#circuitpython-dev
lone axle
tulip sleet
righto
click ok, and click the 3 dots in the datafile section
choose the bootloader-feather_m0...bin file
let's erase the whole chip to make the fuses be 0xffffffff so they'll get repaired
so after you choose the data file, click Erase Chip
lone axle
That popped up a window ran some progresses and then closed
did not seem like any errors
tulip sleet
IN the log you might need to scroll it to see if there are any errors
then click "Program device", which will load the bootloader
check for errors there too, you'll need to move the horizontal scroll bar to see
lone axle
Ah scrolling in log I see:
Selected file: path/to//bootloader-feather_m0-v3.9.0.binConnecting to J-Link...Connecting to target...Erasing...Done
ok, now disconnect the jlink from the header and double-click
tulip sleet
single-click will probably work too, because the user program has been erased
(worked for me)
lone axle
same results with double and single (only blinks yellow LED once)
tulip sleet
are you disconnecting the J-Link ribbon cable from the box header breakout?
tulip sleet
no reason to use wall power at all, just leave it on the hub the whole time
lone axle
I will try again with it in the hub the whole time
tulip sleet
before you do that, plug in the j-link back in, and let's look at memory. Run JMem.exe
i mean do move it to the hub, but don't repeat the programming yet
lone axle
Okay, jmem open
tulip sleet
could you screen shot?
lone axle
tulip sleet
click ok
lone axle
tulip sleet
actually no, choose ATSAMD21G18 first in Target device
sorry
then screenshot the first page of the memory dump
Also choose Fixed Speed (4000)
(I am getting core dumps without that)
lone axle
tulip sleet
click ok
lone axle
lone axle
tulip sleet
type 1CC0 in the Go to file and you should see the version string on the right, it should be 3.9.0
then Go to 2000, and it should be all FFFF's
lone axle
tulip sleet
go to 804000 to see the fuses
lone axle
Looks like they did get set back
Not sure how most of this works, perhaps my unplugging from wall power before clicking reset could've been a problem?
tulip sleet
i don't think that's true, but you can certainly skip that.
the fuse settings are still causing infinit resets
but the bootloader itself is in good shape
we can go back to JLink.exe and try writing it again. We could also try writing it with gdb if that doesn't work
tulip sleet
ok, I'm trying it on mine, and the writes aren't really happening, that's a problem. so wait a bit and let me try it with gdb
lone axle
Okay will do
tulip sleet
is this the only computer you have? do you have an RPi or some other Linux machine?
lone axle
I do have many other computers including RPi and a Linux laptop
tulip sleet
could you fire up the Linux laptop and download and install the J-Link software there? What kind of Linux?
have you done CircuitPython builds already? DO you have the arm gcc toolchain isntalled somewhere?
lone axle
I did recently complete some basic circuit python builds. It was on a Debian VM inside of this win7.
I think the other Linux ones I have are Debian as well. I will check
tulip sleet
lone axle
Debian 9.5 on another machine I'll try it there.
tulip sleet
it appears that JLink.exe does not know how to write to the fuses properly; it requires some special commands different than a regular write to flash
segger software download is here : https://www.segger.com/downloads/jlink/JLink_Linux_x86_64.deb
Then do sudo dpkg --install on the .deb
aha, I found a trick way to write the fuses:
mem32 0x00804000, 1
mem32 0x00804004, 1
w4 0x4100401c, 0x00402000
w4 0x41004000, 0xa505
w4 0x00804000, 0xD8E0C7FA
w4 0x00804004, 0xFFFFFC5D
w4 0x41004000, 0xa506
you can do this with JLink.exe. (from https://electronics.stackexchange.com/questions/438386/samd21-how-do-i-write-to-nvm-fuses)
this would be easier than setting up gdb unless you're mostly there
(it's sending raw commands to the flash write hardware to do the work)
lone axle
It took me a minute to get discord up to get the links. So I am here now on the linux machine but with nothing installed yet. I'll try that way first
(with the trick you found, back on windows)
lone axle
it poped up some progress bars really fast on the last one. disconnecting the jlink ribbon now
still the same, yellow LED when press reset but never red
tulip sleet
alright, i'm going to take a break for a couple hours. you can too, or ping me when you have the toolchain installed
(family matters call)
we are close, but these tools are not great
lone axle
Okay. Thank you for all your help. I'll work on getting that set up and I have some other things I can work on for a while, including dinner.
tulip sleet
i may modify the bootloader a bit to be less strict about when it thinks the fuses are busted
@lone axle this version of the bootloader should fix your fuses. Try doing the loadfile and verify, and then disconnect and reset as before.
if it doesn't work, disconnect the board so it doesn't wear out the flash trying to fix the fuses
or you can use JFlashLite to flash it, at 0x0. that's less tedious
lone axle
@tulip sleet I had unplugged the feather when I stepped away a moment ago. Just came back and plugged the feather back in (to the USB hub) and I have pulsing red now without touching reset button at all.
Oh, and FEATHERBOOT
Everything working as in intended now as far as I can tell. Thank you!
slender iron
@crude blaze do you have meson experience?
lone axle
@solar whale Success!!
Waiting for packets...
Received (raw header): ['0x2', '0x1', '0x0', '0x0']
Received (raw payload): bytearray(b'Startup message 0 from node 1')
Received RSSI: -30.0
``` 📻@lone axle that's great! I'm confused why it's working, or did you use the new bootloader I sent you?
The best way to make sure the fuses are OK is to now update the bootloader with itself: https://github.com/adafruit/uf2-samdx1/releases/download/v3.9.0/update-bootloader-feather_m0-v3.9.0.uf2. Copy that to FEATHERBOOT and wait for FEATHERBOOT to reappear.
lone axle
I can't be sure, but I think the trick that you found for writing the fuses did actually work but at least in my case not until I unplug / replugged as opposed to using the reset button.
It was plugged into an externally powered USB Hub at the time (not sure if that matters).
So I did not use the new bootloader you sent.
When I came back and plugged it in (was about to try the bootloader you sent) I was met with pulsing red on first power up.
@tulip sleet Okay I've copied this update-bootloader uf2 onto FEATHERBOOT and it has reappeared.
lone axle
Thank you so much!
tulip sleet
thanks for your perseverance. I was going to write up how to write the bootloader using the JLink tools, and this has been very informative.
it also gives me some hints about writing some fuse-fixing code in a future version of the bootloader
you're welcome and thanks again
lone axle
Is it possible to use the BOSSA tool to write the uf2 bootloader onto the other one of these (that I didn't mess up the offset on the first time)
tulip sleet
@lone axle you can use the https://github.com/adafruit/uf2-samdx1/releases/download/v3.9.0/update-bootloader-feather_m0-v3.9.0.ino Arduino sketch. Just compile and upload it in Arduino, and it will replace the bootloader. Using bossac won't work due to the boot protection fuses.
lone axle
Ah, neat I'll use this arduino sketch
tulip sleet
it was broken until this release, but I rewrote it and I think it should work for you.
lone axle
Hmm, what if It's already running CircuitPython
tulip sleet
you'll have to reload circuitpython, and save your CIRCUITPY contents beforehand
lone axle
I've got everything I need from it.
Will Arduino be able to see it when it's in Circuit Python mode? Or do I need to go to the bootloader or something?
tulip sleet
CircuitPython can respond to the upload request from Arduino. Or more simply, double-click to get the Arduino bootloader, make sure the port is chosen in Tools->Port, and then upload the sketch
lone axle
My arduino IDE is pretty ancient I'm going to install a newer one
manic glacierBOT
lone axle
@tulip sleet I got this output in arduino:
Sketch uses 19476 bytes (7%) of program storage space. Maximum is 262144 bytes.
