working.....
idle owl
@slender iron How do I build from something still in a PR?
tidal kiln
@pulsar bloom probably best to post in the forums then.
https://forums.adafruit.com/viewforum.php?f=58
pulsar bloom
Thanks suspect win10 problem
slender iron
you can add my repo as a remote or I can just post a build for you
idle owl
Ooh. Build for me. I want to figure out adding yours as a remote but not right now.
My friend is currently sewing neopixels at the moment so I have bit before back to teaching. She wrote her first CircuitPython today! And when she tried just threading them together without fabric, nearly smoked one of the NeoPixels. Been a day filled with firsts! 😄
slender iron
metro m4 test
lofty topaz
Hello all who flip bits. Kibitzin' again. Amazing stuff. Quick Question? How in the world do you keep from breaking the build? That was sacreligious in my day. Saw developers fired on the spot for breaking the build.
tidal kiln
@lofty topaz break in what way? by changing source code?
lofty topaz
@tidal kiln By checking in code that does NOT allow the build to perform. You never ever break the build.
A syntax bug.
I always wondered how "open source" kept from breaking the build.
tidal kiln
@lofty topaz in general, by forking the working code and modifying that. then merging it back in to working (main) branch is done once reviewed by those who will make sure it doesn't break. hopefully.
idle owl
@slender iron Hmm. Am I testing the USB issues with this update? Or something else... Because it's still acting weird.
lofty topaz
@tidal kiln Sounds like a job for Superman. 😃
idle owl
On the mac first. It mounted ~10 times in a row after the reset button was hit, on the next one, didn't want to mount. Physically unplugged it, Finder complained about circuitpy disconnecting (even though it wasn't mounted). Windows USB3, nothing, Windows USB2, still 2 unnamed drives.
tidal kiln
@lofty topaz the tools help a lot. also the fact that we are not pushing live code to a mars rover or something.
lofty topaz
Curious tho, in this day and age, do your builds take 3 to 4 hours like ours did over night?
@tidal kiln Oh yes, the tools are most helpful. I wish I had the tools back then that we have today.
slender iron
@idle owl so you are plugging it in and then hitting reset after it shows up? are you ejecting?
@lofty topaz they take 10 -15 minutes and are run by Travis CI
lofty topaz
@slender iron Now that's what I call a luxury.
slender iron
@lofty topaz thats slow too. we'd love it to be faster
umbral dagger
Works been intense this week, but I plan to upgrade to the latest CP3.0 asap and start poking at sensors and such.
I'm also aiming to get home in time for Show&Tell tonight.
lofty topaz
@slender iron Patients, have patients. The time will come when it takes no time at all.
I save to the f: drive and walla, the CPX is running it. In seconds.
And to think I used to wait for 30 minutes while the EEPROM bathed in UV light before dumping hex code into it. Then plugging it into the S100 board and hoping the Z80 based system will run.
Yeah, I'd say things are a tad easier these days.
slender iron
😄 I'm happy the iteration cycle is a bit faster now
stuck elbow
Nice. I used to compile the kernel for a week on my 386...
Kids have it easy today 😉
lofty topaz
I once esent a 386 to our plant in Mexico and an FBI agent called me and said that technology was too advanced to export. Your under arrest. 😃
We worked it out but my boss wasn't too happy.
stuck elbow
I recently had to file some documents to be able to order an infra-red sensor from mouser
because they are "restricted export" and they wanted to know exactly what it is for
lofty topaz
@stuck elbow I can believe it. Do the paperwork. It pays.
stuck elbow
funny thing is that Adafruit is selling them on breakout boards and you don't need any documents for that
it was the AMG8833
idle owl
@slender iron I was ejecting every time.
tidal kiln
@stuck elbow did you see the post about the AMG being mis-classified?
stuck elbow
I have not
tidal kiln
(hmmm maybe that wasn't posted)
umbral dagger
@stuck elbow I bought some raw AMG8833s from Digikey and had to do the declaration. I believe that requirement got dropped just after I got them.
stuck elbow
nice, good to know
umbral dagger
MetroM4 is happily updated to master
tidal kiln
stuck elbow
thanks, that's interesting
I was trying to imagine what they think you could possibly build with this that would be dangerous, that you couldn't build with a bunch of phototransistors
heat-homing systems come to mind, of course
idle owl
Also after sitting for a bit, it froze Finder. Trying the second build you sent.
So here's the other thing that happens whne it starts acting odd and doesn't mount: the main.py that's running loads and runs really slowly.
It's a rainbow animation and it goes from zipping along to a really slow fill and slow rotation.
But that's even only 1 in 3 times that it didn't mount. And this time, at least if I reset it again, circuitpy shows up. Before once it did that, it wouldn't show up again at all until it had been physically removed.
stuck elbow
I noticed that even on 2.x Trinket M0 runs slower when connected to USB
especially when doing anything cpu-intense
idle owl
Couple of things that look like errors in dmesg: disk3: media is not present. disk3: media is not present. disk3: media is not present. disk3: media is not present. disk3: media is not present. msdosfs_fat_uninit_vol: error 6 from msdosfs_fat_cache_flush USBMSC Identifier (non-unique): AF8C5FF815E47535020202D473C480FF 0x239a 0x8021 0x100, 1
USB device 239A802114130002 - fConsecutiveResetCount = 1.
disk3s1: device/channel is not attached.
USBMSC Identifier (non-unique): AF8C5FF815E47535020202D473C480FF 0x239a 0x8021 0x100, 1
SandboxViolation: AssetCache(18795) deny(1) file-read-metadata /Volumes
USB device 239A802114130002 - BOT reset completed with kUSBHostReturnPipeStalled.
USB device 239A802114130002 - fConsecutiveResetCount = 1.```@slender iron I can test it now - unless it has already been done.
idle owl
Most of the rest of it is the same thing you've told me previously was fine.
@solar whale Test it. I'm still having issues with it. Could be on my end.
solar whale
@slender iron is it inthe file you posted above or should I download it.
idle owl
The first uf2 he posted is the one to try first.
slender iron
@idle owl was the latest uf2 better than the first?
I'm seeing the issues here too
@idle owl and @stuck elbow main gets slowed down by usb interrupts I think
idle owl
@slender iron The latest was much smoother for me.
slender iron
hrm, smoother how?
idle owl
It mounts more consistently, when it stops mounting it at least mounts again after a reset or two, before it didn't. It did freeze Finder once, but mounted immediately again after unplugging it physically.
slender iron
hrm
solar whale
@slender iron wa thsi the correct version to test ?? ```Adafruit CircuitPython v1.9.3-1401-g971bd7e9a on 2017-11-08; Metro M4 Express with samd51j19
Oh.
When I plugged it into windows, the onboard NeoPixel went white.
