A GitHub release..?
#circuitpython-dev
plucky flint
plucky flint
IIRC you have some automation stuff that builds / deploys things.
Yes I do.
(have access to that circup repos)
idle owl
Ok. Yeah so go to https://github.com/adafruit/circup/releases and click "Draft a new release"
idle owl
Once you make a release, that's when the automation kicks in.
plucky flint
So, you want me to do a draft release, or just a regular release (I do releases all the time with other projects / gigs)? Just wanna know I'm doing what's right for you folks.
idle owl
For the release notes, please include what updates have been made, and you can thank Steven there as well.
Do a regular release, bump it to 0.0.8 to keep up with where you're at
plucky flint
All release notes are in CHANGES.rst
idle owl
Ah ok, fair enough
plucky flint
π
idle owl
Well include something in the release notes anyway π
idle owl
Since I have no idea what it does with empty release notes. Probably fine, but we avoid it.
plucky flint
idle owl
Perfect! Thank you!
plucky flint
Does this get automagically pushed to PyPI as a pip installable package..?
idle owl
It should, yes.
idle owl
Yes that as well goes with any updates.
plucky flint
Stonking stuff... Automate all the things... I, for one, welcome our new automated build script overlords. π
idle owl
@plucky flint
plucky flint
Right... I'm AFK, helping Mary (my wife) to set up for a remote instrumental lesson she's giving.
idle owl
@plucky flint Alright, have a wonderful day!
onyx hinge
@slender iron there's that array primary_display_t , which has a union-of-busses and a union-of-displays. Is it guaranteed that the i'th bus and i'th display are related?
(and if so, where is that guarantee maintained?)
idle owl
@lone axle If I make a simple change to Adafruit_CircuitPython_Bitmap_Font can you test it? It's removing 3.0.0 backwards compatibility from the main file.
lone axle
@idle owl yep I should be able to test it. Probably not until a bit later this evening though.
idle owl
That works. Thanks. I'll tag CPLibians on the PR and you specifically as well.
lone axle
Okay cool ty.
slender iron
@slender iron there's that array
primary_display_t, which has a union-of-busses and a union-of-displays. Is it guaranteed that the i'th bus and i'th display are related?
@onyx hinge No, I don't believe so.
ionic elk
is there a way to declare a static array at a specific memory address? I could relocate my H743 filesystem cache to one of the AHB SRAM banks (128 each) but I don't really get how to do it... Micropython seems to fake it and their implementation is pretty complex.
timber mango
Can I get some help on making a python program display the contents of a txt file?
@timber mango You will probably find more and better help on #help-with-circuitpython . π
onyx hinge
@slender iron I can take notes today
slender iron
thanks @onyx hinge !
timber mango
If somebody closes an issue or PR, could you please add a comment saying why this was done? Thank you. Without knowing why this was done it is making me feel like I do not want to bother anymore and feels like a slap in the face.
pastel panther
@timber mango I thought my explanation was pretty clear. I don't think offense is an appropriate reaction but that is your choice. We value all contributions.
timber mango
@pastel panther I had not seen your last comment yet. I still disagree but it does not matter.
pastel panther
@timber mango it matters if you feel like you've received a slap in the face, and if you disagree with our choices I am happy to discuss it at a later time
slender iron
@ionic elk let's sync up later and we can chat linker scripts
serene warren
Hey y'll, gonna be lurking today - as usual.
modern wing
I may be joining late, and if so, I'll be happily lurking π
raven canopy
I have a conflict, and no real notes. Group hug to all!!
serene warren
Red text?
ivory yew
I'll be lurking today
half sedge
lurking
modern wing
Indeed -- stay safe, do what you can to be happy, and help each other if you can, when you can. 
Indeed -- stay safe, do what you can to be happy, and help each other if you can, when you can.
serene warren
Hey y'll, gonna be lurking today - as usual.
onyx hinge
I'd better get started listing lurkers in the doc
serene warren
OK figured it out. Red text = try again. Not sure why it happened in first place.
modern wing
It's discord -- glitches be abound in times of high usage.
I had the same thing happen -- patience...wait a few, try again π
serene warren
Hello
errant grail
Hello! I'll be text only today.
timber mango
lurking today
inland tusk
Hello I am lurking today
turbid radish
Lurking, working on tomorrow's CircuitPython Newsletter
lone axle
Thank you π had no idea the "Let's Check" button on the settings page unmuted me in the room.
bleak tiger
lurking today thanks
lone axle
That is probably how I've wound up accidentally un-muted a few times lately. π
serene warren
Notes doc today?
tidal kiln
lurking
serene warren
Tanks π
onyx hinge
@slender iron maybe mention to order adafruit stuff from digikey right now (somebody's adding it to the doc right now)
crimson ferry
Notes doc is now in the pinned messages each week
simple pulsar
lurking occasionally
turbid radish
Fill Phil's shoes
onyx hinge
ivory yew
yay f-strings!
lone axle
stevenabadie did some of the recent work on circup.
idle owl
@gilded cradle I have a question from JP: with displayio - it's very slow to switch out a line of text from, say "Wave to start" to "Scrub time: 20" the first time it is used. It's super fast the second time, onward. Is there a pre-load of glyphs or something to be done to make it equally fast the first time?
idle owl
File an issue on the newsletter repo, submit a PR, or even email us.
crimson ferry
@idle owl font.load_glyphs(b'ABC123') ?
turbid radish
Send me things via Twitter at @anne_engineer
ivory yew
Steak of CircuitPython
solar whale
state of the snake
ivory yew
π I'm always happy to do code reviews
idle owl
@crimson ferry Possibly? Where does the font come from in that?
onyx hinge
@simple pulsar lurking or talking today?
crimson ferry
@idle owl font = load_font('ondisk.bdf')
gilded cradle
@idle owl you can preload the characters ahead of time and it won't seem so slow. JP can look at the Apple Media Service example I collab'd with him on for an example.
onyx hinge
@idle owl thanks
onyx hinge
I'll get back on task
slender iron
https://docs.google.com/spreadsheets/d/1FdapBL3wIJf5T_LqwEcglIMMMRelu_9UjnY62IaQBFA/edit?usp=sharing
ivory yew
I'm better at reviewing Python than C. π
simple pulsar
Just lurking, with a bit of eating and bit of guide authoring.
slender iron
@simple pulsar did you want to review my midi library changes? (or did you already?)
tidal kiln
@idle owl if using with adafruit_display_text need to be sure to also include M for spacing:
https://github.com/adafruit/Adafruit_CircuitPython_Display_Text/blob/master/adafruit_display_text/label.py#L112
idle owl
@tidal kiln Not sure how it's being done - so thanks, I'll pass that on in case that's how.
slender iron
@zinc thistle are you lurking?
zinc thistle
Yup...discord newb
onyx hinge
@zinc thistle we're happy to have you listen in
gilded cradle
@zinc thistle feel free to drop back into the voice chat
idle owl
@zinc thistle No worries π We all were at some point.
zinc thistle
haha...thank you...I'll drop back in for a min...
simple pulsar
@slender iron looks ok
half sedge
@slender iron About FPGA, I have a TinyFPGA BX and an Icebreaker "collecting dust", so I would love to make some use of it. Don't hesitate to share CircuitPython FPGA stuff.
slender iron
@half sedge nice! I just got an icebreaker. are you on the 1bitsquared discord?
half sedge
Nope, but I have a very low attention span for the moment. π¦
onyx hinge
As a counterpart to "keep it up", remember to share the load, which means to take a break when you need it.
idle owl
@onyx hinge Exactly. That's the best part about having a team. We can back each other up when needed.
simple pulsar
?
ionic elk
Aw dang now I have to get animal crossing
slender iron
I got it last night and have been liking it
lone axle
π
simple pulsar
Is the process to update the bootloader well documented for a typical user?
half sedge
@tulip sleet I have been trying to get a BLE oxymeter, but found none in stock with reasonable delivery time. But advise on product with CP potential are welcome.
tulip sleet
@half sedge The BM1000C models (maybe BM1000E also?) are what work well. But the US stockers of these are sold out, so the ebay vendors that claim to have them are all in China. EU may be different, but probably also sold out.
half sedge
I am stable but in a country that only test severe case. So just staying home, waiting and monitoring my health.
tulip sleet
The CMS50D-BT is very flaky.
slender iron
@simple pulsar for updating the bootloader there is usually a uf2 that you can drag over to update it
ionic elk
ShenZhen IO and Oxygen Not Included also come to mind
More good games that are like work but harder and you have to pay for them π
half sedge
@tulip sleet Probably sold out too or delivered in 6-8 weeks where I hope not to need that anymore!
gilded cradle
@ionic elk NinjaFlex is supposed to work well in the e3D Hemera hotend on just about any printer
ionic elk
Currently unavailable due to COVID-19 supply issues. Returning soon. NUUUU
dag nabbit dangit
gilded cradle
Aww, ok.