Atmel SMART device 0x10010005 found
Device : ATSAMD21G18A
Chip ID : 10010005
Version : v2.0 [Arduino:XYZ] Mar 5 2016 17:46:52
Address : 8192
Pages : 3968
Page Size : 64 bytes
Total Size : 248KB
Planes : 1
Lock Regions : 16
Locked : none
Security : false
Boot Flash : true
BOD : true
BOR : true
Arduino : FAST_CHIP_ERASE
Arduino : FAST_MULTI_PAGE_WRITE
Arduino : CAN_CHECKSUM_MEMORY_BUFFER
Erase flash
done in 0.818 seconds
Write 19732 bytes to flash (309 pages)
[====== ] 20% (64/309 pages)
[============ ] 41% (128/309 pages)
[================== ] 62% (192/309 pages)
[======================== ] 82% (256/309 pages)
[==============================] 100% (309/309 pages)
done in 0.138 seconds
Verify 19732 bytes of flash with checksum.
Verify successful
done in 0.022 seconds
CPU reset.
An error occurred while uploading the sketch
SAM-BA operation failed
that's probably just an upload issue. Double-click to get the pulsing bootloader, and then look in the Tools->Port menu to see which one is the board, and select that one (it should have a check mark). Then try again.
oh, wait, it looks like it worked
is the bootloader still there?
the "error" is that the board reset, which is supposed to do
lone axle
Actually yep it seems like it did work. I double clicked and it came up to FEATHERBOOT
lone axle
Thanks again
solar whale
@lone axle glad you got it working!
lone axle
I will try out the ack ones tomorrow.
lone axle
Playing with it so far makes me interested to try to make a basic text message app where the two devices ill be able to type and send short messages to each other.
manic glacierBOT
crude blaze
@crude blaze do you have meson experience?
@slender iron sorry, very little. I probably wouldn't be of much help
onyx hinge
I'm interested in #2762 because there was a weird unexplained problem with the ulab submodule when I worked on actions@v2
I also figured out something else about actions failing due to tags
onyx hinge
wonder if this matters, seen while using pip to install dependencies of circuitpython's build: ```botocore 1.15.39 has requirement docutils<0.16,>=0.10, but you'll have docutils 0.16 which is incompatible.
awscli 1.18.39 has requirement docutils<0.16,>=0.10, but you'll have docutils 0.16 which is incompatible.
and no, that wasn't the secret answer to whatever is wrong with ulab 😢
oh wait maybe I didn't rebase my branch as I thought
onyx hinge
huh I thought maybe using a github action with a Dockerfile that installs all dependencies would speed things up, but in fact it rebuilds the docker image everytime. https://help.github.com/en/actions/building-actions/dockerfile-support-for-github-actions
onyx hinge
And in fact this run has weirdness for multiple submodules, it's not ulab that is special
@lapis hemlock can you make a pull request to circuitpython with all the ulab improvements? I can walk you through it if you would like. Maybe Monday before or after the meeting.
lapis hemlock
@lapis hemlock can you make a pull request to circuitpython with all the ulab improvements? I can walk you through it if you would like. Maybe Monday before or after the meeting.
@onyx hinge Absolutely! In return, you've got to merge the latest PR inulabitself😉 I'm not working on Monday, so I am flexible.
@lapis hemlock can you make a pull request to circuitpython with all the ulab improvements? I can walk you through it if you would like. Maybe Monday before or after the meeting.
@onyx hinge I guess, your question was triggered by this issue https://github.com/adafruit/circuitpython/issues/2762, which, in the meantime, has been resolved.
onyx hinge
maybe I got it this time. 🤞
wary ember
newby here, just got a m4 airlift lite, trying to move from arduino. Can't find any libs or simple examples to send/receive email, like umail or maybe wifinina. Where do I look? I'm doing IoT home apps to send alerts etc when a monitored condition happens. Thanks.
crimson ferry
@wary ember You'll want the ESP32SPI and Requests libraries, for starters, for HTTP requests, TCP sockets and the like. They're in the library bundle here: https://circuitpython.org/libraries. There are not currently higher-level libraries for email that I'm aware of (I do a request to my server to send email etc).
lone sandalBOT
remote dawn
Total noob here; I have a Metro M0 on order but who knows when they'll ship it. Meanwhile trying out Mu and starting to learn CircuitPython. But my question is this: does the Metro M0 have to be plugged in to computer USB port to work? I'd like to develop a compact little box that does a few tricks that I can carry around and not have tethered to a computer.
main meteor
No, once the code is loaded, it can run standalone (as long as it doesn't require the serial interface to be up)
That's one of the really nice things about these little CircuitPython boards, you can program them and then include them in stuff like props to do their thing.
remote dawn
Perfect! That's what I was assuming. But somewhere in the docs for the board it says "no EEPROM" and that made me wonder. Because of course storing the program so that it stays in there even with the power off does use EEPROM. But I'm an engineer who has used EEPROM's since the 1980's. I guess nowadays in the Arduino world EEPROM refers only to an area where parameters can be changed without changing the underlying logic of the program.
crimson ferry
Yeah, I remember UV EPROM. Now it's mostly NAND flash, often with a bit of emulated EEPROM.
marble hornet
BUG?: I am unable to import the nvm module in 5.2 on the metro m4, I grabbed it from microcontroller but is the lack of the nvm module intentional?
timber mango
If you just want the nvm module, you need to do:from microcontroller import nvm
marble hornet
@timber mango that is an instacne of a bytearray from the nvm module is it not? i'm curious if not being ale to import nvm as a module is intentional.
manic glacierBOT
zinc maple
https://www.tindie.com/products/makerfabs/circuitpython-pewpew-m4-kit/ This yours? @stuck elbow
lone axle
@zinc maple yep. The Pew Pew M4 (and original Pew Pew) were created by de∫hipu
stuck elbow
@zinc maple yes
tulip sleet
@idle owl I am going to cookie-cutter a couple of new libraries. cookie-cutter looks up to date with black, etc. Is it fine to use as is now?
idle owl
@tulip sleet Should be yeah. Will still require running Black on a few things as we couldn't quite get it all proper with the template bits in it.
tulip sleet
thanks! I'll run a whole-tree black before committing anything to the new libs
modern wing
Good afternoon all you wonderful code slingers -- happily lurking today 🙂
solar whale
we need a lurking emoji
turbid radish
Hey all, weekly reminder about tomorrow's CircuitPython Newsletter - subscribe at adafruitdaily.com, send news, projects, etc. to me on twitter @anne_engineer or email anneb@adafruit.com. I don't monitor Discord for CP news (yet).
old smelt
Lurking
turbid radish
I'm lurking while doing the newsletter
errant grail
Lurking
timber mango
lurking today
fathom trellis
Listening in today
turbid radish
Hey Nina!
fathom trellis
👋
turbid radish
omg, yes
ivory yew
purple gang represent
sterile bronze
lurking
fathom trellis
👍 to getting pinged
ionic elk
speaking of notes where the notes at?
idle owl
@ionic elk It's also pinned to the channel. https://docs.google.com/document/d/1brXruQuOrHT1cdZ6Do-4O9wlcwYZDUV8MIQPYe_1gdo/edit#
tidal kiln
lurking
turbid radish
Yea @ivory yew !!!!!!!!!!!!!!!!
fathom trellis
@ivory yew 🎉✨
raven canopy
👏 @ivory yew! 🎉
ivory yew
Thank y'all. To celebrate, I want 30 minutes of Scott trying to pronounce slavic names.
onyx hinge
idle owl
Congrats @ivory yew!
thorny jay
Lurking and no notes! Sorry, did nothing this week. While feeling much better and totally recovered from Covid, I could not get started doing anything.
lapis hemlock
@ivory yew From me, too!
ivory yew
💜
onyx hinge
raven canopy
where is the :1000: emote? 😄
slender iron
raven canopy
i really should get some Nitro. haha
onyx hinge
onyx hinge
inland tusk
lurking for status update
slender iron
tiny oriole
Apologies for the late arrival. If i dont get it typed in in time, Group hug to all!
inland tusk
enjoyed the Tiny USB for ESP32 S2
lapis hemlock
while True:
hug(@jepler)
is there an esp32-s2 dev board available?
idle owl
@tiny oriole Are you text only?