Plugged back into Mac, and the system's been erased. No files.
slender iron
hrm
idle owl
Mounted once as circuitpy. On reset: disk was not readable. Took a ton of resetting to get it to mount as circuitpy again, still empty though. Only has boot_out.txt.
slender iron
are you copying files to it?
solar whale
@slender iron I just overfilled it and it responded correctly - no space on device - recoved fine
idle owl
I wasn't at the time of any of this happening. Just ejecting and mounting to it.
solar whale
INterstingly, the whole lib/ actually fit - had to copy it twice to fill it
slender iron
hmm, sounds troublesome still
solar whale
@slender iron why the 1.9.3 tag?
idle owl
Tried copying another file to it with a main.py already there that wasn't working, Finder is frozen.
slender iron
its the latest tag (from micropython)
solar whale
ok - it seems to handle overfilling the FS well - at least for one test
solar whale
who?
idle owl
Mounted 20 times, didn't mount one time, but mounted again after reset on that one. Some error in my main.py apparently, but eh.
slender iron
I think the blinky error code is wrong when main is missing
idle owl
This seems to be working better than before.
solar whale
also works fine for me - reported no space on devive - recoved fie when I made space
idle owl
Oh. Also, it erased the code again.
slender iron
k, thats with the revert that dan proposed
idle owl
So there's no main.
slender iron
was there ever a main?
solar whale
so far it appears to be automounting much more reliabley on linux.
nevermind - ppushed my luck -- not always automounting, but working fine.
idle owl
There was before, but it's since been erased.
Lost some new code too, lol. Should have known better.
stuck elbow
I try to never edit the code on the board.
Always edit it in the repo on my disk first, and copy over.
Unless it's some trivial experiments, of course
idle owl
Yeah I'm usually better about it. But I get complacent over time, and every so often, lose something. Then I'm on it for a while again, repeat.
solar whale
I have learned never to assume any system will come back after a reboot ... too many times
I haven't learned to proofread my typing....
@idle owl waht is the issue you are testing?
idle owl
@solar whale USB stability. It just also erased my board in the meantime.
solar whale
😦 any particular scenario you want me to try to reproduce?
idle owl
I was just ejecting and then pressing the reset button repeatedly. 😄
idle owl
I had one on there at the start.
Then after doing all the resetting, it erased at some point.
Then I added another one.
But there was error status lights when Finder said the board was empty
solar whale
Is teh issue only on windows machines ?
solar whale
one eject reset cycle OK on my Mac
slender iron
I need a medieval break from usb
idle owl
@slender iron Sound reasoning, good sir.
slender iron
😃
idle owl
😉
solar whale
@slender iron at least the FS is respnding well to overfilling now. Thanks!
slender iron
if its ever been solid on usb it'd be good to know what commits it is
solar whale
both have worked for me - now using Adafruit CircuitPython v1.9.3-1401-g971bd7e9a-dirty on 2017-11-08; Metro M4 Express with samd51j19
I had some funky USB issues last night --- dropped connections and the FS issues - None so far this evening.
slender iron
this is the code for that last uf2: https://github.com/tannewt/circuitpython/tree/align_usb
solar whale
is it hte align_usb branch?
slender iron
yeah
solar whale
thanks
slender iron
no problem!
solar whale
@slender iron is the cache enbale or disabled with the setting to 1?
slender iron
disabled I believe @solar whale
solar whale
OK - not intuitive 😉
solar whale
I guess its a "disable" flag, not an "enable" flag - sigh...
solar whale
OK thanks - will take a look ...
royal ridge
Is upgrading circuit python to 2.10 going to wipe flash on CPX?
idle owl
It shouldn't, but it's always good to backup your code.
royal ridge
Hrm. also, the install section directs to 2.0... should it direct to 2.1?
idle owl
Can you post the link please?
royal ridge
tidal kiln
@royal ridge weird. that link shows 2.1 for me.
slender iron
@royal ridge I fixed the download uf2 link
royal ridge
I ❤ you @slender iron
idle owl
Thank you @slender iron
tidal kiln
well. if it's fixed, it's fixed. nvmd.
solar whale
@slender iron Do you have a link to the schematic for the Metro_M4 board?
slender iron
@solar whale its not public yet but I have a copy
solar whale
ah - ok - just curious
slender iron
do you have a specific question?
solar whale
After reading the errata, I was wondering if there was an exteranl pullup on the RESET - I see one on the Metro_M0 so I assume it is on the M4.
my board is under my metrowing - so I can't look at it 😉
solar whale
thanks - I'll verify next time I take the wing off.
slender iron
😃
solar whale
ugh - forgot how small the traces are - I'll take your word for it!
really enjoying the RAM on the M4 - I't's great to be able to use the OLED display and be able to runs sensors at the same time! the Metrowing has been a great tool for plugging in boars to test.
slender iron
yeah, I think the hardware errata is dealt with
we may be able to increase the heap size too
idle owl
Science cannot move forward without heaps!
solar whale
I have a test ruunnig with 2 Temp sensors displaying temperatures and gc.mem_free on the OLED - it cycles from > 114K down to 15K as it runs
slender iron
😃
drowsy geyser
FYI, I just built new firmware and flashed the M4. Still can't get to the REPL and no mass storage mount.
slender iron
@drowsy geyser from master? did you see my branch go by?
drowsy geyser
Yessir, from Master. I did not see your branch (sorry). Been in a conference all day.
Just got home and tried the build... 😃
slender iron
drowsy geyser
Ahhh, thank you, Scott!
slender iron
np
tidal kiln
@slender iron just fyi, i closed ads1x15 issue
slender iron
thanks @tidal kiln !
tulip sleet
@slender iron should I try latest from align_usb branch on Windows?
slender iron
if you like, I still don't think its fixed
been just getting frustrated with it
gonna make this trip a vacation from usb
idle owl
Understandably.
tulip sleet
that's a great idea. The only reason I submitted the PR this morning was to get Windows working again. Maybe if we just take that for now? and you can come back refreshed
slender iron
yup, thats fine with me
feel free to have other folks give you reviews while I'm travelling
tulip sleet
maybe I will add the alignment fixes to that PR, since those are clearly issues, and then leave it at that. I plan to work on the other busio stuff, and that doesn't need merging soon at all. I am ordering a some breakouts for the other protocols to test (e.g. OneWire.) Anything urgent that comes up I will get someone to test and review
slender iron
kk, sounds good
tulip sleet
Go eat dinner and pack and relax before your fliight tomw
slender iron
😃 I have chores to do. my mom is staying at our house so I have to do some cleaning. 😃
idle owl
Kitties will have a friend though!
tulip sleet
catsitter
idle owl
Works out all around
tulip sleet
I had a lot of coding epiphanies while driving home an hour from Palo Alto to Berkeley instead of being poised over the keyboard. And then I did something else entirely, cause I couldn't work from hoem.
idle owl
Those are great!
tulip sleet
actually I hope you forget about it for days or at least hours at a time. bye - happy travels!
idle owl
I agree with Dan! Have a wonderful time!
slender iron
thanks all!
solar whale
@slender iron have a great trip !
umbral dagger
Loking ahead to the weekend my goal will be to get a rudimentary debugger running on ATSAMD CP (I was experimenting in the Unix port).
idle owl
@umbral dagger Nice!
idle owl
@tulip sleet I've had the M4 serial connection up, the rainbow animation main.py running a ring and a strip, and main.py open in Atom since before the live streams and they're all still running smoothly. I don't think it made it this long earlier.