lone axle
@ionic elk It's definitely not the cheapest option but I have a Lulz Bot Mini and have had good luck using the "FlexyStruder" tool head with it to print ninjaflex.
onyx hinge
inland tusk
@ionic elk check with noe and Pedro. They use that filaments
turbid radish
MacBook Pro ESC keys
ionic elk
who are noe and pedro
onyx hinge
first photo is some veggie broth
turbid radish
Noe and Pedro do 3D printing for Adafruit. They have a show on it Thursdays
lone axle
I like the lego castle case
tulip sleet
also ask in #help-with-3dprinting
tulip sleet
@ionic elk I have a FlashForge Finder that the Ruiz bros (Noe and Pedro) say works well with NinjaFlex (I have not tried it). It does not have a heated bed. I bought it good-as-new from somebody on craiglist who checks and refurbs Amazon returns of FlashForge printers, bought from someone else (!) on craiglist, both are local
turbid radish
Thank you @idle owl for all your work on the weekly videos
ionic elk
CATALINAAAAA
idle owl
@ionic elk Pretty much. :sigh:
lone axle
I will update some devices tonight to 5.1.0 rc0
idle owl
@turbid radish You're welcome. Thanks for your help with uploading it. JP was a huge help.
serene warren
Great meeting. Keep safe to all.
errant grail
Thank you. Stay safe!
onyx hinge
@turbid radish clocks https://discordapp.com/channels/327254708534116352/327298996332658690/694255638062301185
turbid radish
Perfect thanks jeff
onyx hinge
(I don't know if those internal links really work, I've had mixed luck -- try https://youtu.be/5wBRmIr8cD4 too)
fierce girder
Sad that Europe has gone daylight savings, being an hour closer was nice the last couple weeks
solar whale
do you have a link to the calendar?
slender iron
fierce girder
For awhile there it was only 5 hours ahead of to New York, so nice π
ionic elk
@lone axle does the flexystruder come with the lulzbot or is it separate?
lone axle
It is separate (and a few hundred $ iirc)
ionic elk
I'm leaning toward the Lulzbot mini 2 or the FlashForge 3D Printer Creator Pro, but I'm open to other suggestions
lone axle
the stock extruder has the drive gear a little bit too far away from the hot end. So flexible filament can loop over and start coming out the side instead of getting forced down into the hot end.
The flexy struder changes the design a bit to move the drive gear basically as close as it can get to the hod end so that the fillament cant go anywhere but down and in to the hot end.
I've got a Lulz Bot mini 1 but I've seen the 2 in use and def have eyeballed it a few times. The belt driven axis's make it way quieter and a fair bit faster moving in the Z orientation.
ionic elk
On closer inspection it doesn't appear the mini 2 even has a flexystruder attachmetn
Hmm it says it just prints them out of the box though I guess
lone axle
Definitely not the cheapest machines (especially now adays). But I really like them for their "Plug and Play" nature. I was printing my first print within 10 minutes of unboxing. And even all these years later with the Lulz Bot mini 95% + of my prints are load it up click print and come back later to a finished part.
idle owl
@tulip sleet Is BLE_iBBQ ready for the bundle?
tulip sleet
sure, thanks, jp hasn't tried it yet, but it works fine for me, and we can always update it as necessary.
idle owl
@tulip sleet Alright. Thanks.
lone axle
I've got a few other printers as well, and have experience with several that aren't mine and I find that most of them require lots more tweaking and tinkering in order to get them working properly.
idle owl
@tulip sleet helper or driver
tulip sleet
same as heart_rate, etc. (I forgetwhich)
idle owl
ok I'll check thanks
lone axle
Oh nice, I didn't realize the 2 has a tool head that does flexy out of the box. That is cool. Definitely do heed the caution about the PEI build sheet + ninja flex though. Ninja flex sticks EXTREMELY well (too well in fact) to the PEI build plate. You definitely need to either use plain glass, or give it a good smear of PVA glue on top to act as a layer inbetween
ionic elk
@lone axle guess the Lulzbots are out of stock pretty much everywhere though
idle owl
@tulip sleet I tagged you as a reviewer on the PR. Obvs it needs to go through the checks.
lone axle
@ionic elk darn, I could see that though, they recently got taken over by a new company in North Dakota I believe. Moving their operations from CO. If you have a Microcenter nearby you might check there, my local one has usually got Lulzbots in stock.
I originally ordered it from Amazon I think. Though it was long ago, not sure if they still sell them or whether the recent takeover would have affected their supply.
idle owl
oh right. Release.
idle owl
@tulip sleet It's failing Run actions/checkout@v1 with git checkout --progress --force d310cce85d178b1b4e69005219f1d84788cfeef0 ##[error]fatal: reference is not a tree: d310cce85d178b1b4e69005219f1d84788cfeef0 Removed matchers: 'checkout-git' ##[error]Git checkout failed with exit code: 128 ##[error]Exit code 1 returned from process: file name '/home/runner/runners/2.165.2/bin/Runner.PluginHost', arguments 'action "GitHub.Runner.Plugins.Repository.v1_0.CheckoutTask, Runner.Plugins"'.
This is one I've never run into.
tulip sleet
re-running, that might be temporary
idle owl
I tried. I'll give it a few minutes and try again.
tulip sleet
are you up-to-date from upstream?
tulip sleet
fetch and merge from adafruit/master
slender iron
@ionic elk don't order from digikey yet. what imx rt boards do you have handy?
slender iron
ok, you'll need to make another order then. π get a 1010, 1050, and 1060. once we have an stm32f7 foundation we'll have you swap to imx
idle owl
It passes on my fork. Sigh. Rerunning again.
@tulip sleet Still failing. I don't understand. It passes on my fork of the bundle.
I'll trash the PR and recreate it. I guess see if that works.
idle owl
@tulip sleet That worked apparently.
tulip sleet
i am baffled
the issue supposedly fixed in checkout@v2, but we had trouble with that in circuitpython. We could try it in other places that put less stress on the github actions system
@onyx hinge If you feel there's a specific thing to complain about for checkout@v2, an issue in https://github.com/actions/checkout would probably be helpful to them
onyx hinge
@tulip sleet In the end the failures were just so weird I didn't even understand them. I'll try to characterize it better if I try again someday.
slender iron
@ionic elk ready when you are
@idle owl when you have a chance can you share your meeting notes template with me. I don't think you have yet
idle owl
@slender iron Oh. Yeah for sure.
slender iron
thanks!
I should have asked a long time ago
Here is the notes document for next Mondayβs CircuitPython Weekly meeting. Everyone is encouraged to attend! Please add your hug reports and status updates even if youβll be attending the meeting - itβs super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and weβll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1YZYYrODJwUpKD4PWWzjpPhzBAvf5Mf196r4z7uo6dns/edit?usp=sharing
slender iron
Here is the notes document for next Mondayβs CircuitPython Weekly meeting. Every...
ionic elk
@slender iron shoot sorry I closed my discord tab without realizing it and missed your pijng
I'm around now
slender iron
np, works for me. want to video or just audio?
ionic elk
Just audio is ok unless there was something you wanted to pull up
#define PLACE_IN_DTCM_DATA(name) name __attribute__((section(".dtcm_data." #name )))
STATIC uint8_t _flash_cache[0x20000] __attribute__((aligned(4)));
lapis hemlock
@onyx hinge I am here now.
onyx hinge
hello
lapis hemlock
Peek-a-boo.
So, the question is, whether both ** and power should be implemented.
Or one of them will suffice.
lapis hemlock
Then you can't factor it out, because it will be a method of ndarray.
onyx hinge
array([ 0, 1, 4, 9, 16])
``` _cough_okay, then make it power(), I was choosing ** not for any particular reason
3, 3 ways to take a power in numpy
lapis hemlock
** is more elegant, and I see arguments in favour. The question really is, whether it is worth the extra 1.5 kB.
onyx hinge
** costs 1.5kB more than power?
lapis hemlock
arctan2 cost around 1.5 kB, and the structure of the problem is very similar.
onyx hinge
or just in general either one costs 1.5kB
lapis hemlock
No, what I meant is that it would cost around 1.5 kB, and if it is **, then you can't switch it off.
onyx hinge
I think either way will be fine in circuitpython. just a matter for documentation.
I need to go afk
manic glacierBOT
Looks like the timing issues were caused by the code being in flash, moving it to ITCM and making sure no flash functions are called in the timing-critical section fixed the issue :)
Closes #2476
raven canopy
listening to today's meeting: i play a fair bit of Destiny 2 these days. hmu for Steam friending.
lone axle
I've never run across a script quite like this: https://github.com/adafruit/Adafruit_CircuitPython_Bitmap_Font/blob/master/examples/bitmap_font_simpletest.py
it is accessing sys.argv[1] and other things that make it seem like it's for CPython, and not CircuitPython... Is that the case?
slender iron
@lone axle I think it is for CPython
lone axle
Thank you
timber mango
I really like the idea of combining various modules with an MCU on the same Feather, such as the Feather M0 RFM69 and M0 WiFi. I really hope there might be M4 versions of some of these in the works at some point. π
crimson ferry
There are radio 'Wings for a lot of the variants.