tiny oriole
for the moment. possibly not for the weekly update though
had a work conflict for the first part
idle owl
@tiny oriole No worries, got it in the notes
tiny oriole
thanks!
thorny jay
@solar whale There was on in the video from the desk of lady Ada. But the stock was at zero for me.
lone axle
Go ahead and read it sorry
solar whale
thanks -- will keep check the video and keep an eye out
lone axle
I think I've got the mic settings straightened out. I will give it another try for status reports.
inland tusk
@onyx hinge mute
idle owl
@spice crypt Are you lurking?
onyx hinge
good to hear your voice @raven canopy
fathom trellis
I have to drop off! Thanks folks, this was awesome. 🙌
idle owl
No worries! Thanks for joining us!
spice crypt
how do I set "lurking"
idle owl
By telling us 🙂
raven canopy
@onyx hinge the "perks" of tele-work. 😄
turbid radish
Just post the word Lurking here in chat
spice crypt
lurking
solar whale
Do the recent changes to CP github workfow (like checkout v2) require or recommend any changes to users doing local builds
onyx hinge
@solar whale no, shouldn't. if you want to build the new litex port you need to install a new compiler though, because it's based on RISC-V instead of ARM.
solar whale
Do you have a link to the RISC-V toolchain that is needed?
onyx hinge
@solar whale the build uses https://static.dev.sifive.com/dev-tools/riscv64-unknown-elf-gcc-8.3.0-2019.08.0-x86_64-linux-centos6.tar.gz even though it is using ubuntu.
old smelt
Have to drop early today. Great sitting in for a bit again. Have a great week all!
idle owl
@old smelt Thanks for joining us!
raven canopy
@tiny oriole i was glued to NWS last night. we were on the tail end of that system... 😬
tiny oriole
yeah, i got a rando text from my Spotter friends yesterday telling me to look over at radar and was greeted with a system so large you could see a pretty well defined eye on the radar.... which for those of you not familiar with NWS Radar, being able to see a tornado signature is usually a matter of "knowing what to look for"... seeing what is called a "debris ball" signature, let alone a debris ball with a clear eye, is extraordinarily rare.
lone axle
Ok, go ahead and read them, sorry
inland tusk
@tiny oriole I was watching the MS saga through a friend;s website. Saw the major Tornado as it went by his house to north.
The trick to seeing it is to look for winds on the boundry between the red areas and a greeen areas.
they will be going in opposite directions
onyx hinge
@slender iron I have the 32x16 and 64x32 so yours will be a little different. but m4 feather + that wing is the quickest way to get a working display.
yay! belated hug for getting the "current" sphinx working, I missed that going by
ionic elk
@tulip sleet how often does background_tasks call the flash flush?
onyx hinge
idle owl
Or run: black --target-version=py35 .
raven canopy
Black commit-hook: https://black.readthedocs.io/en/stable/version_control_integration.html#
onyx hinge
tannewt turned into a robot
it got better
@lapis hemlock do you want help making a pull request to update ulab in circuitpython? If so, would you like to do it now or at some other time?
lapis hemlock
@onyx hinge are the issues "ulab can give erroneous results when sorting 2D arrays", and "ulab can give erroneous results with expressions like a[a<1]" new? (5.2 release notes https://blog.adafruit.com/2020/04/09/circuitpython-5-2-0-released/)
turbid radish
Thanks Scott and all!
gilded cradle
thanks
lapis hemlock
@lapis hemlock do you want help making a pull request to update ulab in circuitpython? If so, would you like to do it now or at some other time?
@onyx hinge I am free now.
errant grail
Thank you!
onyx hinge
@lapis hemlock they were in 5.1 and 5.2, I noted them because those were fixed in ulab master branch but we didn't take those fixes .. and still haven't.
lapis hemlock
@lapis hemlock they were in 5.1 and 5.2, I noted them because those were fixed in ulab master branch but we didn't take those fixes .. and still haven't.
@onyx hinge OK, thanks! So, what have I got to do now? With the PR, I mean.
onyx hinge
@lapis hemlock so I think we want to take all the changes in ulab master branch into circuitpython masater branch
thorny jay
@tulip sleet Is there a BLE driver/example that does serial / UART over GATT. For some hardware, there seems to be a TX and a RX UUID/characteristic that I need to write to/read from. Would you have any code that drive BLE hardware that behave like that?
lapis hemlock
What has changed? In the past it used to happen automatically, didn't it?
onyx hinge
@lapis hemlock so in a git clone of circuitpython, you'd cd extmod/ulab; git fetch origin; git checkout origin/master so that the ulab submodule is pointing at the head of the master branch
@lapis hemlock no, it didn't happen automatically in the past
while I was working on the initial pull request into circuitpython, I was regularly updating
anyway, after that, you can run whatever local tests you want, then git add extmod/ulab, commit, and create a pull request in the usual way against adafruit/circuitpython
There are a number of steps but it is a fairly mechanical process
lapis hemlock
OK, I can do that. I am cloning at the moment.
onyx hinge
thanks, I appreciate it. I think right now is a good time to do a merge, it'll go out with this next set of stuff that we'll call 5.3
lapis hemlock
What does this mean ```You are in 'detached HEAD' state. You can look around, make experimental
changes and commit them, and you can discard any commits you make in this
state without impacting any branches by performing another checkout.
If you want to create a new branch to retain commits you create, you may
do so (now or later) by using -b with the checkout command again. Example:
git checkout -b <new-branch-name>
HEAD is now at 1f17bdb4f Merge pull request #2765 from jepler/checkout-v2-subomdules
hmm I already left out a step 🙂 so let's back up a moment.
manic glacierBOT
onyx hinge
check out the "master" branch again, go back to the top level, and git submodule update --init. Because I left this step out, the git checkout command actually changed the main circuitpython tree, not the ulab tree
I could tell that because of the commit message, which was a circuitpython commit message and not a microlab one
lone axle
@thorny jay maybe these? https://github.com/adafruit/Adafruit_CircuitPython_BLE/blob/master/examples/ble_uart_echo_client.py and https://github.com/adafruit/Adafruit_CircuitPython_BLE/blob/master/examples/ble_uart_echo_test.py
solar whale
@tulip sleet any issues going to 20.04?
onyx hinge
As for that message in general .. there's a distinction in git between checking out a local branch git checkout master and checking out other things like tags and remote branches git checkout origin/master. That's simply what it says when checking out a thing that is not a local branch. Usually, in submodules, you have a specific commit checked out, rather than a local branch. So I don't think about it much, and I'm not surprised when I see that message. I think you could git checkout master; git pull and get about the same effect as git fetch origin; git checkout origin/master except you would not see that message
lapis hemlock
@onyx hinge Do we have to merge the last PR from ulab before going ahead?
onyx hinge
I can give it a quick look before we proceeed
lapis hemlock
OK.
HEAD is now at 12d2c91 Merge pull request #92 from v923z/transpose is the state before the PR.
onyx hinge
This doesn't seem right. ```>>> ulab.numerical.argmin([1.2, 3.4])
1
Thanks @lone axle , I need to check the implementation of adafruit_ble.services.nordic UARTService .
onyx hinge
This seems weird too ```>>> ulab.numerical.argmin([1])
1
ulab.numerical.argmax([1])
1
It seems that serial over BLE has not been standardised. But many old protocol were serial and ported to BLE.
lapis hemlock
This doesn't seem right. ```>>> ulab.numerical.argmin([1.2, 3.4])
1
@onyx hinge Indeed. That was a stupid indexing error. I have pushed a fix.
Sorry for the confusion!
github is done with the checks.
slender iron
Here is the notes document for Monday’s CircuitPython Weekly meeting. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868>https://docs.google.com/document/d/1YIp-LKPdIzsN2LN4yRo4FzxzTTnLuFeDOJPboQKDVhc/edit?usp=sharing
onyx hinge
@lapis hemlock I still see some weirdness and put more details in the pull request. I need to step away and go grocery shopping now.
lapis hemlock
@lapis hemlock I still see some weirdness and put more details in the pull request.