I'm going to leave it overnight.
pastel panther
@idle owl You were having mac problems with the m4, right? Is there something that I should/could test on mine?
idle owl
@pastel panther I was ejecting it and pressing the reset button. Repeatedly. Eventually it won't mount, or in a couple of cases, erased the board. A couple of days ago I managed to corrupt it bad enough that it would only show up as NO_NAME - that took the JLink to recover from. But the issues I had earlier were simply from ejecting and resetting. Finder would freeze completely and I would have to physically remove the plug to get everything to work again.
idle owl
Then about half the time, when I plugged it back in, it would take resetting 10 times to get the drive to mount.
pastel panther
How fast were you reseting? LIke 1hz or so?
idle owl
The build I have on there now seems to be much more stable than it was earlier.
pastel panther
Is that the build from Scott's fork
idle owl
I waited until CIRCUITPY disappeared from Finder before resetting. To give it more of a fighting chance, I would give it a couple of seconds after that for some of the testing.
Yeah, he sent it to me, so it's from his repo.
I don't know what's been committed to master since.
pastel panther
Ok, I'll test it out after I'm done tearing down this TV (yay dumpster gold!)
tulip sleet
the commit above is on master and rolls things back a bit so that Windows is usable again, and also includes one obvious fix. When Scott gets back we can try some more stuff.
tulip sleet
I wanted to test Scott's latest on Windows but my test Windows box was tied up with an update. Will try again in a a few minutes.
idle owl
Ooh! Shows up on Windows 10 USB2 and USB3, immediately.
tulip sleet
that's good - I'll be back in a minute after rebooting this machine, which also has windows
idle owl
I need to head out. Goodnight!
tulip sleet
gn
align_usb branch is ok too.
<@&356864093652516868> if you wish, try the latest adafruit/circuitpython master, and also https://github.com/tannewt/circuitpython, branch align_usb. Let me know if one seems better than the other and I'll make that one be on master for the next days. @drowsy geyser both seem OK on Windows for now, but only lightly tested
the odd v1.9.3 version you see on the align_usb version is just due to merging from micropython. The version uses the latest tag available. I retagged in master so it's back to 3.0.0-alpha.something
drowsy geyser
@tulip sleet Maybe I'm doing something wrong (likely), but I did a git pull git submodule update --init --recursive and flashed the new firmware (I did see your new usb module go in). I'm not seeing any behavior change.
So Master is still not working for me.
Do I need to delete and re-clone the repo?
solar whale
@drowsy geyser Did you also try the align_usb branch from https://github.com/tannewt/circuitpython ? It has one change in it.
drowsy geyser
@solar whale I haven't been able to get it to build. I get tons of errors on make. I'm pretty sure I built my source tree wrong, but I haven't gone back and tried to fix it yet. I just did a git clone from that repo, followed by a submodule update. The build environment looks totally differnt.
solar whale
ddi you do git checkout align_usb before submodule update and build?
solar whale
git clone https://github.com/tannewt/circuitpython
cd circuitpython
git checkout align_usb
git submodule update --init --recursive
cd mpy-cross
make
cd ../ports/atmel-samd
make BOARD=metro_m4_express clean
make BOARD=metro_m4_express
Thanks! Added to my OneNote. 😃
solar whale
the alighn_usb branch is just for this test - with mainline repo always checkout master.
drowsy geyser
Yup! I built them in separate directories.
solar whale
I finally made separate dirs for master, 2_x- it gets really messy switching between 2.x and master since the layout is differnt.
drowsy geyser
@solar whale Results: Scott's branch works fine, whereas Master does not. I'm assuming someone will do a merge at some point to get Scott's changes back into Master?
solar whale
yes @tulip sleet is collecting input - please update him. Great news!
drowsy geyser
Ok, will do, @solar whale . Thank you again for helping with my newbie build problem!
solar whale
Glad to help but somehow newbie seems a bit misplaced for you!
drowsy geyser
LOL. Newbie to GIt! I have to get up to speed since we use it at work for everything....
solar whale
It rminds me of "learning" emacs -- always seems like I'm just scratching the surface!
drowsy geyser
Totally with you on that. I'm a dyed in the wool vi guy.
Started using it in '85 and it's just ingrained. I can't tell you how many times I've typed ":w" in Word just from muscle memory....
Dan updated in DM. Now the question is, "do I sleep before the conference, or solder up a BME280 and see if it works on the M4?"
solar whale
My BMP280 works fine!
drowsy geyser
Well then, no need to test it. Lessee, what else do I have here?
solar whale
you will need to regenerate the .mpys in the libraries you use - moy format changed in master
I tried mcp9808 and an OLED ssd1306 - all good
drowsy geyser
Hmmm, I have a TSL2561 here. I think someone already did the driver for that....
solar whale
the extra RAM is amazing - nice to have room for drivers and code!
Its in the bundle now - with the extra RAM you may be able to test with the .py insteasd of the .moy
drowsy geyser
Yeah, I can't wait for these to get into production! I really want to use CircuitPython in my rocket flight controller, but the libraries wouldn't fit in it before. 😃
solar whale
THis is the first time I've been able to use an OLED display on CP and still have room for drivers.
drowsy geyser
Nice! I definitely want to have the OLED on my systems. I use them a lot!
solar whale
neopixels work too! It's great to have all these new features working on the M4 - the update speed for the neopixels is great!
drowsy geyser
I have a pile of stuff I need to solder up and there's a NeoPixel ring in the queue. I definitely want to try it on the M4!
solar whale
I have done a few tests with a 16 pixel ring and it is very nice - the "color wheel" demo I have been using takes 16 secs to go areound the loop on the M0 - about 6 on the M4!
drowsy geyser
Wow! That's a significant improvement!
tulip sleet
@drowsy geyser I got your results. Could you do do git log in your adafruit/circuitpython repo? It should look like this:
commit b82cfcb064e7a01548abcde5983b6d4be6fd9e9c (HEAD -> master, upstream/master)
Merge: b4fc464a4 0e6928b08
Author: Dan Halbert <halbert@halwitz.org>
Date: Wed Nov 8 23:38:13 2017 -0500
Merge pull request #424 from dhalbert/revert_182a946_msc
Revert "atmel-samd: Tweak mass storage code to finish read sooner."
commit 0e6928b087805f95520de34647d2836278fcdbf9 (tag: 3.0.0-alpha2, tag: 3.0.0-alpha.1, origin/revert_182a946_msc, revert_182a946_msc)
Author: Scott Shawcroft <scott.shawcroft@gmail.com>
Date: Wed Nov 8 10:47:22 2017 -0800
atmel-samd: Align SCSI responses.
commit bfe7e055e9b1dac0411fe975fdf2412029f03b33
Author: Dan Halbert <halbert@halwitz.org>
Date: Wed Nov 8 11:17:10 2017 -0500
Revert "atmel-samd: Tweak mass storage code to finish read sooner."