I could go for a CircuitPython-capable LTE FONA (preferably Feather form factor).
The ESP32 S2 is likely to facilitate a single-board CP MCU+WiFi.
bitter bronze
I have tried following the suggestions of Radomir, but have not been successful. I end up with a MAKECODE drive instead of the CIRCUITPY drive with current.uf2 with no date. There is a SPIFLASH folder which is empty. When all the lights are green, there is no drive listed in File Explorer. When I single click reset, the led next to the "On" LED is bright red for a few seconds, then switches to green and I then see the MAKECODE drive. Copying the adafruit-circuitpython-circuitplayground_express-en_US-5.0.0.UF2 file to the MAKECODE drive I see the file on the MAKECODE drive. I was expecting the drive to disappear at that point based on the directions, but that does not happen? I am missing some understanding and I am not sure what questions to ask now?
crimson ferry
@bitter bronze What happens when you double-click the reset button?
bitter bronze
Double clicking reset shows redlight for a few seconds and then the same light is green. Drive listed is MAKECODE.
tulip sleet
@bitter bronze If you are running Makecode, just single click to get into the bootloader. Then you should see CPLAYBOOT. Drag the .uf2 to that.
crimson ferry
Sorry, didn't realize MAKECODE had different behavior.
tulip sleet
it's confusing that it's different
bitter bronze
As far as I can tell, MAKECODE is not running. When disconnecting and reconnecting the CPX again, all the lights are green and no drive is seen either MAKECODE or CPX. Clicking reset once brings up the MAKECODE drive.
crimson ferry
What Windows version, and did you need to install drivers? https://learn.adafruit.com/adafruit-circuit-playground-express/adafruit2-windows-driver-installation
bitter bronze
I am running Windows 10. I was wrong. MAKECODE was running. Stopping MAKECODE, unplugging CPX and plugging CPX in again (all green lights), clicking reset, I again see the MAKECODE drive.
tulip sleet
if you single press reset, do all the neopixels go green?
as opposed to double-click
i have a vanilla CPX here, not seeing MAKECODE with a simple MAKECODE program. I wonder if yours has an odd bootloader for some reason. What files do you see in the MAKECODE drive?
i see CPLAYBOOT if I single-click, or if i plug in the board fresh
bitter bronze
yes single click, all green, and MAKECODE drive disappears. I am sure that I am doing some thing the wrong order but I can't figure out what. DIsconnect CPX, reconnect CPX, all lights come on green, sigle click reset, shows one bright red light for a few seconds and then it turns to led red (I am color limited, so she said that it is red and not green as I stated earlier. It is next to GND, The smaller led nearer the edge is green.
tulip sleet
when you single-click, is there a pulsing led to the right of the usb connector (with usb at top)?
just yes or no is fine
bitter bronze
single click results in the D13 light slowly pulsing red (I think).
all other lights are green
tulip sleet
ok, good, so that's the bootloader, but you're not seeing CPLAYBOOT for some reason, so let's debug that. Are you running any third-party disk utilities like Acronis backup, Hard Drive Sentinel, or are you using Kaspersky antivirus?
bitter bronze
I am running Kaspersky and NovaBACKUP.
tulip sleet
Kaspersky can mess up the BOOT drive. https://learn.adafruit.com/welcome-to-circuitpython/troubleshooting#cplayboot-trinketboot-featherboot-or-gemmaboot-drive-not-present-20-5 https://learn.adafruit.com/welcome-to-circuitpython/troubleshooting#windows-explorer-locks-up-when-accessing-boardnameboot-drive-20-12
do you have another machine you could try this on that's not running kaspersky?
also if you have a really old version of our drivers installed on the machine, that could be an issue, as described in the link above
but it would be several years old
bitter bronze
OK, I will look at the links. I will try that later on another machine this evening. Need to go now. Thanks again.
tulip sleet
sure, good luck!
manic glacierBOT
timber mango
Now, I just need to get a few SD card breakout boards and/or AdaLogger FeatherWings. Not anytime soon though. π¦
slender iron
@atomic summit nice esp-idf video! I too would love to understand cmake π
atomic summit
Cheers @slender iron - CMake seems unnecessarily verbose and totally awkward, and the general docs, including those specific to the IDF are PANTS!
I hope the term PANTS is something you use in the US. If not...
PANTS means "Rubbish" or "Trousers" or "Underpants"
lone sandalBOT
onyx hinge
@atomic summit I've never heard of a system for controlling the software build process that is universally liked. The ones in wide use must each have their strengths, I figure, so I try to be positive about them.
sometimes it hurts that C/C++ don't have a unified "story" for building software like so many newer languages have.
but at this point I don't think it can be fixed for C
these actions failures are really frustrating. Unable to connect to azure.archive.ubuntu.com:http:
timber mango
PANTS means "Rubbish" or "Trousers" or "Underpants"
@atomic summit I like it! I also like the Aussie terms of "stuffed up" (not working) and "fallen over" (completely SNAFU)." π π You may know me as hybotics on Twitter and elsewhere. π
@onyx hinge The only reason make hangs around is that it has been there forever from the Bell Labs days of UNIX. π
solar whale
@onyx hinge The only reason make hangs around is that it has been there forever from the Bell Labs days of UNIX. π
@timber mango so have I π
onyx hinge
-536 bytes free in flash firmware space out of 253440 bytes (247.5kB).
OK now we get the real errors. Looks like select boards will need Protomatter disabled, but that's not unexpected.
@slender iron if I find I need to selectively enable FramebufferDisplay then it needs to go into its own module, right? Do we want displayio.framebuffer.FramebufferDisplay now that we can properly do submodules?
or just framebufferdisplay as a parallel top level module
lone axle
I did try out 5.1.0.rc0 on a few devices last night and did not run into any trouble. I have a few more I intend to try but might not get to them until tomorrow. I'm teaching programming fundamentals remotely starting tonight so it will take a good chunk out of my evening.
orchid basinBOT
timber mango
@slender iron Do we want displayio.framebuffer.FramebufferDisplay now that we can properly do submodules?
@onyx hinge I would go with this in order to keep the associations and connections clear. One can always dopython from displayio.framebuffer import FramebufferDisplay
marble hornet
@tulip sleet i was thinking about the safemode issue (I;m having it rn w/ the watch when it runs out of battery (prototype)). i think a good solution might be running a different script when it boots on safemode instead of the normal one. that way the coder could choose to just run the normal script or run some sort of error reporting script or a check to see if anything is broken script.
thoughts?
onyx hinge
@timber mango that was more or less my thinking, I'm happy we're on the same wavelength
slender iron
or just framebufferdisplay as a parallel top level module
@onyx hinge This looks good to me. we can't enforce correct submodule import so best to avoid it
ie, you can use microcontroller.pin by just import microcontroller
onyx hinge
true
slender iron
we just fixed it so import microcontroller.pin works too
onyx hinge
maybe someday we'll fix the other part π
idle owl
Note to self: if git fetch is taking ages, git clone will not be any faster. π
raven canopy
git clone --show=True? π
idle owl
Ah crap I did forget to call show!
slender iron
I'll be out for an hour or two. it's grocery shopping day
idle owl
@slender iron I sent you an email regarding RTD, so you're aware. No rush on it, my priorities are elsewhere at the moment.
raven canopy
May the odds be in your favor!
onyx hinge
we just did our grocery shopping too. our local grocery has stuff pretty much in stock so that's nice
obsidian compass
I have a question, is it really necessary to have a neopixel LED on custom designed CP boards ?
idle owl
@obsidian compass It's not necessary, no. It is handy to have a status LED though.
Some of our boards use DotStars if that's any help.