@onyx hinge Too bad. I will try to find out what the problem is.
onyx hinge
feel free to ping me tomorrow, after 1200GMT is good for me.
thanks for keeping at it!
lapis hemlock
OK, will do.
tulip sleet
@thorny jay @lone axle pointed to what I would point to
the Bluefruit Connect applications all use BLE "UART"
@solar whale I have not seen any significant 20.04 problems, but I regularly get "System Problem detected" popups. I am using the cinnamon desktop and haven't taken advantage of any new mainline UI updates. I was hoping I might see fewer USB hangs, but if anything they might be slightly worse. It's nice having newer default python etc. but it's not miraculously better.
solar whale
Thanks -- I'll try it soon.
lapis hemlock
@onyx hinge I think I have sorted it out. And thanks for the test script! Next time I will have to implement the hug function with infinite recursion, like so: ```python
def hug(whom):
while True:
hug(whom)
hug(@jepler)``` Though, I don't quite know what happens in such cases. It might break my computer. 😆
onyx hinge
you must have fixed the bug in my test script too
lapis hemlock
Yes, I have. Was that intentional?
onyx hinge
no
lapis hemlock
Such a shame! It would've made for a good prank.
onyx hinge
there was no poisson d'avril this year
lapis hemlock
We can merge it, then, right?
onyx hinge
I just did
lapis hemlock
Thanks! Do you want to proceed with the circuitpython stuff?
I can stay for a couple of minutes, but not longer than half an hour.
I have pulled in the changes, this is my status now: HEAD is now at 22813d6 Merge pull request #91 from v923z/argmax
onyx hinge
that looks right
onyx hinge
right. do you have a fork of circuitpython yet? Or, if you have a micropython one, that works.
lapis hemlock
I can fork CP.
onyx hinge
you create a branch, "git add extmod/ulab", push the branch to v923z/circuitpython, and create the PR using the web interface
idle owl
@tulip sleet Can you take another look at this please? It's the last Black PR. https://github.com/adafruit/Adafruit_CircuitPython_CPython/pull/11
onyx hinge
"git add", "git commit", push
lapis hemlock
If I try to fork CP, I end up on my https://github.com/v923z/micropython-property
onyx hinge
if you use "git gui" it can show you a summary of the ulab commits that are being added
yes, that's more or less expected. Because circuitpython is forked from micropython, github doesn't create any new forks when you do that
(I think it's not ideal, but they have reasons for doing it that way)
lapis hemlock
OK, then I would have to add ulab to that.
onyx hinge
you started by cloning adafruit/circuitpython, right?
lapis hemlock
That's correct.
onyx hinge
and now you need to put the changes up to your fork, which is https://github.com/v923z/micropython-property
lapis hemlock
Is it OK, if I simply copy my ulab directory to https://github.com/v923z/micropython-property/extmod and then push it to github?
onyx hinge
Not really, because you need to start with the circuitpython master branch, not something from micropython
Let's do it this way. git remote add v923z git@github.com:v923z/micropython-property.git ; git fetch v923z. That should print something that looks successful.
In your circuitpython clone
lapis hemlock
Do you want all the bells and whistles?
onyx hinge
not if it looked succesful
lapis hemlock
It didn't complain, but I don't quite see anything familiar.
onyx hinge
you can paste it if you're not sure
lapis hemlock
remote: Counting objects: 100% (3722/3722), done.
remote: Compressing objects: 100% (6/6), done.
remote: Total 13303 (delta 3716), reused 3719 (delta 3716), pack-reused 9581
Receiving objects: 100% (13303/13303), 6.54 MiB | 4.91 MiB/s, done.
Resolving deltas: 100% (9759/9759), completed with 1019 local objects.
From github.com:v923z/micropython-property
* [new branch] cross -> v923z/cross
* [new branch] dynamic-native-modules -> v923z/dynamic-native-modules
* [new branch] esp-extra-scripts -> v923z/esp-extra-scripts
* [new branch] iabdalkader-cc3k-update -> v923z/iabdalkader-cc3k-update
* [new branch] make -> v923z/make
* [new branch] master -> v923z/master
* [new branch] parse-bytecode -> v923z/parse-bytecode
* [new branch] reentrant-gc -> v923z/reentrant-gc
* [new branch] travis-clang -> v923z/travis-clang
* [new branch] travis-esp8266 -> v923z/travis-esp8266
* [new branch] travis-stackless -> v923z/travis-stackless
* [new branch] travis-testing -> v923z/travis-testing
* [new tag] v1.10 -> v1.10
* [new tag] v1.11 -> v1.11
* [new tag] v1.12 -> v1.12
that's great
lapis hemlock
I am glad that you are so excited.
onyx hinge
those are all the branches you created previously and pushed to micropython-property.git(as well as some tags of micropython versions)
we'll get there
let's see .. did I have you create a branch in your circuitpython clone? If not, let's do that now: git checkout -b update-ulab
lapis hemlock
Done.
onyx hinge
at some point I hope this gets close to the familiar pull request workflow you already knew well enough
if you didn't git add / git commit (or your favorite way of creating a commit, such as git gui), do that next.
onyx hinge
extmod/ulab
lapis hemlock
OK.
[update-ulab 965ef48a3] added ulab
1 file changed, 1 insertion(+), 1 deletion(-)
git push --set-upstream v923z update-ulab
lapis hemlock
This is weird: ```
ERROR: Permission to v923z/micropython-property.git denied to deploy key
fatal: Could not read from remote repository.
Please make sure you have the correct access rights
and the repository exists.
that is weird. we just ran right past the edge of my git knowledge 🙃
do you usually use "git@github.com" when you pull/push things for ulab generally?
onyx hinge
want me to just complete the pull request process this time around and we can worry about it some other time?
lapis hemlock
I am fine, I don't mind, if you want to work on this for a bit.
But if I am wearing you out, then...
onyx hinge
I'm not sure how to proceed, I'm unfamiliar with this message
also I can't reproduce this locally to look for resolutions to the problem
lapis hemlock
We can let it drop, if you want to.
onyx hinge
for now anyway, I'll put in a pull request so we actually get these bug fixes and improvements you worked on in the past, and then next time there are some accumulated changes you'd like to see in CircuitPython we can revisit it.
onyx hinge
In the long run we'd like to have you be comfortable creating a pull request to circuitpython with ulab improvements (or other stuff 🎁)!
lapis hemlock
What I don't quite get is, why this failed with this: fatal: Could not read from remote repository.
Everyone should be able to read that repository.
onyx hinge
the "git remote add" command I had you run must have been wrong, except that when you did a "git fetch v923z" it looked like success to me. I'm confused.
lapis hemlock
So am I.
onyx hinge
I'll let you go for now. Have fun with your projects, I look forward to hearing about them when they're sufficiently ready.
lapis hemlock
I will keep you posted. 👋
@onyx hinge How come ulab fails on circuitpython, but passes with micropython?
onyx hinge
uh oh! I don't know (yet)
nothing interesting. ```$ ../ports/unix/micropython -X emit=native ../extmod/ulab/tests/argminmax.py
NotImplementedError: native yield
But yield has been in micropython for ages, I am pretty sure it is also in circuitpython.