This reverts commit 182a946b0ef12f6f031e846af788cdeb2eecd3a6.
@tulip sleet That's is what it looks like word for word. I can paste it here if you like... (Sorry it took so long to respond; I was fussing with some sensors and didn't notice the message)
opaque patrol
Its quiet today....is it like this every thursday?
tulip sleet
@opaque patrol in the morning it's usually quiet
@drowsy geyser ok, I'll bring align_usb into master today. I tried the master HEAD on Win10 last night, and it seemed to work OK, but there may be other issues.
tidal kiln
@tulip sleet for testing non-master branches, do we need to have the ability to build locally?
tulip sleet
@tidal kiln It looks like the pull request builds aren't saved, so yes, you need to be able to build locally. Do you have a setup or do you need a recipe? If the latter, what OS's do you have available?
tidal kiln
@tulip sleet thanks. no don't have a setup yet, but was planning on doing that here at some point. i'm on linux, so should be pretty easy. i did build mpy-cross, but that was just a matter of cloning the repo and running make.
tulip sleet
@tidal kiln
sudo add-apt-repository ppa:team-gcc-arm-embedded/ppa
sudo apt update
sudo apt install gcc-arm-embedded python build-essential
git clone https://github.com/adafruit/circuitpython.git
cd circuitpython/
git submodule update --init --recursive
cd ports/atmel-samd/
make clean BOARD=metro_m4_express ; make BOARD=metro_m4_express # or whichever you want
For Scott's repo and this branch:
git clone https://github.com/tannewt/circuitpython.git tannewt-circuitpython # or wherever you want it
cd tannewt-circuitpython
git checkout align_usb # get to the right branch
git submodule update --init --recursive
[ same as above]
sweet. thanks. why the need for adding the ppa?
tulip sleet
to get the arm gcc compiler, and keep it up to date. Better source than downloading manually from ARM.
tidal kiln
and the git submodule step just brings in latest of all the submodules, which aren't really part of the main repo?
tulip sleet
right - they are in separate repos. For instance, we put asf4 in a separate repo since it might be shared with other projects, and we might want different snapshots of it depending on the progress of a particular port.
tidal kiln
neat. pretty straight forward. and the non-main branches are being held in personal repos? that's why the second one comes from scott's?
tulip sleet
we do development and bug fixes in our private repos so we don't clutter up the main repo with temporary branches or mess it up by accident. When we're ready for something, we submit a PR and ask for a sanity-check review.
you have to set up your private repo with adafruit/circuitpython as a remote so you can re-sync as necessary. It takes a bit of time to learn the worflow
tidal kiln
gotcha. are the ones in the s3 bucket basically the same as what i'd get from the first recipe?
tulip sleet
Yes, they're the same
since they're built in travis by cloning adafruit/circuitpython and doing the build pretty much the same way you would
tidal kiln
so really only need to build for testing latest from the private / in-work repos, otherwise just grab from s3
tulip sleet
yeah, though you have to make sure you have the right commit. Since I'm debugging all the time, I don't bother with the aws repos.
tidal kiln
cool. just got a successful build from first recipe.
tulip sleet
great! if you have a multi-core processor, you can try make -j4 or however many cores you have. It will speed things up somewhat.
right now a build takes me about 45 seconds for 3.0. 2.x is more like 20-30 secs
idle owl
@tulip sleet M4 still connected to serial, still running main.py still able to interact with files using Atom, still mounted in finder and not frozen. This is from last night.
tidal kiln
yeah. that's about what it looked like. i got a x4 core, so will try that next. i'm going to see if i can build scott's branch...
tulip sleet
@idle owl thanks! and that's align_usb, right? I am probably going to bring master up to date with that
idle owl
How can I tell for sure? I'm pretty sure, yes, but I'd rather be certain.
It hasn't lasted overnight connected like this in days. This is a huge improvement.
tulip sleet
what's the git id that's shown in the initial prompt in the repl -g<something>
idle owl
Adafruit CircuitPython v1.9.3-1401-g971bd7e9a-dirty
tulip sleet
that's one commit behind what's in align_usb now: https://github.com/adafruit/circuitpython/pull/425/commits: see what's on the right-hand column
idle owl
I am using the last thing Scott posted to me last night. I don't have his repo built locally yet so I can't build it myself.
Do you want me to test the current commit?
tulip sleet
could you clone his repo and build? see recipe I gave cater above. thanks. I'm not sure about the ef9a87f2 commit. I thought it wasn't working well on Windows, but @drowsy geyser said it worked for him
idle owl
This commit worked perfectly on Windows for me yesterday. I'll test the current one. (Might still need help, but I'll try the recipe first.)
Only the second thing you sent right? Because I would already have ppa from cloning the master repo?
tidal kiln
yep
idle owl
Thank you
tulip sleet
you just enable the ppa and load the arm gcc stuff once. It gets installed in /usr/bin, etc. It's not related to the repo
tidal kiln
you don't need to reinstall the tool chain, but that's not part of the repo
tulip sleet
great minds think alike
tidal kiln
boom. got a build of scott's repo.
idle owl
I can already build firmware in the circuitpython/ repo... I assume I've done the first recipe before. It was over that weekend when I got the M4, so ... been long enough that I wasn't sure. But realised you hadn't built anything yet, so that's why I verified.