But it's not required to have a status LED to be considered a CircuitPython compatible board.
obsidian compass
Yes i have thought about that but can't find that via any local distributor here in India
idle owl
Understandable.
obsidian compass
Thanks @idle owl π
indigo wedge
yes it's a "dumb" RGB LED
obsidian compass
So I am guessing we need to change the type of LEDs while compiling the bootloader for the custom board and we will be good to go?
onyx hinge
though there are some problems with the status LED on the particle boards, https://github.com/adafruit/circuitpython/issues/2707
tulip sleet
@marble hornet Thanks! A safemode.py is an interesting and simple solution. I'm going to get an issue discussion going more after I finish some bootloader stuff.
timber mango
I set my two Feather M0 RFM69 boards up with Circuitpython 5.0.0 and a few libraries I need. There is no 'time' library though, which I need for timing functions.
onyx hinge
@timber mango should be built in https://circuitpython.readthedocs.io/en/5.0.x/shared-bindings/time/__init__.html
solar whale
I posted in #help-with-radio that time is present on mine
onyx hinge
thanks @solar whale for confirming and moving to the right place π
bitter bronze
@tulip sleet I tried it on a different machine and I was able to get it working. Thanks.
tulip sleet
@bitter bronze great!
bitter bronze
@tulip sleet are there any plans to work with Kaspersky or look into a work around for this problem. I have sent a message to Kaspersky letting them know about this problem and suggesting how they could support your work. I requested that they look into the issue and possibly provide a work around or code in their software.
tulip sleet
I filed a ticket with them or something like that a long time ago. Some other utility software mfrs were more responsive, e.g. AIDA64.
i would make sure it's Kaspersky by disabling or temporarily uninstalling. It could also be something else like the backup program.
the previous issues we had with Kaspersky were a hang, not just the drive being invisible
bitter bronze
Thanks. I will try turning off Kaspersky. The backup program is only started when I want to do a full dump once a week. I restarted the system after the last backup and it is not running now, but I still can't see the CPX drive. I have not had any hangs.
tulip sleet
do you have AIDA64? you might have installed it a long time ago and forgotten about it. Also Hard Drive sentinel and other HDD monitoring programs
bitter bronze
I do not have it, but looked it up after you mentioned it. I do not have the Hard Drive Sentinel or other explicit monitoring programs.
ionic elk
@slender iron
Traceback (most recent call last):
File "../../tools/build_memory_info.py", line 71, in <module>
ram_region = regions["RAM"]
KeyError: 'RAM'
any ideas?
I don't have a RAM region on the H743 - how did you get around this on the imx, which also lacks said region?
Also, I'm embarrassed to admit that the closer I look at the i.mx linker file the more confused I get. I've been hacking at it all morning and feel even more lost than I was when I started...
Why are DTCM and ITCM 32kb, despite the datasheet saying they're 128k? How can you fit the instruction code in the ITCM if it's either of those sizes? Is the DTCM being used for more than the stack? Why would the stack even have a data/bss section if not? How is it determined what gets put in the DTCM data/bss vs the normal region?
And am I right that the i.mx autogenerated most of that reset-handler code? STM has a startup file written in assembly and a system file in c, but neither of them even reference dtcm, even though the default STM32H7 linker appears to use dtcm for both the stack and heap.
lapis hemlock
@onyx hinge Jeff, I have implemented the ** operator, and binary operands can now be swapped, i.e., a = array([1, 2, 3]); a + 1; 1 + a are valid. Could you, please, look at the changes: https://github.com/v923z/micropython-ulab/tree/power .
@onyx hinge Relational operators cannot be turned around, so 1 < a won't work. These are not implemented in micropython, so if you want to support those, we would have to request upsteam modifications. I am not against that, but perhaps there is a strong reason for the missing implementation, which I don't see at the moment.
And I haven't the foggiest idea as to why the test fails with the slices... I haven't changed anything there.
slender iron
@ionic elk back from grocery shopping and lunch. want to voice chat?
onyx hinge
@lapis hemlock I would like to take a look but it will most likely be tomorrow. thanks as always for the updates
lapis hemlock
@onyx hinge No problem. I think I could locate the error, I am working on a fix.
slender iron
lapis hemlock
@lapis hemlock I would like to take a look but it will most likely be tomorrow. thanks as always for the updates
@onyx hinge After the fix, tests pass. Are you very busy at the moment?
onyx hinge
Glad you found the fix
lapis hemlock
I had a question about github actions, but it is not urgent at all, we can defer it.
onyx hinge
okay, I am trying to finish up a part of this protomatter project and then call it quits for the day π
lapis hemlock
OK, then some other time.
onyx hinge
false, // pixels_i|n_byte_share_row (unused for depths > 8)
this typo crept in a LOT of copy paste places
orchid basinBOT
idle owl
@onyx hinge Am I reading right, f-strings are enabled for everything but SAMD21? or am I conflating things
onyx hinge
@idle owl right. there may be some other very full board it was also disabled for, but it's sure off for the M0s
idle owl
Ok good enough
onyx hinge
and it's supposed to be on for all the newfangled boards like stm32 and such
idle owl
right on. Thanks
ionic elk
@slender iron Hi sorry I missed you today, got pulled into a project and then cooking. Would you have some time tomorrow morning?
slender iron
@ionic elk I should but not sure of the timing
ionic elk
Ok I'll check back with you then
slender iron
kk, I have to check with Becca about it
orchid basinBOT
orchid basinBOT
onyx hinge
wowow I missed the fomu port landing. π₯
slender iron
yup! merged it earlier
welcome @half geyser
and then updated my lower_power branch and needed to update the litex port for my changes
half geyser
Yay! Thanks. Super happy to have circuitpython for Fomu built upstream.
slender iron
π
my test code.py:
import digitalio
import neopixel_write
import time
import microcontroller.pin
# This pin should be in board. The microcontroller.pin should be the FPGA's name
# for the IO.
d = digitalio.DigitalInOut(microcontroller.pin.TOUCH1)
while True:
neopixel_write.neopixel_write(d, b"\x00\x88\x00")
time.sleep(1)
neopixel_write.neopixel_write(d, b"\x88\x00\x00")
time.sleep(1)
@half geyser I added a short README.rst with instructions too
onyx hinge
@slender iron is the general term for this kind of LCD panel interface we were discussing "LVDS"?
slender iron
no, I don't think so
LVDS is a differential signal I believe
I think the ones we were looking at are TTL
onyx hinge
ok it must be different than what I was reading about raspberry pi displays that use a ribbon connector then
slender iron
ya, probably. I haven't looked too much into it
onyx hinge
looks lke the evk has a 565 RGB interface (there's spots for 888 on the 40-pin connector, but the low pins are just GNDed) and some clock/sync signals
slender iron
π
marble hornet
@slender iron i'm having the following error building any version of the low power module. looking into fixing it now
Use make V=1, make V=2 or set BUILD_VERBOSE similarly in your environment to increase build verbosity.
make: *** No rule to make target `extmod/ulab/code/create.c', needed by `build-TG-Watch02A/genhdr/qstr.i.last'. Stop.```@marble hornet update the submodule(s)? looks like it doesn't have the files in ulab.
marble hornet
@raven canopy i downloaded instead of cloned. do you know how to clone a specific branch from someone's fork? specifically https://github.com/tannewt/circuitpython/tree/lower_power
i tried --single-branch
and googling it
raven canopy
git clone -b <branch name> <url> should work.
marble hornet
thnx
marble hornet
@slender iron expect an estimation of battery power draw difference by the end of tomorrow
spice crypt
Since lower_power is the last discussion, I'm also having issue compiling it.
(cloned https://github.com/tannewt/circuitpython.git)
git checkout lower_power git submodule sync; git submodule --init; git submodule update --recursive make V=1 SD=s140 SERIAL=/dev/tty.usbmodem1411 BOARD=feather_nrf52840_express all make: *** No rule to make target 'extmod/ulab/code/create.c', needed by 'build-feather_nrf52840_express/genhdr/qstr.i.last'. Stop.
raven canopy
@spice crypt you need to make sure the submodules have been populated.
git submodule sync
git submodule update --init
yes, I have.
raven canopy
oh...just saw that. my bad. π
raven canopy
oh. try with out the submodule update --recursive. we've stopped using that i think.
spice crypt
Okay, do I have to delete anything beforehand?
raven canopy
i would use a clean again. (the ulab submodule is a fork, which i think recursive causes confusion)
spice crypt
Success!
raven canopy
yay! π
spice crypt
Looks like submodules are cloned all over, so probably no easy way to "refresh"
raven canopy
the two submodule commands should track with the current working tree.
slender iron
@spice crypt thanks for the sponsorship!
slender iron
π
spice crypt
Having the same compatibility issue (like several days ago) when I flash the lower_power build over my present 5.0.0 libraries. If this has something to do with ble --> _ble rename, how do I mitigate this?
main.py output: Traceback (most recent call last): File "main.py", line 87, in <module> File "adafruit_ble/__init__.py", line 275, in name AttributeError: 'Adapter' object has no attribute 'name'
bitter bronze
Second problem that might be related to Circuit Python or NeoTrellis hardware or maybe something else. I have adapted John Park's code to connect NeoTrellis keys through AutoHotKey to supply character streams to Emacs, Windows Command Prompt and other programs. It works for varying periods of time and then the NeoTrellis stops working until I unplug the USB cable and then plug it in again? Is this an unusual occurrence for NeoTrellis or CircuitPython. Are there any timeout issues for the hardware or CircuitPython? Thanks.
lone axle
@bitter bronze I'm not certain that I understand 100% but I do think it may be unusually. I have several Circuit Python devices around the house that are plugged into various computers sending HID key and mouse events to the host computer. They run for days or weeks at a time without needing any sort of unplug / re-plug.
if you can go ahead and post a link here to your code on gist or somewhere online if possible.
lone axle
@idle owl I can work on making the change for that Bitmap Font PR later today if it can be helpful to take that one off of your plate.