I know you split off at v.1.94, but that is not too old.
onyx hinge
"-X emit=native" means, roughly, turn all python code into native (in this case, x86_64) code. it's not the default.
lapis hemlock
Let me know if you want me to re-write the test script.
onyx hinge
so in that case we just need to skip it. (It still gets tested in circuitpython without -X native). There are many tests that have this flag set.
lapis hemlock
Or that.
onyx hinge
ah, tests
lapis hemlock
OK, so there is nothing I have to do on my end.
onyx hinge
nope
lapis hemlock
OK, then see you!
onyx hinge
👋
crimson ferry
The CircuitPython TFT display libraries are basically write-only to init the drivers. Is it possible, and desirable, to add some kind of read status back from the drivers so that CircuitPython code can verify the presence and function of the attached display?
slender iron
@gilded cradle https://forums.adafruit.com/viewtopic.php?f=60&t=164335
crimson ferry
This is generally a feature I like in peripherals, but atm this is driven by my 240x135 (ST7789) going blank after a couple of days. The 2.4" and 3.5" displays don't do that. If I ^C ^D, it will start displaying again.
slender iron
@gilded cradle https://forums.adafruit.com/viewtopic.php?f=60&t=164372
thanks @ionic elk and @gilded cradle for the forum follow ups
slender iron
folks who are waiting for a PR review from me will have to wait until tomorrow. I'm barely through email now
will be grocery shopping for a bit in the morning too unless I go tonight
solar whale
Be careful out there!
slender iron
they're orchestrating things well. I'm not worried about it
raven canopy
i scored toilet paper today. the benefit of not going at night, like i usually do. 😄
slender iron
I'm on the waitlist here: https://blog.whogivesacrap.org/home/goodnews/stock-updates
they send a big box so we've got plenty left
manic glacierBOT
onyx hinge
the "upgrade ulab in circuitpython" PR had more loose ends than I expected, in a way I'm glad that I hit a snag working with v923z to do it.
half geyser
I'm looking through the circuitpython source, and I see this function cpu_get_regs_and_sp(regs) that gets called a lot. It assumes there are 10 registers.
- This isn't the case on RISC-V, which has 30 registers, and
- The
regsargument doesn't appear to ever be used.
Could this be replaced with a functioncpu_get_sp()? Or is the register list actually used somewhere that I'm not seeing?
tulip sleet
@half geyser cpu_get_regs_and_sp(regs) gets called two places, when the stack is allocated, and in gc_collect(). You're right, it seems like it doesn't use regs; perhaps it used to check the register contents for pointers to gc. Could you open an issue? Thanks.
sour lynx
@slender iron I was watching your livestream about ESP32-S2 support in micropython, and had a couple of pointers/ideas on integrating with the ESP-IDF.
slender iron
@sour lynx that'd be awesome! I'm not sure where to start since we use make and the idf uses cmake
sour lynx
Right... some time in the future, if you start using CMake in circuitpython, it will be possible to import all the IDF stuff as a CMake library, as in this example: https://github.com/espressif/esp-idf/blob/master/examples/build_system/cmake/idf_as_lib/CMakeLists.txt.
slender iron
I was considering it but I'm not sure it'll improve things much for us
sour lynx
Until then, there is an alternative: create a "hello_world" project with IDF, and build it with the IDF build system. Then add a CMake target to dump the CFLAGS and LDFLAGS of the main component into some file. Then you can read this file and use the CFLAGS and LDFLAGS with make.
slender iron
ya, I was considering that using a tinyusb example
sour lynx
The idea is that you write a Makefile which calls CMake to build this simple IDF project. This produces all the static libraries of IDF components. An extra rule in CMakeLists.txt saves the compiler and linker arguments to some file, which your makefile can read.
slender iron
I think @atomic summit prefers how loboris did his fork: https://github.com/loboris/MicroPython_ESP32_psRAM_LoBo
that sounds like a good approach
sour lynx
yes, although that was based on GNU Make build system (which was used in IDF 3.x and before)
i think you could do something similar with CMake
Let me try it locally and see if it works...
manic glacierBOT
slender iron
I'm not sure it's getting many changes now
do you have any pointers for integrating with freertos? that's the other thing I'm wondering about. right now we don't have an rtos at all
sour lynx
I'm not sure it's getting many changes now
@slender iron Yes, this fork has been dormant for a while...
Regarding FreeRTOS, you probably need some "main" task where the interpreter runs, this is the easy part.
slender iron
heh, ya and it's micropython code is old too.
do you recommend the idf as a submodule or is there another way we could pin ourselves to a version of it? I know there is a docker for CI
sour lynx
A submodule is not a bad option. Keep in mind that if you want your users to be able to rebuild circuitpython with a slightly different IDF commit (e.g. to try out a fix we have released), they will need to manually change the submodule commit.
sour lynx
OTOH, submodules give better control over the specific revision than the docker image. We only tag docker images for release tags and the tips of release branches, so if you want a specific commit on a release branch, then the docker image might not exist for that commit. Note that you can still used the docker image for the respective release branch. Just git checkout the right commit of IDF inside the docker image, before building.
slender iron
cool. we try to avoid things shifting out from beneath us.
sour lynx
The docker image is useful also because it comes with all the tools and python packages already installed; you can also run install.sh in your CI job, but this might take longer than fetching the docker image.
slender iron
I do want to make sure I set it up so that we can easily update versions
ya, we need to explore that for our other toolchains too. our arm install is non-trivial
are you using github actions for CI?
sour lynx
We don't have pre-made Github Actions for IDF projects yet, although one team member is working on it.
But the docker image is pretty close. Let me show an example...
slender iron
docker image is enough. mainly just asking about the provisioning
we're always looking to speed up our builds
slender iron
👍 nice!
sour lynx
This runs on a slightly different image, igrr/idf-qemu, but the only difference from espressif/idf is that there is Qemu added to the image.
slender iron
right
ok, so I'll 1) add idf as a submodule 2) write make rules to build the idf using cmake and output the flags and 3) link those libraries into the main build
any idea if the IDF works with LTO? we use it on arm for small builds
sour lynx
No, unfortunately LTO doesn't work at the moment because we place certain functions into IRAM (for reasons) and LTO can accidentally move them to Flash.
slender iron
oh interesting!
sour lynx
however Flash size is usually not a concern with ESP chips (since you have plenty of external flash), so the gains are not so important for the app
we are looking at enabling the LTO for the bootloader though, since it has to fit into RAM.
slender iron
we use it for the way we do translations now but we can fit both the english and translations into flash
I thought there was a ROM bootloader. is it just part of the loaded image?
we're keen on having a uf2 bootloader
sour lynx
There is indeed a ROM bootloader (a.k.a 1st stage bootloader), and the bootloader in flash (a.k.a. 2nd stage bootloader). The ROM bootloader is always there — and it always allows flashing the app over UART or USB (CDC/DFU). The bootloader in flash doesn't handle flashing. It is mainly responsible for:
- OTA updates — switching between copies of the app, when the app self-updates over Wi-Fi.
- security features, such as flash encryption, secure boot, rollback protection — although you will likely not use these in circuitpython
There is an overview of how IDF manages flash layout in the remaining space after the bootloader: https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-guides/partition-tables.html
sour lynx
You don't have to use partitions, you can access flash at any address directly, but it recommended to establish partitions for your data storage, such as FAT.
The minimal number of partitions you need for a functional application is probably 2: one for the application itself, another for "NVS", a non-volatile storage library which IDF uses, for example for Wi-Fi credentials storage.
slender iron
we usually do it through linker scripts at the moment
but we're pretty new to having circuitpython on an external flash chip
(though we do have the filesystem on internal flash in some cases)
sour lynx
I see. The flash memory on the ESP32 is not directly mapped into the address space. That is, it can be mapped by the MMU, but since MMU is configurable, there is more than one possible mapping. This means that theoretically you can store the application at any 64kb-aligned offset in flash, and the MMU will map it into the spot in the address space for which the application was linked.
This allows having two "slots" for the application in flash, and use them on a round-robin basis for OTA updates. The application runs from one slot, downloads a new version into the other slot, then restarts, asking the bootloader to switch to the new slot.
The application is linked in the same way, regardless of where it will be flashed. The MMU deals with mapping the application into the correct location in instruction and data memory.
There are some notes at https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-guides/general-notes.html, which you may find helpful. The address values are slightly different for the ESP32-S2 (we haven't updated the page yet), but the logic is the same.