@tulip sleet Why make clean and not make BOARD=etc clean ? I've been running it with the board in the command every time.
tidal kiln
do both. clean just deletes everything so you're starting from a known ground zero state.
idle owl
Ah.
tulip sleet
you're right, that's a bug, you have to specify the BOARD for both
I edited the recipe above
tulip sleet
make clean without a BOARD spec uses the default board.
tidal kiln
thanks. was just about to look in makefile...
tulip sleet
if you want to use jlink to look inside and set breakpoints, etc. then DEBUG. Otherwise don't bother, because it's a better test of what will be shipped
idle owl
Ok, sounds good. I'll do it without.
tulip sleet
DEBUG adds some additional info to the compiled program to make life easier for the debugger. In some cases it also changes the optimization level
idle owl
Makes sense.
tulip sleet
compiling without optimization makes debugging easier because the code doesn't get reordered, variables don't get optimized away, etc. But then the code is slower and larger. For M0, it's usually too big to fit in flash
idle owl
So not a problem here. But it's not an accurate test either.
tidal kiln
here's what i get from scott's build:
Adafruit CircuitPython v1.9.3-1402-gef9a7f2 on 2017-11-09; Metro M4 Express with samd51j19
>>>
(or more specifically, my build of scott's latest)
tulip sleet
there can be issues @tidal kiln that's good, that's the latest commit on that branch
tidal kiln
neat. thanks for getting me running on this.
tulip sleet
yw
tidal kiln
knowing about that ppa probably would've been half a day of googling
tulip sleet
it's confusing; the ppa has gone through several incarnations
idle owl
Alright. Loaded, mounted immediately, main.py running smoothly. Adafruit CircuitPython v1.9.3-1402-gef9a7f262 on 2017-11-09; Metro M4 Express with samd51j19
tidal kiln
weird. slightly more verbose?
idle owl
I have 2 more digits in the build number.
tidal kiln
yeah
idle owl
Hmm, I closed the window while in the REPL, and now it's acting like it's still connected to the REPL (white onboard LED, main.py not running). I can't reestablish a serial connection.
tulip sleet
that string is generated by a git command: git describe --dirty --always --tags
tidal kiln
so some OS build environment dependency?
idle owl
I might have something in .bashrc that's altering my default git stuff. A friend of mine magicked it up for me with some git magic, could be that. Don't know if that would affect it.
tulip sleet
could be a bug; is that different from the build you tested overnight? If so add a comment to the PR
git --version
idle owl
git version 2.14.3
tidal kiln
git version 2.7.4
idle owl
@tulip sleet I don't know if I tried this exact sequence with the other build. But it's consistently repro'd.
Can you link the correct PR please. I had it up yesterday but I don't think I do anymore
idle owl
Ok. Be a minute.
tulip sleet
@tidal kiln I think it was lengthened in later versions: git describe --dirty --always --tags gives me v1.9.3-1402-gef9a7f262
idle owl
Same thing.
tulip sleet
ok, so not a very new problem
idle owl
I have three more of Scott's builds from yesterday, I'll try those as well
tulip sleet
back in a few mins
tidal kiln
@tulip sleet yep. that's it. i get:
v1.9.3-1402-gef9a7f2
Same in the third one too.
If it's unrelated to the current PR, should I file an issue instead of commenting on the PR?
The third one froze everything too, so there's still a marked improvement over Scott's earlier commits.
tidal kiln
how do you get it to freeze? closing while in REPL?
idle owl
No, on the earlier commit as soon as I try to interact with CIRCUITPY by clicking on it or attempting to eject it, Finder freezes until I physically remove the M4 from USB.
tidal kiln
so this v1.9.3-1402-gef9a7f262 build shouldn't have an issue?
idle owl
Correct. I tried ejecting and resetting a bunch before leaving it over night, and it didn't freeze while actively testing it, and it was still connected successfully when I checked today.
tidal kiln
ah. so i'm late to game and only confirming it's fixed?
idle owl
Oh! lol. Yes 😄
tidal kiln
cause it seems to be running fine
idle owl
This build should be working across the board, Linux/Mac/Windows, USB2/3, REPL connect/disconnect, physical power off/onboard reset button.
Theoretically 😄
tidal kiln
problem shows up in current main build though? i'm seeing Dan's comment:
"Let me know if one seems better than the other and I'll make that one be on master for the next days"
idle owl
Yeah. Master is failing quite a few things.
tidal kiln
ok. let me try that.
idle owl
Even with Scott's last 4 commits, at least for me, the current one was the only one without issues. I can post the ones Scott posted yesterday if you want to test those too. (I definitely don't know how to build from different commits.)
tidal kiln
does it matter? or is the only concern comparing scott's latest to current master?
idle owl
I don't know. I think they wanted to test the few changes Scott made to each other because each of them had one tweak related to stability. I'm not sure though. @tulip sleet Should @tidal kiln test each of Scott's builds from yesterday or just master vs current build?
tidal kiln
ok. i'm on current master/g. seems to work ok. got a recipe to break it?
idle owl
Hmm. For me it was interacting with it by editing my main.py in Atom (it froze Atom pretty consistently) and in the file explorer. Ejecting and reset using onboard reset usually made it pretty unhappy pretty quickly.
By trying to open it or eject it from the file explorer, it would freeze Finder and required physical removal to get it back. Reset using the onboard reset button instead of powering off caused issues.
tidal kiln
k. let me try changing blink rate in atom...
idle owl
Wouldn't even mount to Windows at all. Showed up as 2 unnamed and untouchable drives.
tidal kiln
working. for ref, i'm running:
Adafruit CircuitPython 3.0.0-alpha.0-1157-gb82cfcb on 2017-11-09; Metro M4 Express with samd51j19
>>>
I don't know how anyone got it to break on Linux. I updated my Linux machine but haven't set it up for testing yet. I know Dan uses it and he was having issues. I don't know what he did to get them though.
tidal kiln
hmmm. ok. will keep punching on it and see if anything odd happens...
discord apparently accepts sed find and replace commands....THIS IS A CARROT
neat
tulip sleet
back: so I'm not sure, you're saying the latest on align_usb is the most stable? I did the rollback after a lot of testing on Windows becuase Hukuzutuna said it was broken. As for the REPL bug, yes, a separate issue would be good.
tidal kiln
@tulip sleet for me, not seeing an issue with either
tidal kiln
linux
tulip sleet
yes, linux has been fine. It was Windows that was wonky
tidal kiln
:tell_me_something_i_dont_know_emoji:
tulip sleet
i'll try both master and align_usb on windows myself. just haven't had time to repeat the experiments. I have slow spare windows box that was updating late last night
idle owl
For me, align_usb has been the most stable. Carter and I were just figuring out what he should be testing.
tidal kiln
@tulip sleet on another note, how about this behavior, i mentioned it before, not sure if it got noticed:
Adafruit CircuitPython 3.0.0-alpha.0-1157-gb82cfcb on 2017-11-09; Metro M4 Express with samd51j19
>>> import time
>>> t1 = time.monotonic(); time.sleep(0.1); t2 = time.monotonic(); t2-t1
0.0991211
>>> t1 = time.monotonic(); time.sleep(0.01); t2 = time.monotonic(); t2-t1
0.00878906
>>> t1 = time.monotonic(); time.sleep(0.001); t2 = time.monotonic(); t2-t1
0.0
>>>
curious about both the accuracy and the apparent lower limit
tulip sleet
@tidal kiln each time.monotonic tick is a millisecond. But that's just our implementation - there's no guarantee
that it's any particular unit
tidal kiln
is there something higher resolution?