idle owl
@lone axle That would be brilliant. Thank you.
slender iron
@spice crypt I'll look into that today
ionic elk
@slender iron Latest push to PR has commented out itcm and dtcm code that failed, if you'd like to take a look
ionic elk
I'm really stuck on it because STMCube's linker file also plainly makes use of the itcm and dtcm sections, but it has no reference to them anywhere else - not in the startup file, not the system file, nowhere. So I have nothing to base it off of like i.mx
orchid basinBOT
slender iron
@ionic elk I'm free for the next 3.5 hours and can chat whenever you can
slender iron
k, I'm free too
bitter bronze
@foamguy Thanks. Just knowing it is not a known problem helps me know where to look. It is not a critical issue and I will unplug it and plug it back in to get it to work again. I will do some testing on a second machine to see if it might be related to other software running on the first machine. I have not had any experience with posting code, so I will have to research that. I will also clean up the code before posting. By the way what code editor do you use? I have tried Mu but it is a bit crufty, I have also have used the cumbersome Python IDLE. And a few others. Thanks again.
lone axle
@bitter bronze I'm pretty flexible with IDE / Editors. If am working directly on a CIRCUITPY device I tend to use Mu so that the serial pane is together with the editor. If I am editing code thats local on my PC and not on a CIRCUITPY device yet I'll use either Notepad++ or Atom (I'm trying to decide which I like better atm) especially if it's just a quick look or small edit within a file. If it's a full project that is in a git repo and I intend to push changes and make a PR from it then I use PyCharm typically because it has lots of fancy utilities built-in that make git easier to understand and use for me.
slender iron
@spice crypt what is in your main? I just printed the adapter name through a BLERadio object and it worked just fine
spice crypt
@slender iron
`from adafruit_ble import BLERadio
from adafruit_ble.advertising.standard import ProvideServicesAdvertisement
from adafruit_ble.services.nordic import UARTService
import time
import board
import digitalio
import analogio
import displayio
import terminalio
from adafruit_st7789 import ST7789
from busio import I2C
import adafruit_ds3231
from adafruit_bitmap_font import bitmap_font
from pulseio import PWMOut
import neopixel
import storage
import json
#from supervisor import reload, runtime
from adafruit_display_text import label
from adafruit_display_shapes.rect import Rect
#from adafruit_display_shapes.circle import Circle
from adafruit_display_shapes.roundrect import RoundRect
#from adafruit_display_shapes.triangle import Triangle
from adafruit_display_shapes.line import Line
Bluetooth Init
ble = BLERadio()
ble.name = "nrf52840ex"
uart = UARTService()
advertisement = ProvideServicesAdvertisement(uart)`
spice crypt
The problem really extends beyond that. Last time I almost down this rabbit hole (for me), even when I comment out that, othe errors come up
Let me go get my device, brb
slender iron
kk, which device is it? I can grab the same
spice crypt
It's based on the NRF52840_express
slender iron
I'm testing on a circuitplayground bluefruit
spice crypt
My device also happened to get corrupted! Maybe a sign of other things. copying over files now after formatting
spice crypt
Ok, back up and running
slender iron
and still having troubles?
spice crypt
Yes, even with barebone code
care to share your test code?
`from adafruit_ble import BLERadio
from adafruit_ble.advertising.standard import ProvideServicesAdvertisement
from adafruit_ble.services.nordic import UARTService
import time
import board
Bluetooth Init
ble = BLERadio()
#ble.name = "nrf52840ex"
uart = UARTService()
#advertisement = ProvideServicesAdvertisement(uart)`
from adafruit_ble import BLERadio
from adafruit_ble.advertising.standard import ProvideServicesAdvertisement
from adafruit_ble.services.nordic import UARTService
import time
import board
Bluetooth Init
ble = BLERadio()
#ble.name = "nrf52840ex"
uart = UARTService()
#advertisement = ProvideServicesAdvertisement(uart)
`from adafruit_ble import BLERadio
from adafruit_ble.advertising.standard import ProvideServicesAdvertisement
from adafruit_ble.services.nordic import UARTService
import time
import board
Bluetooth Init
ble = BLERadio()
#ble.name = "nrf52840ex"
uart = UARTService()
#advertisement = ProvideServicesAdvertisement(uart)`
main.py output:
Traceback (most recent call last):
File "main.py", line 11, in <module>
File "adafruit_ble/services/nordic.py", line 57, in init
File "adafruit_ble/services/init.py", line 50, in init
TypeError: cannot create 'Service' instances
spice crypt
FYI. My adafruit_ble library is from adafruit-circuitpython-bundle-5.x-mpy-20200228
slender iron
what version? there should be a __version__ on the module
I believe I'm using something later. Its pulled from my copy of the bundle repo
spice crypt
I didn't find version but I did find this in the bundle root in VERSIONS.txt
https://github.com/adafruit/Adafruit_CircuitPython_BLE/releases/tag/5.2.0
slender iron
ah, that works too
for version you'd do python import adafruit_ble print(adafruit_ble.__version__)
and you are using my lower_power branch? do you have your own copy with your own board define?
spice crypt
5.2.0
Yes, your branch. Default nrf52840_express board config
I'm using the feather
slender iron
is the REPL version: Adafruit CircuitPython 5.1.0-rc.0-104-g77562a2ac on 2020-04-01 with a different date?
spice crypt
Adafruit CircuitPython 5.0.0-alpha.1-8-g9fb635b60 on 2020-03-31; Adafruit Feather nRF52840 Express with nRF52840
from yesterday
Built with: make V=1 SD=s140 SERIAL=/dev/tty.usbmodem1411 BOARD=feather_nrf52840_express all
slender iron
you are on my master branch
which is not my lower_power branch
my master is 9fb635b60dd681331a6e8a7c1f38d30cd1cd5f83 from September 2019
spice crypt
hmm!
tstr@tstr-VirtualBox:~/CP_lowpower/circuitpython/ports/nrf$ git branch
- lower_power
master
Must have copied the wrong UF2
slender iron
lower_power is your branch name
what is the first line of git log?
mine is commit 77562a2acde3c82e5949096533535b211f747c6d (HEAD -> lower_power, tannewt/lower_power)
spice crypt
haha, was already getting that
`commit 77562a2acde3c82e5949096533535b211f747c6d (HEAD -> lower_power, origin/lower_power)
Author: Scott Shawcroft scott@tannewt.org
Date: Tue Mar 31 18:27:55 2020 -0700
Update the supported ports
`
slender iron
that looks right
spice crypt
cp ./build-feather_nrf52840_express/firmware.uf2 ~/shared_tmp/
that's the UF2 I copied
slender iron
my make command is: make -j 17 BOARD=circuitplayground_bluefruit
spice crypt
I think I copied the wrong UF2, there's and -s140 appended to another folder
let me update again
slender iron
my folder doesn't have the -s140
spice crypt
Hmm. anyway, things look better so far
slender iron
great!
slender iron
π no problem
spice crypt
Is it expected for the flash contents to be corrupted after UF2 updates?
It's happened the last two times
Didn't before IIRC
slender iron
we did change something in 5.0.0 betas that could have caused it. It generally doesn't.
spice crypt
Ok, will report if happens consisently, obviously. Trying my full code shortly
so as far as expecting the nrf52 to go into sleep mode. I should expect it during a time.sleep()? How to utilize a GPIO toggle to do the same?
slender iron
it will only do it during time.sleep with this PR.
external interrupt support will come later
(I have to decide on an API)
spice crypt
I have a nice DMM on loan, I'll note what I see.
slender iron
cool cool. sleep is only idling the CPU so its not a huge improvement. the main work was getting time keeping moved onto the RTC
spice crypt
everything counts in large amounts
slender iron
ya, definitely an improvement
spice crypt
slender iron
huh, weird
I see that too
time.monotonic should work
time.time has some mechanics around the RTC source
I need to look into that tomorrow
spice crypt
That'll work for me. The extra resolution is nice sometimes. I have a actual offboard RTC for my project
slender iron
it should be 30.15 us resolution
monotonic_ns will always be precise but causes an allocation if the number is > 2**30
spice crypt
This is a very broad question, but should displayio elements be expected to take a long time to initialize? I have 3 screens with 5-10 elements and they all take ~15seconds to initialize
spice crypt
I think I have tried in the past. I don't .show() the global group until all the inits have finished
Same effect?
slender iron
what do you mean by inits then? it should be fast to fill a group that isn't shown
slender iron
if it's not on a screen then I'd expect it to be fast
spice crypt
Good to know
slender iron
though python is only so fast π
if you are printing anything and the display are inited then you will be slowed by the prints
tulip sleet
@slender iron I'm confused about the write-synchronization for BOD33. It has a B33SRDY bit (other write-sync'd regs have a SYNCBUSY bit). The datasheet says:
The BOD33 register is Enable-Protected, meaning that they can only be written when the BOD is disabled (BOD33.ENABLE=0 and STATUS.B33SRDY=0). As long as he Enable bit is '1', any writes to Enable-Protected registers will be discarded, and an APB error will be generated. The Enable bits are not Enable-Protected.
slender iron
the bit docs for it says it isn't enable protected
tulip sleet
I thought I needed to spin on B33SRDY until it's 1
which thing isn't enable protected?
slender iron
ENABLE isn't enable protected
tulip sleet
right, so I sent ENABLE to 0, and then I need to wait for B33SRDY to go to 1 or to 0?
slender iron
the doc says 1 is sync complete
tulip sleet
and when I set ENABLE to 1, do I need to wait similarly?
tulip sleet
but the quote above says you need STATUS.B33SRDY=0, I would have thought =1
slender iron
ya...
tulip sleet
i've been looking for some examples. I found maybe one on the whole www
slender iron
where is the quote from?
slender iron
tulip sleet
right, so maybe the first quote is wrong
tulip sleet
also note that paragrph only talks about writing 1 to enable
i guess it's simple enough to try; it will hang if I get it wrong π
there is no sample code or driver that actually does this, it appears
raven canopy
Bit 2 β B33SRDY BOD33 Synchronization Ready Interrupt Enable
----------------------------------
Writing a '0' to this bit has no effect.