(except that there is only one CPU)
slender iron
cool cool! thanks for all of the pointers
I'll likely stream when I work on it next so folks can give feedback. 🙂
I've gotta head to the store now
thanks!
manic glacierBOT
manic glacierBOT
slender iron
folks may be interested in my reply here: https://forums.adafruit.com/viewtopic.php?f=60&t=164439&p=807845#p807845
talking about longer term plans
slender iron
@onyx hinge hey, I'm trying to get protomatter going
onyx hinge
@slender iron ahoy
slender iron
import _protomatter
import framebufferio
import board
pm = _protomatter.Protomatter(32,
3,
[board.D6, board.D5, board.D9, board.D11, board.D10, board.D12],
[board.A5, board.A4, board.A3, board.A2],
board.D13,
board.D0,
board.D1,
doublebuffer=True,
framebuffer=None)
fb = framebufferio.FramebufferDisplay(
pm, width=32, height=32
)
print("hello")
looks plausible, assuming you didn't pull the latest changes to make all args kwargs-only
slender iron
this isn't doing anything for me. is the next debugging step a logic analyzer?
onyx hinge
no error, no LEDs?
slender iron
I've got a feather m4 and 32x32 display
nope
Adafruit CircuitPython 5.2.0-47-g5466ddcfd on 2020-04-13; Adafruit Feather M4 Express with samd51j19
downloaded it from github
didn't see your kwarg changes pushed
is that the right pinout for the feather?
onyx hinge
Author: Jeff Epler <jepler@gmail.com>
Date: Tue Apr 14 10:02:08 2020 -0500
Protomatter: Make all arguments kw-only, add rgb count and optional height checking
yes that pinout matches what I've been using
slender iron
kk
onyx hinge
ouch the typos in that commit message 🤢
besides looking for pulsing with scope or LA, another option would be to try the arduino protomatter demo.
onyx hinge
usually I just get a messed up display then, not no-display
slender iron
bingo
onyx hinge
it's possible that going straight to console mirror doesn't work, I don't think I tested that
oh really!
slender iron
no blinka but I see a >
onyx hinge
I have had missing blinka too
but I have really not concentrated on seeing the REPL..
@slender iron ok the push for that commit failed to build due to a spurious network error.... intersphinx inventory 'https://circuitpython.readthedocs.io/projects/register/en/latest/objects.inv' not fetchable due to <class 'requests.exceptions.HTTPError'>: 502 Server Error: Bad Gateway for url: https://circuitpython.readthedocs.io/projects/register/en/latest/objects.inv
but I'll try the re-run button
slender iron
onyx hinge
is only half the panel working?
onyx hinge
you could pass in a framebuffer of b'\xff' * 32 * 32 * 2
to get all white
and not construct a displayio
slender iron
I'm not sure my power supply can handle that
onyx hinge
and there may be bugs if you try to use it that way, I haven't since the very beginning
(I know there are bugs, but you can see pretty lights and usually not hit them)
slender iron
Turns out display ruler was already on the drive
onyx hinge
are the colors wrong?
onyx hinge
or just translating weird
ok that's what's in the file. looked weird, I'd never seen it before
deliberately lo-fi?
oh for pity's sake, re-run doesn't work because I'm still branched from the point where checkout@v1 was used.
slender iron
that's an old version
for eink
rebase time!
color looks wrong
blinka's eye is yellow
onyx hinge
there are two color conversions, first from the internal colors to RGB565, and then to (in your case) RGB333. Double conversion isn't ideal, and it may be causing color artifacts(?)
looks a lot like the blinka I see
slender iron
ya, the bottom right of the eye should be dark blue
tries to find the original image
onyx hinge
is that the only pixel out of place?
onyx hinge
All the LCDs swap the byte order of RGB565 vs what I use
is this drawing going through some path which .. doesn't?
.rgb888 = 0x0736a0,
.rgb565 = 0xf501,
slender iron
red seems weird
onyx hinge
this looks pre-swapped (from the palette of built in blinka)
onyx hinge
in displayio_colorconverter_convert I added code to swap(-or-not) color values when going to the RGB565 colorspace
originally, everything was swapped there
slender iron
right, so you changed the default
onyx hinge
but this kind of palette that the internal blinka image has, contains pre-swapped RGB565 colors, and those are getting out
so would it be anything with a palette?
displayio_palette_get_color
slender iron
so I must have standardized rgb565 to bgr565 actually
onyx hinge
whatever to call it, the LCD panels clearly like it
slender iron
well the LCDs can swap them in their init code
onyx hinge
this branch will break palette colors for current LCDs
slender iron
I had trouble just getting them all the same
onyx hinge
though the built in blinka bitmap will not show it
I took the swap out of displayio_colorconverter_compute_rgb565 and moved it up a level (and made it conditional)
but the way displayio_palette_set_color uses it, means it'll stop swapping the color around for LCDs that need it, giving wrong results
slender iron
right
onyx hinge
and the reason that the built in blinka is different is, all those values were precomputed so they could be in ROM, so they're right for LCD and wrong for Protomatter
Do you see the "right" place to put the swapping?
slender iron
the simplest things is probably to reswap in protomatter
if we change the built in blinka then we have to change all other display init
onyx hinge
that may be the best thing right now.
slender iron
it sucks that it's bgr but I had a bunch of trouble getting it all uniform
onyx hinge
I'm sad about doing so many more memory references for every refresh
slender iron
can you do the swap at the same time as the RGB333? then atleast it stays in a register
onyx hinge
alternately, I could move the swap down into protomatter C library where it would just be a register op, but that puts another pull request on the timeline and we have a timeline
(jinx)
slender iron
🙂
don't let a timeline compromise your work
dislikes clocks
for now we can hack it in cp if you like
there are so few pixels it won't really matter
onyx hinge
I think it's possible to merge this on schedule and then not drop the ball on fixing the stuff that was less than perfect.
slender iron
¯_(ツ)_/¯
it's not the end of the world to have it
I doubt anyone will notice
there are plenty of other places to optimize
onyx hinge
the imx framebuffers may force us to face this
slender iron
ya perhaps
pseudo pebble
Hey guys! I am loving CP so far... but I wonder what is the preferred way to track CP code in a git repo? Obviously dont want the repo to live on the microcontroller drive... so I'll need to hold it on y machine and move it over in a build step. Anyone have a preferred method of that with VS code before I just knock something together with build tasks / gulp ?
slender iron
@pseudo pebble I think that ladyada uses git on the fs 🙂
I usually just copy files back over when I'm done
@onyx hinge does palette just need to look at the swap bit in colorspace too?
onyx hinge
mmmmaybe, but how does that fix the rom'd palette?
if you show me what you were thinking about, I can show you what I was coding but haven't tested yet. it's probably better..
I was busy writing a big apologetic comment which is never a good sign
slender iron
the colorspace is dynamic from the display right?
onyx hinge
I think that's right
slender iron
so we 1) default it to swapped and have the cc and palette swap on out
onyx hinge
slender iron
and 2) change the rom representation to rgb565
onyx hinge
right, 1) is half-done, maybe we just need 1.5.
onyx hinge
I have a program that uses OnDiskBitmap only at this point
blinka looks good on pygamer, so does an OnDiskBitmap
OK, that's pushed
slender iron
kk. will look
onyx hinge
on pygamer anyway, an in memory bitmap also looks right
I just set the 0th palette entry to red, green, and blue and each one was right
slender iron
cool. I think we handled all of the cases
onyx hinge
I feel silly for not knowing blinka looked all wrong
onyx hinge
the purple was .. pretty close honestly
slender iron
easier to see on this display
onyx hinge
okay, I am hungry enough to eat a 🌮
onyx hinge
thanks for the testing! It's fine, I planned to totally front load this week for Protomatter's sake...
slender iron
kk, cool
onyx hinge
I'll be back this evening if you have more to tell me on whatever medium, also to deal with stuff that comes from guide moderation
👋
slender iron
kk
slender iron
@tulip sleet can you review https://github.com/adafruit/circuitpython/pull/2754 tomorrow?
slender iron
looks like CircuitPython is still using the Contributor Covenant
simple pulsar
@onyx hinge What res is that neopixel screen you're working on? It looks very cool.
onyx hinge
@simple pulsar it's 64x32. And it's not neopixels, it's this other thing. Sometimes called HUB75. https://www.adafruit.com/product/2277
that's the specific one I've been working with, Scott's got a 32x32 variant, and there are others
simple pulsar
I only ask because I've been messing around writing a program that generates test card images. I just tried it at 64x32 and it's not quite tuned for such a low resolution!
onyx hinge
interesting!
simple pulsar
It was inspired by a misbehaving CLUE board/display discussed in there: https://forums.adafruit.com/viewtopic.php?f=19&t=163189 - I just posted the 240x240 bundle which looks a lot better.
onyx hinge
@simple pulsar that's neat, I can almost see Limor in the tiny one after seeing the full size
idle owl
@tulip sleet It's actually quite simple to patch the CoC on the libraries without creating a lot of havoc. Please submit the fix to the Adafruit Community Code of Conduct repository and we can pull it from there.
tulip sleet
@tulip sleet It's actually quite simple to patch the CoC on the libraries without creating a lot of havoc. Please submit the fix to the Adafruit Community Code of Conduct repository and we can pull it from there.