tulip sleet
time.sleep also uses msec ticks to count, but hold on a sec ...
you can use microcontroller.delay_us(n), but that does a busy-wait loop, I think. Still looking at the code
yes, it's true
tidal kiln
meaning it blocks?
tulip sleet
it's just a assembly loop, but it doesn't seem to disable interrupts, I think. The actual routine is about six calls deep.
this is in 2.x. Not sure what is in 3.x (if anything yet).
do you need accurate timing for a driver?
the time to execute a single python statement could be significant and swamp any delay you insert
tidal kiln
not yet
ran into the behavior porting some arduino code that had a delay(1) in it. turned that into time.sleep(0.001) but got no delay.
tulip sleet
that will actually call the millisecond sleep routine with an argument of 1
tidal kiln
ok. then why 0?
tulip sleet
no, wait, it could truncate, because it does 1000*0.01, and that gets cast to an int, so if it's less than 1.0, it will truncate to 0.
tidal kiln
but 1000*0.001=1.0, so why does it trunc?
tulip sleet
4.76837e-07
it's slightly less than one: difference is >0
you could file an issue saying it should round, not truncate
tidal kiln
int(round()) instead of int() ?
tulip sleet
https://github.com/adafruit/circuitpython/blob/master/shared-bindings/time/__init__.c#L76 calls https://github.com/adafruit/circuitpython/blob/master/ports/atmel-samd/common-hal/time/__init__.c#L37
an implicit cast happens, it looks liek
tidal kiln
from 1000 * mp_obj_t -> uint32_t
tulip sleet
not mp_obj_t: it gets a float: mp_obj_get_float(seconds_o);
tulip sleet
the bottom two bits of the 32-bits indicate whether the value is a pointer, an integer, or a float. we only have 30-bit floats, not 32 bits. ...
right
sad but true
tulip sleet
probably a bug; I don't know what other impls do (e.g. the original utime)
tidal kiln
if you think bug, i can post an issue, and just use that original example i showed
tulip sleet
so time.sleep(0.006) should sleep for a millisecond? or not sleep at all?
Go ahead and file an issue.
tidal kiln
should sleep for 6 milliseconds
and if not, time.sleep() documentation should specify lower usable limit
tulip sleet
sorry missed a zero: so time.sleep(0.0006) should sleep for a millisecond? or not sleep at all?
tidal kiln
good question
tulip sleet
CPython defn: https://docs.python.org/2/library/time.html#time.sleep
there was a change in 3.5 to make it sleep at least as long as requested, not shorter in case of interrupt
tidal kiln
hmmm. but they make no mention of low end behavior. should it rail to the lower limit (like 1ms) or should it drop to zero?
Python 3.5.2 (default, Sep 14 2017, 22:51:06)
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import time
>>> time.time()
1510257853.390322
>>> t1 = time.time(); time.sleep(0.1); t2 = time.time(); t2-t1
0.10014939308166504
>>> t1 = time.time(); time.sleep(0.01); t2 = time.time(); t2-t1
0.010099411010742188
>>> t1 = time.time(); time.sleep(0.001); t2 = time.time(); t2-t1
0.0010828971862792969
>>> t1 = time.time(); time.sleep(0.000001); t2 = time.time(); t2-t1
7.939338684082031e-05
micropython utime has sleep(), sleep_ms(), sleep_us(). since some micropythons don't have float, they needed to supply integer args for shorter times. Scott made decision to always have float and to have our libraries more like CPython.
tidal kiln
probably not much of an issue in CPython land.
just MCU fun
i'll post the issue so it can at least be documented and discussed
in the arduino code i was porting, the delay(1) was in a very tight/fast loop, so the total expected delay would be approx NSTEPS * 1ms
tulip sleet
I looked at the MicroPython impl - it's the same, so the truncation has always been there
tidal kiln
which worked there, but in the CP port it became NSTEPS * 0
that's probably a more likely scenario than a tight timing requirement in a driver
and if that does come up in a driver, time to move to c?
tulip sleet
well, the low-level neopixel code is in C. You could use microcontroller.delay(us), but again, it might get swamped. I wrote a Char LCD driver that has lots of sub-millisecond delays, cause I didn't know how accurate I had to be. I guess those all went to zero, but it still works.
manic glacierBOT
tidal kiln
^ fwiw
idle owl
3.0 issues are still posted to adafruit/circuitpython/issues right?
tulip sleet
@idle owl right, though you might preface the title with "3.0:" if it only affects 3.0, and set the milestone to 3.0beta. Somtimes I do a double-take on an issue that turned out to be 3.0-only.
idle owl
No problem.
tulip sleet
@idle owl @tidal kiln @solar whale @drowsy geyser align_usb from tannewt/circuitpython is now on master on adafruit/circuitpython
idle owl
@tulip sleet Thanks!
solar whale
@tulip sleet will download and test tomorrow. Thanks!
tulip sleet
@timber lion there are a bunch of CPy libraries that need new releases, especially since @fading solstice 's fix for read_into->readinto. I'm unsure of the workflow for libraries that are already wired into travis: should I just create a release that creates its own new tag and the .mpy will be created automatically, or do I need to push a new release tag first?
timber lion
oh just make a new release in github's releases tab and travis will do the rest
as long as the library was setup on travis' side with the GITHUB_TOKEN variable
most are, if you see releases with a .mpy then they're good to go
one thing though what's the story for people on older versions of bus device? they'll get a failure too unfortunately
something you might consider, keep the old read_into working and have it print a deprecation message
it can just call readinto internally
that's a good way in general to make breaking API changes I think
tulip sleet
@timber lion Scott suggested we support read_into I2CDevice for a while, and just change the library. See https://github.com/adafruit/circuitpython/issues/364#issuecomment-342566548. (The whole thread is worth reading.) So the idea is to encourage people to download the bundle.
idle owl
So, I figured out the problem. If I do a screen -ls after closing the terminal window, there will still be a session running. If I then explicitly terminate that session, I can reestablish the serial connection with the board and when it connects it's still in the REPL. At that point, I can ctrl D and reload the serial connection and it successfully exits the REPL and returns to running main.py. So it leaves the board in the state it was in (ie. connected to the REPL). I don't know if it's an issue with screen, or how it handles the serial connection. I'm thinking it's not necessarily an issue with CircuitPython.
idle owl
@tulip sleet Should I still file an issue, include the solution and close it? I don't think it's necessarily a CircuitPython issue, but I can't imagine I'm the only one who has ever run into it. Could be though 😄
tulip sleet
@idle owl That sounds like it's working as desired! screen lets you suspend and resume sessions - that's actually the original inspiration for it. I don't use -ls, so I haven't tried resuming.
idle owl
@tulip sleet Ok, noted. I had to look up -ls. I've never used screen for anything except establishing the serial connection.