Writing a '1' to this bit will clear the BOD33 Synchronization Ready Interrupt Enable bit, which disables the BOD33 Synchronization Ready interrupt.
that's about interrupt enable; i'm not doing interrupts
this is BOD33.B33SRDY
there are interrupts galore for all these bits
slender iron
yay interrupts π
tulip sleet
i'll just try it; this peripheral is unlike all the others
raven canopy
ahhh. sorry about the noise Dan. love that they all use the same register name. π
tulip sleet
np, I searched and found the same thing and got confused at least twice. it just comes before the real description
lone axle
Is there something I could do to fix an actions error with build docs like this?
raven canopy
@lone axle add the module to the automock in docs/conf.py
lone axle
Thank you
raven canopy
i always forget the actual variable name. :D
autodoc_mock_imports.
https://github.com/adafruit/Adafruit_CircuitPython_Bitmap_Font/blob/master/docs/conf.py#L24
tulip sleet
it's kind of funny that when I google the main code examples that come up are the code I already wrote π
raven canopy
so you know the code is correct! search results you can trust. haha
slender iron
ya, that's happening to me more and more
lone axle
Is the CircuitPlayground library baked in to the firmware for that CPX?
tidal kiln
@lone axle yep. here's the full list:
https://github.com/adafruit/circuitpython/blob/master/ports/atmel-samd/boards/circuitplayground_express/mpconfigboard.mk#L22
lone axle
Neat, thank you.
onyx hinge
@slender iron thanks for trying to raise the issue about the CI troubles we're seeing. https://github.community/t5/GitHub-Actions/Install-dependencies-in-Ubuntu-flakes-several-times-a-wekk/td-p/51785
we could try apt... || apt... || apt... and maybe it would succeed on try 2 or 3 .. we could switch out the mirror manually, but it might be rude to use canonical's bandwidth.
slender iron
I don't think it's canonical's bandwidth since the url has azure in it
I assume that actions runs on azure and is configured to a local server
marble hornet
@slender iron i got about an extra 1/2 to an hour out of the lower_power changes
slender iron
nice! the display probably takes a lot of the battery
marble hornet
yes it does. and it dipped power a bit before it died so it went into safe mode and stopped running code for near the end. wasn't a perfect test. but the safemode.py change should fix that issue.
raven canopy
I don't think it's canonical's bandwidth since the url has azure in it
I assume that actions runs on azure and is configured to a local server
i actually spent some time last night trying to dig into that. (i think it was last night. what's today?) looked through both the actions/runner and actions/virtual-environments repos, and didn't get any closer to finding anything.
i also tried to match up the listed update timestamps on the archive to the failed actions. didn't see any correlation.
slender iron
what safemode.py change? I didn't see anything about that go by in my email
lone axle
@slender iron I don't know for certain but I think I saw someone mention the idea of having safemode.py get launched when the device boots in safemode so people can choose more specifically what happens as compares to normal boot up.
raven canopy
marble hornet
thank you sommersoft
I thought a safemode.py running after a safe-mode error would be a simple, practical, and powerful solution to unavoidable brownouts.
spice crypt
@slender iron If I use custom bdf fonts w/ displayIO, my rendering (initialization) time goes up significantly. What gives? I even have auto_refresh disabled until all the groups, label, and shapes are set.
3x bdf fonts: 23seconds
terminalio.FONT only: 3 seconds
from adafruit_bitmap_font import bitmap_font font_arial = bitmap_font.load_font("/fonts/Arial.bdf") font_arial_16 = bitmap_font.load_font("/fonts/Arial-16.bdf") font_helvetica_bold_16 = bitmap_font.load_font("/fonts/Helvetica-Bold-16.bdf")
crimson ferry
@spice crypt Are you preloading the glyphs? `font_arial.load_glyphs(b'ABC123.,:')
spice crypt
No
crimson ferry
That should save time later
spice crypt
I actually just stumbled upon that. I think I tried before but will again. Thanks.
onyx hinge
@slender iron yeah I'm not sure how it works, some of the URLs are azure, some are not, but all of the errors say azure. After this operation, 6,707 kB of additional disk space will be used. Err:1 http://azure.archive.ubuntu.com/ubuntu xenial/main amd64 libunistring0 amd64 0.9.3-5.2ubuntu1 Could not connect to azure.archive.ubuntu.com:80 (52.177.174.250), connection timed out Err:2 http://security.ubuntu.com/ubuntu xenial-security/main amd64 gettext amd64 0.19.7-2ubuntu3.1 Unable to connect to azure.archive.ubuntu.com:http:
raven canopy
and to add more clarification, or confusion, the apt source is clearly a mirror. although it isn't on Cannonical's official ubuntu mirror list that i can see.
# run from an Actions workflow...
> Run ping -R -c 3 azure.archive.ubuntu.com
PING archive-lb.eastus2.cloudapp.azure.com (52.177.174.250) 56(124) bytes of data.
--- archive-lb.eastus2.cloudapp.azure.com ping statistics ---
3 packets transmitted, 0 received, 100% packet loss, time 2029ms
subsequent curl step is successful, so take the ping failure as a false positive.
and, an additional post was made on the issue, to which a "reported to the engineers" response was made.
https://github.community/t5/GitHub-Actions/sudo-apt-get-install-fails-with-Could-not-connect-to-azure/m-p/52300#M8479
onyx hinge
@tulip sleet @slender iron There are several ulab bugfixes since 5.1.0-rc.0. If I want to pull these in, that entails making a rc.1, right?
ionic elk
@tulip sleet I'm running into issues with this HAL function: https://github.com/hathach/st_driver/blob/3fc2e0f3db155b33177bb0705e0dd65cadb58412/STM32F7xx_HAL_Driver/Inc/stm32f7xx_hal_rcc_ex.h#L845
compiler is throwing a shadow variable warning for tmpreg, but isn't the UNUSED(var) => (void)var supposed to counteract that?
do you have any ideas? This seems completely contained in this one HAL file, and if I don't figure it out I'm going to have to turn off the variable shadowing warning
Nevermind, Thach had to turn it off so I assume I do too.
tulip sleet
@onyx hinge right, rc.1, but it doesn't have to soak that long if those are the only changes
simple pulsar
I was about to download rc.0 to have a play. Is it worth waiting for rc1?
tulip sleet
^^ @onyx hinge are you going to do rc.1 imminently?
onyx hinge
I"m working to complete some protomatter stuff first.
simple pulsar
No rush here, i was wondering what one to go for. Are all the ulab bugs noted in GitHub in case I run across them?
onyx hinge
I haven't even tested that updating the ulab submodule works as expected (probability is high, however)
micropython-ulab has its own github issue tracker https://github.com/v923z/micropython-ulab
These fixes would be incorporated: ```0d5dc6b Merge pull request #78 from v923z/argmax
411a5ff Merge pull request #81 from v923z/unistd
26d740e Merge pull request #80 from v923z/copy
272685d Merge pull request #79 from v923z/sort
ac3f03c Merge pull request #70 from v923z/testfix
3ec9f5a Merge pull request #69 from jepler/boolean-cpy
@onyx hinge are they fixes or adding more stuff? remember we can always release 5.1.0 now and update ulab in 5.2.0 beta or whatever
onyx hinge
@slender iron a couple are bugfixes -- sort, boolean-cpy, argmax. not all are.