@idle owl not worth doing separately; I'd wait until you have to do another pan-library edit
raven canopy
its only like 5 tiny steps. grab copy, stick it in a repo, commit, git format-patch, submit a PR, then run it... (ok. 6 tiny steps.)
tulip sleet
it's very unworth it to generate all those PR emails
i don't think anybody has tried to contact me via the old username
raven canopy
its only one PR. the patch is applied as a direct commit.
tulip sleet
one pr per library?
raven canopy
nope. one PR to adabot. the libraries are direct. (except NeoPixel, or any failed patches)
tulip sleet
ah, I see; i was thinking about the black PR's, but those were complicated
raven canopy
yeah, can't patch those since a single diff won't compute.
tulip sleet
anyway, I submitted a pr to the code of conduct repo; thank you
simple pulsar
Going back to the CLUE with the misbehaving screen, there's a bit more data on it which can be gleaned from the new images on: https://forums.adafruit.com/viewtopic.php?f=19&t=163189&p=807915#p807915
It's almost like it has a crazy high gamma / brightness setting and saturates at very low levels. Is there a chance the initalisation is failing? Or that a default value is relied upon and this is a variant of the screen with an unhelpful default?
simple swift
Trying to get a better idea of micropython vs circuit python. I keep seeing videos about the esp8266 but that is micropython is it not? Am I understanding that circuitpython is build and maintained by adafruit?
lone axle
@simple swift I believe the esp8266 can use micropython (but I'm not certain). Circuit Python is a community project that Adafruit created and supports heavily by sponsoring some of the lead developers. But it's not exclusively maintained by Adafruit, there are also many non-adafruit folks working on it. And there are roughly just as many non-Adafruit devices that now support Circuit Python as there are ones made by Adafruit themselves.
simple swift
@lone axle great answer thank you. Is it reasonable to assume circuit python is getting more support than micropython?
lone axle
I am perhaps not the best person to speak on it. Personally I am way more involved and familiar with Circuit Python than Micropython. But I don't think I would say that micropython is lacking support really. They are continuing to make improvements the core and have produced a few new versions of the flagship PyBoard devices.
There is a fair amount of carryover as well. For instance one of the newer features within Circuit Python is called ulab (micro-lab) which is a numpy-like module that allows quick math operations. I believe that was working with Micropython before it was more recently brought in to Circuit Python. So it's an example of improvements to one leading to improvements in the other.
main meteor
It's nice to see: instead of a fork that keeps diverging, it's a fork that stays somewhat parallel and improvements in one get propagated to the other.
simple swift
Ah ok thanks again so much to learn I just want to have my python app start talking to hardware
timber mango
The STM32F405 Feather can run Micropython.
I prefer Circuitpython though because the hardware support for sensors and such is better.
ionic elk
@slender iron is there a specific method to avoiding merge conflicts with translations? It seems like I'm having to constantly go in all these files and change that one line that conflicts every merge
but I don't see them in other PRs that often
slender iron
@ionic elk ya sit tight. On my phone. Will send shortly
ionic elk
Having a failure in the... unix port too now? A recent merge maybe?
slender iron
@ionic elk when merging git checkout --ours -- locale/*
then make translate
the first line gets you back to the merge version
basically reruns your addition (I think)
ionic elk
good trick
slender iron
I should make sure it's preserving the master version
ionic elk
Dan mentioned it might be beneficial to just get rid of the PO Created_date line altogether since it's the only thing that conflicts
which seems reasonable to me
onyx hinge
@slender iron I'm pleasantly surprised at how well OnDiskBitmap performs in this protomatter demo, fwiw. Portions of about 9 20x20 tiles from an OnDiskBitmap are visible at a time, and I'm getting "lotsa" FPS refresh rate.
raven canopy
RosiePi Lives! (shoutout to all of you video content creators...this stuff is tedious)
https://www.youtube.com/watch?v=nN3rocc4dSQ
ionic elk
@slender iron when you get a sec could you check out this unix port CI failure? https://github.com/adafruit/circuitpython/pull/2735/checks?check_run_id=589252573
onyx hinge
@ionic elk your commit (unintentionally, probably!) updates the "ulab" submodule which is probably the cause of this https://github.com/adafruit/circuitpython/pull/2735/files#diff-f08a2dcbca771712247d5e774c01ec57
ionic elk
@onyx hinge again?? I thought I fixed that... is this another failure of my git fu?
I'm not touching the submodule
onyx hinge
I don't blame anyone for having trouble with git, especially where submodules are concerned.
ionic elk
does it get changed on merges or something?
during a git fetch upstream and git merge upstream/master event?
onyx hinge
When it comes to submodules, it feels like there are a number of gotchas. It's hard to guess which one you got got by
ionic elk
Also thanks for pointing it out I would not have guessed. The only way I can seem to get it to be correct is to manually navigate to the submodule and checkout the version I see on github...?
onyx hinge
If you don't update your submodules with git submodule update, they will continue pointing at whatever ref they were at, when you pull, change branches, rebase, etc.
ionic elk
I run git submodule update --init --recursive every time
onyx hinge
If you then use a bulk commit command like "git commit -a", it will record the ref that was checked out and include it in the commit
then my story doesn't apply to your workflow
tulip sleet
@ionic elk is your version of git up to date? it's possible something is fixed
I have 2.25.1
onyx hinge
the unintended change came in as part of a89928c13c748b7f25bd4e734cfe487b0020a25d "Implement requested changes"
ionic elk
2.24.1
tulip sleet
that should be good enough
what i see is that sometimes (all the time?) a merge from upstream doesn't roll forward the submodule commits; is that the essence of the problem here?
ionic elk
@onyx hinge I'm just so puzzled by that because I run a git status after every git add . so it weirds me out that a submodule would be affected without me knowing
tulip sleet
I use this alias:
alias gitsubupdate='git submodule sync --quiet --recursive && git submodule update --init'
i have dropped --recursive on the second half because it seems to mess up tinyusb sometimes
ionic elk
@tulip sleet the essence of that being that I need the submodule sync first? because it's not coming in with the merges?
slender iron
--recursive brings in a bunch of tinyusb submodules we don't need
ionic elk
and yeah I also get issues with tinyusb a lot, I'll try your version
onyx hinge
@ionic elk speculating about what caused the problem aside, do you know how to resolve the problem, or would you like a hand?
ionic elk
I just went ahead and manually checked out the submodule to the master version and committed it, and it's resolved the file difference
we'll see if it makes the Unix problem go away too
onyx hinge
since the errors were about ulab I think the odds are in our favor
ionic elk
Well, thanks all for the help!
onyx hinge
you're welcome
at $OLD_JOB some days it felt like I was simply on call to help developers with their submodule problems. Only 22% joking.
idle owl
@tulip sleet What sommersoft said. It doesn't generate PRs. But we can wait. I'll look at your PRs in a bit here.
raven canopy
PyCon 2020 videos are starting to arrive. 4 so far..
https://www.youtube.com/channel/UCMjMBMGt0WJQLeluw6qNJuA
tulip sleet
@onyx hinge github actions seems to be stuck at the moment. queued jobs are not being run. I'll keep an eye out
onyx hinge
@tulip sleet I appreciate it, thank you!