I was able to use screen -x or -r or something like that to reestablish the original connection successfully.
tulip sleet
so using -ls is new for you? I'm forgetting what the original problem was.
idle owl
The original problem was that if I closed the terminal window while connected to the REPL the board continued to act like it was connected to the REPL and I couldn't reestablish the serial connection using the original screen connection command.
So I was trying to figure out why and with help found out that it hadn't terminated the screen connection, and it could be reestablished.
tulip sleet
if you close the terminal window and do ps -ef, do you still see screen running? I think it may try to survive that. actually I can try that myself.
idle owl
And if you use screen -X -S <session number> quit to terminate the session, the board still stays in the REPL. But you can then reestablish the screen connection using screen /dev/tty.usbmodemXXXXX 115200, and it pops directly back to the REPL.
Alternately, you can eject the board from finder, do a reset with the onboard button, and it will return to it's base main.py-running state.
Checking ps -ef now
tulip sleet
I don't see screen still running. But I almost never close the terminal while screen is running. I do ctrl-A backslash and answer "yes" to close the session
it's muscle memory now, but it was kind of annoying to begin with
idle owl
I was mounting the board, verifying screen would connect, verifying the REPL, then closing the window, ejecting, and resetting the board. Repeat. For testing USB stability. I happened to notice that if I left it in the REPL, the board remained in the REPL state.
My workaround to get main.py running again was to hard reset the board once more.
tulip sleet
in that case I can get back in easily. So there's something odd that happens when you kill screen by closing the terminal instead of exiting gracefully.
idle owl
Yeah.
tulip sleet
So do the ctrl-A \ thing instead and you can go in and out of the REPL without any resetting.
idle owl
I either need to hard reset, or find the screen session and reconnect, or terminate and reestablish.
tulip sleet
well just try what I said now, if you want, it should work
screen ... <type at the repl>. Type ctrl-A, which is the way to get screen to listen to commands. Then \ to close the connection
it will ask you if you really want to quit, and say yes
idle owl
So from inside the REPL do what you said?
tulip sleet
yes, type ctrl-A. That's NOT read by the REPL. instead screen intercepts it and expects you to type a single-chracter command. YOu can type ? to get a list of all the screen commands
idle owl
Hmm.
tulip sleet
did it work?
idle owl
No...
tulip sleet
are you in the repl?
idle owl
I was. I just reloaded the serial connection to try again.
tulip sleet
oh, on the mac, maybe it's cmd-A
idle owl
It's not, terminal accepts ctrl for terminal commands, cmd-A selects all.
Ahhhhhh
I had to keep ctrl held down, and I had to type A \ quickly
I wasn't doing that.
tulip sleet
so type ctrl-A once, then type ? by itself. you don't have to type quickly. hold down ctrl, press a, release ctrl, press ?
idle owl
Board stays in REPL state, but I can reestablish the serial connection successfully afterwards.
tulip sleet
did you see the screen menu of all the commands?
idle owl
Yeah it worked that time. I don't know what was going on before then.
But if I wait after hitting ctrl-A to type , it doesn't respond.
I have to do ctrl+a+\ altogether it seems like.
Or as long as I keep ctrl held down, I can type \ a little later and it works.
tulip sleet
I started up my mac. It's different quit is ctrl-, not plain \ so ctrl-A ctrl-\ then press y to answer the question
it's different between linux and mac
idle owl
Ah ok.
idle owl
That makes a little more sense. I see. I can pause between them and then hit ctrl+\ and it worked.
tulip sleet
but if you move to linux you will have to remember it's different: annoying but I guess as annoying as ctrl-C vs cmd-C
idle owl
Yeah, I have alt mapped to cmd on my keyboard hooked up to my mac laptop, and the whole thing is running Synergy with the Windows machine as the server, so I have to remember when I move to a different screen that cmd won't work. Honestly, using terminal regularly which requires all ctrl commands makes it a little less annoying to switch between them because I use control for so much on the mac. But inevitably I'll open "Tools" in some app on Windows at least once before remembering.
tulip sleet
I will take leave earlier tonight. kind of burned out last night. gn!
idle owl
@tulip sleet Have a good night!
manic glacierBOT
manic glacierBOT
solar whale
@tulip sleet pulled, built loaded master and it seems to be wrking fine on metro m4 - handles overfilling FS gracefully.
manic glacierBOT
solar whale
@timber lion @tulip sleet hmm - as warned the update to adafruit_bus_device has broken drivers for 3.0 - it was nice to have I2C for awhile - What needs to be done to fix this for 3.0?
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "oled.py", line 29, in <module>
File "oled.py", line 18, in <module>
File "libraries/drivers/bmp280/adafruit_bmp280.py", line 154, in __init__
File "libraries/drivers/bmp280/adafruit_bmp280.py", line 67, in __init__
File "libraries/drivers/bmp280/adafruit_bmp280.py", line 138, in _read_byte
File "libraries/drivers/bmp280/adafruit_bmp280.py", line 163, in _read_register
File "libraries/drivers/bmp280/adafruit_bmp280.py", line 161, in _read_register
AttributeError: 'I2CDevice' object has no attribute 'readinto'
glad rapids
Is it easy to switch a ESP8266 to Circuit Python and back to Arduino ??
solar whale
@glad rapids Yes - just reload your sketch via Arduino as normal. - Always fully erase the flash before loading Circuit Python with esptool.
@timber lion Why is there a problem with the attribute readinto? Ii is there in I2CDevice.
@timber lion @tulip sleet never-mind - I just had to RESET the board - all OK now under 3.0. Sorry for the false alarm.
idle owl
That's the best answer though!
solar whale
I rebuilt the bundle for 3.0 and copied it over, but I guess a RESET was needed to get everythin propely updated.
idle owl
Good to know
solar whale
I spent the whole day breaking things so I should have known better. I think I have gotten back to where I started today - Not much progress, but many lessons learned!
idle owl
That's always progress
timber lion
yeah i've noticed you need to reset the board for it to see file updates sometimes
perhaps a bug, not sure if that's by design
in general any time i copy over a new file i soft reset
solar whale
@tulip sleet Ran into an issue building bundles again ```Deleting existing build.
build-2.1.0/lib/adafruit_register 512
build-2.1.0/lib/adafruit_bus_device 512
build-2.1.0/lib/adafruit_hid 512
build-2.1.0/lib/adafruit_waveform 512
build-2.1.0/lib/adafruit_circuitplayground 512
build-2.1.0/lib/adafruit_pca9685 512
build-2.1.0/lib/adafruit_mpr121 512
build-2.1.0/lib/adafruit_lis3dh 512
build-2.1.0/lib/adafruit_ht16k33 512
build-2.1.0/lib/adafruit_rgb_display 512
build-2.1.0/lib/adafruit_ssd1306 512
build-2.1.0/lib/adafruit_max7219 512
fatal: No names found, cannot describe anything.