I think they're minor, so we could toss it under "known problems"
slender iron
that's fine with me. I'd just like to see something new released
onyx hinge
selecting just elements from an array that match a condition, a[a>1] is broken. And something about sorting is broken, but I think it only affects sorting 2d matrices which is a weird operation
slender iron
I would like to do an unstable release as well
so anything in master would be released too
onyx hinge
how does that interact with making 5.1.0 final?
it would make the rc disappear from the website which would be inconvenient I guess
slender iron
right, I can't make a new unstable until 5.1.0 is released or 5.1.0-rc.2 incorporates everything from master
onyx hinge
hmmm
taking that at face value, it makes more sense to release 5.1.0 and add to the known problems section. I'll wrap up my current task and draft release notes...
slender iron
k thanks!
orchid basinBOT
idle owl
@lone axle So the new PR you created includes the fix from the PR I created on Bitmap_Font?
lone axle
@idle owl correct. I forked from yours changed the import and changed the mock import in docs. Then created my PR from that.
idle owl
Perfect. I'll close my PR. Thanks
lone axle
Yep you're welcome.
onyx hinge
@slender iron some of the actions failed at "install deps" for the 5.1.0 release. Can I use the "re-run" button once it becomes available or do I need to take different action?
idle owl
@onyx hinge It's been doing that lately. Rerun all jobs.
It's been doing that lately across the board.
I've had to do it a few times every so often.
onyx hinge
to libraries as well?
idle owl
Yes.
onyx hinge
well now what. The button isn't where I expect to see it.
idle owl
It's not?
onyx hinge
onyx hinge
at https://github.com/adafruit/circuitpython/runs/555487221?check_suite_focus=true or just generally?
idle owl
This is what it says it failed on New boards check 1s ##[error]Process completed with exit code 1. Run python3 -u ci_new_boards_check.py Boards missing in .github/workflows/build.yml: nucleo_f767zi ##[error]Process completed with exit code 1.
onyx hinge
onyx hinge
the first one is the 5.1.0 release
idle owl
Oh hmm
onyx hinge
@tulip sleet any ideas?
idle owl
not getting the button there either.
?
wait that's not right either but I got there via the release link on the Actions main page
onyx hinge
I can try that one. one of them is the one that would upload to circuitpython.org (errr well s3), if I understand correctly
or maybe between those two runs things are okay? I checked one board which failed in each run (aramcon_badge_2019 and xinabox cc03) and they both have UF2s, possibly from the other run? The failures were different in each run.
idle owl
Not sure about that.
onyx hinge
I wish I were more confident. I feel like I should wait and let someone else take a poke at it.
they'll see the mentions in due time
idle owl
I don't think rerunning it will break anything. But you've got me second guessing as well at this point.
slender iron
@onyx hinge you won't break anything π
I just wait to blog and update the website until I'm sure most assets are where they should be
onyx hinge
@slender iron these builds are still missing, it seems. ```electronut_labs_blip HTTP/1.1 404 Not Found
espruino_pico HTTP/1.1 404 Not Found
espruino_wifi HTTP/1.1 404 Not Found
feather_stm32f405_express HTTP/1.1 404 Not Found
makerdiary_nrf52840_mdk HTTP/1.1 404 Not Found
makerdiary_nrf52840_mdk_usb_dongle HTTP/1.1 404 Not Found
meowbit_v121 HTTP/1.1 404 Not Found
pyb_nano_v2 HTTP/1.1 404 Not Found
pyboard_v11 HTTP/1.1 404 Not Found
spresense HTTP/1.1 404 Not Found
stm32f411ce_blackpill HTTP/1.1 404 Not Found
stm32f411ve_discovery HTTP/1.1 404 Not Found
stm32f412zg_discovery HTTP/1.1 404 Not Found
stm32f4_discovery HTTP/1.1 404 Not Found
oh some of those come in .hex form though
blip is there (hex), espruino pico is there (bin), stm32f405 (bin), makerdiary (hex)
slender iron
ya, that looks like all of the stms
onyx hinge
then it's just my quick checking script that's broken and probably they're all there after all
slender iron
is the website update merged? I'll need to follow up with a download count update
onyx hinge
oh right, I missed that
onyx hinge
@slender iron ok you can update your download count thing now I think?
will do forum & blog after lunch
slender iron
ya, think so
slender iron
@onyx hinge what issue hung you up with checkout v2? v1 doesn't work for reruns which would help with the azure server issue
onyx hinge
@slender iron sometimes, when fetching tags for submodules, the command would report that there was no "origin" remote. e.g., this particular run https://github.com/adafruit/circuitpython/pull/2725/checks?check_run_id=533702449
slender iron
I wonder if that was a github issue
onyx hinge
actions@v2 seems to prioritize fetching as little as possible, which is understandable given that some repos contain a LOT of data. but it breaks the very sensible practice we follow, of auto-identifying via git-describe
onyx hinge
I'm sad that contributors like xobs are frustrated especially since we're largely powerless to fix transient github issues
slender iron
ya, I think we should just be more lenient until it's fixed
his PR is for fomu only so it's ok to just merge
it's my PRs you need to worry about π
onyx hinge
(and 1%-chance-per-job issues will hit us 2/3 of the time due to the large number of jobs)
slender iron
yup
onyx hinge
until kattni mentioned that libs were seeing the apt-get problem too I wondered if we were hitting another possibly-undocumented rate limit.
slender iron
ya, that could be it. or it could just be the cloud creaking as we all use it more than we used to
sounds like it finally got escalated though
onyx hinge
forum post made, 5.1.0-rc.0 post un-stickied.
slender iron
thanks!
slender iron
I wonder if it's time to do the great merge of 2020 after the lower_power stuff is in
onyx hinge
blorg post is scheduled
timber mango
Is that any relation to the BORG?
ionic elk
@slender iron what's the big merge? micropython?
slender iron
ya
ionic elk
what are the big features that'd get added
slender iron
I think mpy files get smaller
but don't know of that many things
mainly to just get us closer to them. jeff just backported a test fix for us
idle owl
@lone axle Do you have a few minutes to take care of some PRs for me? They're simple link updates.
onyx hinge
@ionic elk we're based on micropython 1.9.4, so anything major in their changelogs since then would be potential improvements.
old smelt
Hi @slender iron - I saw in the CP Weekly notes that you have been working on some BLE MIDI stuff. Any docs or projects public, yet? This is interesting on a couple of levels. Thanks!
slender iron
@old smelt the libraries are public and the fixes are checked into CP
slender iron
john park's project next week will use it if it works for him
slender iron
someone on the forum was having trouble getting it going
but I haven't been able to debug it with them
old smelt
Did you have a specific piece of hardware in mind when doing the devleopment?
slender iron
I tested on an nrf52840
old smelt
Ok. So a CPB would work?
slender iron
yup!
old smelt
should work
π
Nice. I brought that home in my boxes and stacks o' office stuff.
Thanks!
ionic elk
@slender iron I'm getting a hard crash as soon as I run the OCRAM copy loop in Reset_handler - it's a regular hardfault handler exception, not MPU related.
slender iron
double check that the pointer bounds are ok
lone axle
@idle owl not atm sorry. But if you send them to me I can probably do it later tonight if that is helpful.
idle owl
@lone axle I got it covered, thank you so much for the offer!
ionic elk
@slender iron Cannot access memory at address 0x24000000 interesting.
Obviously this shouldn't happen with the assembly code...
slender iron
what was the variable name?
ionic elk
that's just the flash RAM. but yikes this actually applies past reset to reverted code... It applies even after the chip is reset, and the code reverted to a previously stable build.
welp I bricked my board
so much for the MPU
onyx hinge
eek nobody likes to hear that
ionic elk
Β―_(γ)_/Β―
lapis hemlock
that's just the flash. but yikes this actually applies past reset to reverted code...
@ionic elk Is this a poem?
ionic elk
Rephrased: the memory access issue applies when I revert my commit to one that was previously tested successfully, using the stm32 defaults.
lapis hemlock
@onyx hinge Thanks for checking out the code. I have found another bug in the sorting routine. I have fixed that, in case you have to pull it.
ionic elk
It also persists past full chip erases and power reset
onyx hinge
@lapis hemlock as you may have seen we decided to go ahead with 5.1.0 without those recent bugfixes. I am not fully familiar with how we do releases, but I think this still leaves us open to incorporating bugfixes from ulab in 5.1.1.
ionic elk
IE, something on the chip has locked this memory, possibly permanently, and normal code cannot unlock it. I'll give ST's official utility a shot, sometimes it manages to "super erase" better than Jlink and texane/stlink
lapis hemlock
@lapis hemlock as you may have seen we decided to go ahead with 5.1.0 without those recent bugfixes.
@onyx hinge No, I haven't seen that. I have seen the release candidate only.
So, should I expect higher issue activity in the coming days?
onyx hinge
Hopefully people will start using it more. I'm not sure whether our users will start by filing issues against circuitpython or ulab. I will tag you in anything that seems relevant.
lapis hemlock
I will tag you in anything that seems relevant.
@onyx hinge OK, thanks!
A different question: how can you develop/test github actions?