I'll be AFK for lunch soon but I'll keep an eye on my notifications on all the systems
ionic elk
In my latest CI check failure I appear to have lost bool support on some boards?
@onyx hinge any chance this is related to new protomatter changes?
They aren't my boards, it's a bunch of SAMD ones
nevermind I think it's another submodule issue
tulip sleet
@onyx hinge I tried to restart #2774; it cancelled some but not all of the jobs and now is in some kind of limbo. If we can get #2775 to build all the way through, I'll approve that and we can close #2774 as subsumed by #2775
onyx hinge
@tulip sleet thanks. I had pushed to 2775 the responses to the 2774 review comments anyway, because I wasn't paying attention.
onyx hinge
thanks
lone sandalBOT
tough flax
Hey folks... I screwed up... I meant to order 10 esp32-s2 wroom modules from <@&453561466486915072> and instead ordered the raw chips. Does anyone working on esp32 support want these?
ivory yew
I'm getting a linking error when building the uf2 bootloader for winterbloom_big_honking_button from master:
section .binfo LMA [0000000000001ff0,0000000000001fff] overlaps section .relocate LMA [0
000000000001e90,000000000000205f]
same with others, hrm.
onyx hinge
Still seeing actions hanging for #2775 -- https://github.com/adafruit/circuitpython/actions/runs/79242680 "62 completed and 38 queued jobs" 😦
2020-04-15 20:50:49 (99.3 KB/s) - ‘gcc-arm-none-eabi-9-2019-q4-major-x86_64-linux.tar.bz2’ saved [116802378/116802378]
hmm the one job in progress spent 20 minutes downloading this file DSL-slow
2020-04-15 20:25:08 (97.0 MB/s) - ‘gcc-arm-none-eabi-9-2019-q4-major-x86_64-linux.tar.bz2’ saved [116802378/116802378]
1/1000 the speed of this other download of the same asset
"checkout" took 10 minutes
and then suddenly all the rest of the jobs started and finished
tulip sleet
@onyx hinge I was wondering if we were being rate limited due to the new GitHub pricing (better for us), but sounds like not.
orchid basinBOT
raven canopy
@slender iron what branch were you having frequencyio issues on? 5.x, master, lower_power?
raven canopy
k. starting to dig in...
ionic elk
@slender iron I missed your top level comment, sorry! I'll work on that first thing tomorrow.
my b
lunar crown
Minor bug on Pyportal Titano
Pyportal Titano
UF2 Bootloader v3.9.0 SFHWRO
Model: PyPortal M4 Express
Board-ID: SAMD51J20A-PyPortal-v0
Adafruit CircuitPython 5.2.0 on 2020-04-09; Adafruit PyPortal Titano with samd51j20
The Docs for adafruit_pyportal say
# Backlight function
# Value between 0 and 1 where 0 is OFF, 0.5 is 50% and 1 is 100% brightness.
The backlight seems to be totally extinguished between 0.0 and 0.4.
Very dim at 0.45, increases linearly as expected from 0.45 to 1.0.
So, if you follow the examples in “Making a PyPortal User Interface with DisplayIO” and set the backlight to 0.3, you end up with a dead screen, rather than a dim screen as the example would imply.
I didn’t see any open issues that seemed to apply to this.
--- Graham
slender iron
uf2 on stm?
@ionic elk yes. so the process is the same for all boards
ionic elk
@slender iron something I need to take care of or something you're working on atm?
slender iron
not urgent. just something I'd like to see
may see if @gentle bronze wants to help with tinyuf2
ionic elk
Ohhh right you mean the UF2 bootloader doesn't support H7s right now
slender iron
does it support any stm?
ionic elk
sure, the F4s
UF2 is enabled for the Feather
and any other F4 board that it can reasonably fit on (ie ones with external flash)
slender iron
I may have messed mine up with jlink
ionic elk
I see you meant "misses" as in "I miss it lets bring it back" not "this got missed"
slender iron
ya, I miss it
crimson ferry
Did something in a post-5.0.0 version of CP break this? ```
# ...
# logbox is a displayio.TileGrid(terminalio.FONT.bitmap`
logterm = terminalio.Terminal(logbox, terminalio.FONT)
print("Testing", file=logterm, end='\r\n')
tulip sleet
@slender iron hi, when you're on, could I talk to you about BLE advertisement prefixes? I'm confused about their length fields.
manic glacierBOT
tulip sleet
I'm inquiring in GitHub community forum re stalled actions: https://github.community/t5/GitHub-Actions/Multi-job-check-actions-stalling-with-many-queued-jobs/m-p/54274
but maybe I should contact support directly .. ok, I did
slender iron
@tulip sleet on my laptop now but not at my desk
tulip sleet
can wait
slender iron
spook is being slow about eating his breakfast
slender iron
and now vin is on my lap...
onyx hinge
I wasn't intending to work today, but I did spend time looking into whether we can enable "1 == array" in ulab. It requires code in micropython which isn't in circuitpython. There's not "just one" cherry-pickable commit that adds the core code necessary, so I think we'll have to put it off until/unless we put in the time to merge micropython into our tree again.
@lapis hemlock ^ you may be interested (but dismayed) to know
slender iron
@tulip sleet at my desk now
slender iron
@onyx hinge looking at your protomatter pr now
onyx hinge
🔍
slender iron
are there two? 2775 does have height in RGBMatrix's constructor
tulip sleet
@slender iron ok, just finished lunch if you are free
slender iron
yup!
lapis hemlock
I wasn't intending to work today, but I did spend time looking into whether we can enable "1 == array" in ulab. It requires code in micropython which isn't in circuitpython. There's not "just one" cherry-pickable commit that adds the core code necessary, so I think we'll have to put it off until/unless we put in the time to merge micropython into our tree again.
@onyx hinge I am open to any reasonable alternatives, e.g., if you conditionally want to switch off1 == arrayincircuitpython, or I could factor out the code for this into a separate function, and then we could attachequal: https://docs.scipy.org/doc/numpy/reference/generated/numpy.equal.html. At least,circuitpythonwould then have a workaround. Unfortunately,==andequalare not exactly the same thing innumpy...
onyx hinge
@lapis hemlock we can also just live with the limitation for now
lapis hemlock
@lapis hemlock we can also just live with the limitation for now
@onyx hinge The snag is that it is not simply a limitation, but the function behaves entirely differently in the two environments.
ionic elk
@slender iron analyzing the .bin file from the H7, it's normal right up to the end - it's just got 3.2 billion extra zeroes.
slender iron
@ionic elk ya, I'm looking now too
ionic elk
there's nothing actually at the end, though, you could chop it all off and it'd be fine
slender iron
it's the .dtcm_bss section that is the issue
imx gets around it by explicitly including sections in the bin
ionic elk
ah
slender iron
-j includes a section explicitly
the stm makefile includes all sections it thinks it needs
ionic elk
I see, this binary -j .flash_config -j .ivt -j .text -j .ARM.exidx -j .data -j .itcm -j .dtcm_data $^ $@?
slender iron
right and it doesn't include dtcm_bss
haha, gross: https://stackoverflow.com/questions/6252812/what-does-the-aw-flag-in-the-section-attribute-mean
ionic elk
I just swapped in $(Q)$(OBJCOPY) -O binary -j .isr_vector -j .text -j .data -j .itcm -j .dtcm_data $^ $@ and it works/has a reasonably sized bin
slender iron
its the "progbits" that causes it to end up in the bin
ionic elk
any reason I shouldn't just plop a flag in there and swap it?
slender iron
+1 just include the flags to be explicit
ionic elk
could you clarify?
slender iron
include the -js
the line above looks right
ok, I gotta exercise. will keep testing after lunch
ionic elk
Still don't totally get what you mean but I'll put something together and just push it so you can suggest what you're thinking of
Sorry still learning this makefile stuff
slender iron
I think you are thinking what I'm thinking. change the objcopy lines to list the sections to include
ionic elk
Ok new fix is in
manic glacierBOT