Stopping at 'libraries/drivers/ads1x15'; script returned non-zero status.
lib/README.txt 117 512
lib/update_linux.sh 646 1024
cunning crypt
@solar whale I often count lessons learned as more progress than actual progress. It often helps more in the long run.
idle owl
I have some audio recording code, and every so often it gives me a "length mismatch" that I've narrowed down to mic.record. The mismatch seems to be with the bytearray that's used as the buffer for recording, but I'm not sure why.
The code works. It also does the length mismatch every so often.
idle owl
It's set to 160 in the code, initially using math and a variable, but I changed it to the number. Changing the bytearray to 200 like in the example PDMIn code causes it to crash to safemode. Changing it to 180 causes a spammed constant length mismatch vs the every so many seconds length mismatch.
manic glacierBOT
umbral dagger
@idle owl I think I recall you mentioning a guide to writitng CP libraries... I looked around but came up empty.
idle owl
@umbral dagger Yes! Let me find it.
This may also help, though I'm not sure what your plans are, so it also might not. https://circuitpython.readthedocs.io/en/latest/docs/design_guide.html
umbral dagger
Thanks. not an official library at the moment, but I want to make it easy/familiar to use.
tulip sleet
@solar whale that library doesn't have travis integration yet. Glad it doesn't break the whole bundle. Thanks. EDIT LATER: now it is itegrated with travis.
vague monolith
@umbral dagger A library for?
umbral dagger
@vague monolith The dotstar featherwing. I'll be postiing on my blog very soon.
vague monolith
Ahh cool
umbral dagger
I debated whether to publish this so soon after last night's post, but I've been working on it for a while and want to empty the queue. So here it is. https://daveastels.com/2017/11/11/using-the-adafruit-dotstar-featherwing-with-circuitpython/
idle owl
The length mismatch is an error written into CircuitPython, it's part of audiobusio. I still can't figure out why I'm getting it.
idle owl
The next question is.... why is the exact same code not working on my other CPX. Still get the length mismatch every so often but I get no lights. LEDs work, I tested putting a pixels.fill in the code elsewhere and it worked. Mic works, I tested it with other recording code and it played back the recorded sample.
I didn't see how the length mismatch issue could have been hardware related, but at this point I was willing to try it anyway, and instead found out the code doesn't work on the second one. 😄
cunning crypt
Probably-obvious question: If I set up a Neopixel string with, say, 10 pixels and run code on it but with only one actual neopixel, it'll still work, right?
idle owl
Yeah. If it's an animation it just takes longer to get back to 1 if you make it with more than are present.
cunning crypt
IE, I set up something to flash each of the 10 pixels on a CPX, and then put that code on a Feather M0 Express, it'll still "Work"
idle owl
So if I have a strip that I'm running a rainbow on and double the LED number in init, it only shows half the rainbow at a time.
cunning crypt
Yep
So, question: Poking around lately, it seems that, effectively, all of the Circuit Python examples for hardware exist. Considering A) I don't want to make two copies of the same thing and B) The CP repo is, well, software, I've been thinking of shifting the focus of the repo's examples from hardware stuff to software stuff. Particularly things like the differences between Arduino code and CP code.
Do you guys think this is a good way to do it, or should I keep my original idea of more hardware-based examples?
idle owl
The real answer is you do what most interests you. But I agree that duplicating things if that's not your intention isn't the way to go. There's often two drivers, one Arduino and one CircuitPython, but I don't know that there's much comparison of the two. There's one guide written on porting from Arduino to CircuitPython but it assumes a pretty significant amount of knowledge of both on the part of the user. Hardware based examples are nice, especially if you pick up on a way to use it that doesn't already exist, but some stuff has a pretty singular use. I think if you think you'd enjoy embarking on something different, go for it. I really feel it's more about what you want to do, because you should enjoy it.
cunning crypt
Well, yes, but it should also be useful.
idle owl
So far everything you've mentioned is.
cunning crypt
I think a lot of the quirks of Python in general need to be brought up. Especially things that Arduino just simply Does Not Allow.
Most of what comes to mind is things with variables.
You can swap a variable for whatever you please in Python.
idle owl
I don't know much about Arduino. By which I mean basically nothing.
cunning crypt
You also don't have to declare them, and when defining functions you don't have to set yourself up for just one variable type. VAriables can be, well, variable.
idle owl
Yeah
cunning crypt
C-variant langauges (Like Arduino) have to have the variables declared as a specific type. Python simply cares not.
Also, man is Python so much easier for working with strings.
idle owl
I lucked out then. I started with Python.
cunning crypt
String manipulation is why Python was my first programming language.
And it was super awesome when I learned that Python was coming to Microcontrollers.
idle owl
Yeah! Agreed
I learned about CircuitPython before finding out about MicroPython. I was the only person at a conference at the end of July who knew about MicroPython - a guy did a talk on it, and he was so stoked when I talked to him about it afterwards.
umbral dagger
Interesting conversations there. Makes me think... I’m learning Python at the same time I’m learning Haskell.
cunning crypt
I first heard of MicroPython in July of last year, when it was, essentially, just the PyBoard. I fiddled a little when I found it was on the ESP8266, but there wasn't enough going on to make it worthwhile. When I hopped on the Adafruit discord and saw the CircuitPython stuff, I knew I couldn't just sit by and watch anymore
idle owl
That's great!
cunning crypt
@umbral dagger You're learning two languages at once? That seems confusing.
Hmm, CircuitPython having a built-in easily computer-accesible filesystem is also super useful.
Especially for Express boards.
umbral dagger
@cunning crypt And they're so very different.
@cunning crypt Haskell doesn't really have "variables"
@cunning crypt Immutability is the rule
@cunning crypt The file system access really makes a huge different.
cunning crypt
It opens up so much more on even the most basic CP board that Arduino simply doesn't have
umbral dagger
The tradeoff for now is memory limitations. Python has significant overhead compared to C++.
cunning crypt
That's why it's on the M0
cunning crypt
Exactly. It's effectively the same thing as the lower-sized Arduino chips
umbral dagger
The x20 chips will be amazing for CP. 1M flash & half a M ram.
cunning crypt
x20?
umbral dagger
The SAMD chips end in a letter & number.