I mean, private users have an allowance of 2000 minutes per month.
onyx hinge
by pushing over and over again until it's not broken π
lapis hemlock
So there is really no test platform for that.
onyx hinge
I am not aware of one.
the 2000 minutes per month limit I think applies to private repositories on the free plan. I didn't think a minutes-per-month restriction applied to public repos, no matter the plan. I could be wrong, of course.
GitHub Actions usage is free for public repositories. For private repositories, each GitHub account receives a certain amount of free minutes and storage, depending on the product used with the account. For more information, see "About billing for GitHub Actions." [https://help.github.com/en/actions/getting-started-with-github-actions/about-github-actions]
lapis hemlock
I didn't even know that you can have free private repositories. People used to scramble their code, so that they can push it to github without actually making the content public.
onyx hinge
search turns up some possibilities for at least partially replicating github actions locally: https://github.com/nektos/act https://github.com/phishy/wflow https://stackoverflow.com/questions/59241249/how-to-run-github-actions-workflows-locally
hum no github pages are loading for me this second
idle owl
@lapis hemlock The free private repos is a recent thing.
@onyx hinge GitHub is experiencing issues.
onyx hinge
ah it must be thursday
write as much as possible in CI-agnostic way (BASH scripts, PowerShell scripts, Gradle scripts, NPM scripts, Dockerfiles, Ansible scripts - anything you know)
idle owl
Well relatively recent anyway.
lapis hemlock
One could, of course, have one's own worker, working somewhere else.
onyx hinge
this is good advice
lapis hemlock
Well relatively recent anyway.
@idle owl Thanks for the info!
idle owl
You're welcome!
lapis hemlock
I am not sure I want to hide anything, though... I am a totally open-source chap.
idle owl
Also fair π
onyx hinge
Occasionally I try to convince my spouse that we should use a private github to track household issues ... she doesn't believe me.
kitchen needs swept? you should have filed an issue. I check those first thing in the morning everyday
idle owl
I can see the skepticism in her face from here.
lapis hemlock
@onyx hinge "write as much as possible in CI-agnostic way (BASH scripts, PowerShell scripts, Gradle scripts, NPM scripts, Dockerfiles, Ansible scripts - anything you know)" while I agree with this, it is actually not the scripting that is hard, but the triggers. What triggers what and when and under what conditions.
onyx hinge
true
lapis hemlock
@onyx hinge "kitchen needs swept? you should have filed an issue. I check those first thing in the morning everyday" π€£
slender iron
but would you get up and sweep then?
lapis hemlock
but would you get up and sweep then?
@slender iron No, he would look for an assignee.
lapis hemlock
@onyx hinge No offence meant, Jeff!
onyx hinge
ttyl
slender iron
thanks @lapis hemlock have a good night!
lapis hemlock
π
raven canopy
@lapis hemlock its not exactly a perfect solution, but when Actions was still in beta I actually created a repo purely for testing various things, so i can keep the noise outside of projects. https://github.com/sommersoft/actions_testing
having actions/checkout (even with its flaws) makes it easy to pull in outside projects to setup a similar environment.
idle owl
GitHub's back up apparently.
onyx hinge
π
obsidian compass
Can someone confirm if this can be used to program/ flash CP bootloader on ATSAM and NRF chips ?
ivory yew
@obsidian compass probably, but I can't find any documentation on it
obsidian compass
@ivory yew right, but it's quite cheap so I think I should take my chances π
If doesn't work I'll buy the expensive sega j-link π€
obsidian compass
Oops I meant segger, but my keyboard like sega
ivory yew
I mean, I'm a huge sega fan, so if Sega made a j-link that's the one I'd have.
obsidian compass
Yup, I loved your talk from supercon on Sega Inspired synths π
lone axle
@turbid radish Possible topic for "news from around the web" in the newsletter: Circuit Brains Deluxe is available for backing / pre-order on Crowd Supply. https://www.crowdsupply.com/null-byte-labs/circuitbrains-deluxe (let me know if there is a better way to send you stuff for the newsletter)
ivory yew
I backed this so fast my keyboard broke
tiny oriole
im super excited about the CircuitBrains.
timber mango
@ivory yew Are you getting 1, 2, or 3 of them?
ivory yew
2
timber mango
Tomorrow is my payday... π I think I will go for 2. I really want to see this succeed!! π π
ivory yew
me too
tiny oriole
i went for just 1 for starters, but might bump it up to 2
timber mango
Even numbers ROCK! Just DO it! π π
tiny oriole
alright. second one ordered xD
timber mango
There ya go! You are now a member of the Two-Fer Circuitbrains Deluxe Club! π I will be joining soon!
I bet if each of us Circuitpython people backed this by getting two that we could get this funded in no time!
ivory yew
looks like his MOQ goal is 150.
I think we can do that!
I have the non-deluxe circuitbrains and it's great (& SO tiny)
lapis hemlock
@lapis hemlock its not exactly a perfect solution, but when Actions was still in beta I actually created a repo purely for testing various things, so i can keep the noise outside of projects. https://github.com/sommersoft/actions_testing
@raven canopy Thanks! I think you are right, this is the way to go.
timber mango
@ivory yew What is "MOQ?"
lapis hemlock
I guess, minimum order quantity.
simple pulsar
I found https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/10-useful-tips-to-using-the-floating-point-unit-on-the-arm-cortex--m4-processor last night. I thought tip 8 was interesting, "Turn off the FPU if not used". I'd imagine that's rather tricky for CircuitPython to do?
There's some intriguing hacky techniques in the comments by Jens Bauer about borrowing the FP registers for other things.
solar whale
Gaa -- "alternate use" of the FP registers was something done long ago on a SPARC system I worked with -- it was clever, but it sure caused some confusion ...
ionic elk
@slender iron why is the stack top forced to equal the heap top in stack.c even after i.MX? Does that need reworking?
slender iron
@ionic elk can you link me to the context you are referring to?
ionic elk
@slender iron https://github.com/adafruit/circuitpython/blob/e6f11947cb053fc4a9da844a3689c6af5881590d/supervisor/shared/stack.c#L49 forces the port_stack_get_limit function in port.c to return the _ld_heap_start, even though in both the H7 and i.mx case they're in completely different memory regions and have no relation to each other.
slender iron
@ionic elk I don't think it forces it. it just exempts the stack from the supervisor managed allocations
it's ok for it to return there
ionic elk
Oh, so the following code isn't really necessary when the stack and heap aren't in the same RAM region?
ionic elk
@slender iron why do you set it to be the heap start in i.mx then?
uint32_t *port_stack_get_limit(void) {
return &_ld_heap_start;
}
k
slender iron
that function isn't actually called by anything in the mimx port
ionic elk
@slender iron just got back from lunch
Oh I see what you mean I mixed them up but that doesn't actually affect anything
ionic elk
@slender iron or @tulip sleet, Micropython makes reference to some object files as SRC_O, but they don't actually exist? Do you know what's up with that? They only exist as uncompiled S or C versions
tulip sleet
the makefile will build them automatically based on default rules
but I don't know why they're called SRC_O
tulip sleet
if there's a rule like .c.o: then that will implicitly be used when referencing the .o. But what are you trying to build? What's your make command? I'll try too
or .s.o:
@ionic elk I made PYBV11 from the tip of micropython master successfully
ionic elk
sorry I had to step out for a sec to help my brother with something
@tulip sleet I'm trying to get my makefile to reference this default assembly file in the st drivers:
st_driver/CMSIS/Device/ST/STM32$(MCU_SERIES)xx/Source/Templates/gcc/startup_$(MCU_VARIANT).s so I don't have to have a bit long list of startup files in my boards directory
Right now I have SRC_S = supervisor/cpu.s boards/startup_$(MCU_VARIANT_LOWER).s which works file
but if I swap that second part to the long st_driver path above, it doesn't work
says it "can't find the object file". And it won't find it if I add the SRC_O and build rules for it either
error: build-nucleo_h743zi_2/st_driver/CMSIS/Device/ST/STM32H7xx/Source/Templates/gcc/startup_STM32H743xx.o: No such file or directory
tulip sleet
@ionic elk use find to see if it built startup_STM32H743xx.o somewhere else besides the path you gave
timber mango
Hi there, I have a question. I am trying to load a small project on my Feather Express M0 board with circuitpython. This involves a oled dislay (shield ssd1306) and a temp pressure sensor BMP280. I recently upgraded to CircuitPython5.0 and now the project no longer works. In python REPL i have loaded the script step by step and then encountered a memory issue. Which appears to be my issue. I really only have the bare essential libs on the device, and the program is really smal. It feels like some of the board memory is blocked or leaked away. Is there a way where I could completely prime the device, formatting it so to say? And if this is possible. What would be the procedure? I am curious if this is a known issue.
tulip sleet
@timber mango moved to #help-with-circuitpython
ionic elk
@tulip sleet no dice. What's funny is I also can't find a cpu.o for supervisor/cpu.s. But if that wasn't being generated, wouldn't it mess up the whole build? That's used to fetch registers for the stack allocation.