Looking at the examples, I would just put them inside the class and put an underscore on the front.
#circuitpython-dev
tulip sleet
tidal kiln
and the answer to that question decides location? that's kind of how i see it also
but not sure if that's residual c brain getting in the way of the pythonic way
tulip sleet
pick the narrowest namespace scope needed, which is inside the class
tidal kiln
and lo, PEP8 sayeth unto the world "Constants are usually defined on a module level and written in all capital letters with underscores separating words."
tulip sleet
i guess the lookup will be faster because you won't have to say self.CONST
inside the class
tidal kiln
which is probably more important than the namespace protection? thus the pep8 wording?
tulip sleet
tidal kiln
ha! has pretty much every possible question been asked on s.o.???
idle owl
Yeah I'd say so
tulip sleet
trying an experiment... hold on..
tidal kiln
but has no discussion or mention about your self. point
tulip sleet
_MODULE_CONST = const(1)
class C:
_CLASS_CONST = const(1)
print(_MODULE_CONST, self._CLASS_CONST)
$ strings ~/foo.mpy
<module>
/home/halbert/foo.py
/home/halbert/foo.py
__name__
__module__
__qualname__
print
self
_CLASS_CONST
so the _MODULE_CONST disappears, but the _CLASS_CONST does not. So for CPy/MPy the module consts would be faster, but I would say it's better programming practice to put them in the class.
so it's a dilemma
we are worried about space, so module consts are probably better, if they are really const()
tidal kiln
whats the better prog pactice argument?
tulip sleet
put them in the namespace that uses them
tidal kiln
yep
tulip sleet
but our code is small, so it's all readable anyway, one way or the other
tidal kiln
interstingly, those two examples seem to do the opposite
the one that could be class level is module level
^^ ds18x20.py
while onewire.py hides some globally useful onewire defines in the class
tulip sleet
i wouldn't read a lot into the code there - it doesn't necessarily reflect best practice either way
wait, the self._CLASS_CONST isn't necessary, I think.
tidal kiln
yep. but i'm refactoring it. so kind of have to deal with it. one way or another.
mpy-cross must take care of that then?
tulip sleet
trying something else... hold on...
kinda a violation of what you'd expect to work:
from micropython import const
_MODULE_CONST = const(1)
class C:
_CLASS_CONST = const(1)
def __init__(self):
pass
def pr(self):
print(_MODULE_CONST, _CLASS_CONST)
that works in MPy/CPy, but this doesn't work in regular python:
_MODULE_CONST = 1
class C:
_CLASS_CONST = 2
def __init__(self):
pass
def pr(self):
print(_MODULE_CONST, _CLASS_CONST)
python3
Python 3.5.2 (default, Nov 23 2017, 16:37:01)
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import foo1
>>> c = foo1.C()
>>> c.pr()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/halbert/foo1.py", line 9, in pr
print(_MODULE_CONST, _CLASS_CONST)
NameError: name '_CLASS_CONST' is not defined
>>>
so the substitution happens at compile time, kinda bizarre
tidal kiln
but understandable
tulip sleet
so I'd do module-level, so that the code could transfer to regular python, without an extensive rewrite aside from removing the const(). Then we gain the benefit of the .mpy's and are still closer to compatible.
but keep thinking about it as you rewrite
tidal kiln
i think you've answered it
also, 99.9% of whats been written is module level
thanks for the input!
tulip sleet
yw! (watching a live guitar lesson at same time)
tidal kiln
planning to try out for Toolchain Pain?
tulip sleet
only acoustic - no heavy metal, more like Python, Paul, and Mary
(took a long time to come up with a mediocre jjoke)
manic glacierBOT
idle owl
@tulip sleet You're in for Toolchain Pain Unplugged.
@tulip sleet Do I remember right that we have erasers for nonexpress boards?
tulip sleet
yes: https://github.com/adafruit/Adafruit_Learning_System_Guides/tree/master/uf2_flash_erasers; also mentioned in the Troubleshooting guide (as revised in master)
wrong, in 2.x Troublshooting: http://circuitpython.readthedocs.io/en/2.x/docs/troubleshooting.html
tulip sleet
I'll merge that into master when I do the next update merge from 2.x
manic glacierBOT
manic glacierBOT
azure wigeon
Hi. I am having some trouble installing CircuitPython on my Circuit Playground Express. I am following the UF2 install instructions but when I copy the UF2 file onto the CPLAYBOOT disk it doesn't work as expected. The flash seems to work, but the new disk which appears has no name (should be CIRCUITPY) and I can't copy any files to it because it claims to be full. I can connect to the repl using screen, though.
1/ Is this the right place to ask for help with that?
2/ What might be the problem here?
azure wigeon
Trying to write a file to the storage from python gives OsError 30, (ie read-only FS). I'm worried that I have damaged the flash storage π¦
solar whale
@azure wigeon you probably have corrupted the file system. Welcome to the club π you can erase it and start fresh. The instructions are here. https://circuitpython.readthedocs.io/en/latest/docs/troubleshooting.html#file-system-issues
azure wigeon
Thanks I'll try that
stuck elbow
by the way, to write files from python, you need to switch in yout boot.py to read-only for USB
it's to prevent corrupting the files when both sides write something
solar whale
@azure wigeon great! If you want to write from CircuitPython, there is a guide. Just need a few minutes to find it.
@azure wigeon found it https://learn.adafruit.com/cpu-temperature-logging-with-circuit-python/writing-to-the-filesystem?view=all
tulip sleet
<@&356864093652516868> A user in the forums asked about using MQTT with CircuitPython: https://forums.adafruit.com/viewtopic.php?f=60&t=127356. I'm not familiar but I referred them to @bronze geyser 's writeup (https://bitknitting.wordpress.com/2017/11/04/using-mqtt-on-circuit-python/) and also to here.
formal plover
@tulip sleet Awesome. We also have information on using Adafruit IO API calls
I'm not sure if it's officially documented though.
Since I've been out
idle owl
@formal plover Hey Kurt!
idle owl
Totally understandable!
@tulip sleet Yeah HappyDay's writeup is the only MQTT thing of that nature I'm aware of. That's what I would have suggested as well.
stuck elbow
from the micropython forums I remember that the mqtt stuff worked pretty well, but I never got into IoT myself
manic glacierBOT
opaque patrol
So, my plans for today was to look at the problem with the mu-editor on the com port, but I got an email from them about an issue that Lady Ada just fixed. Good news - works wonderfully, bad news - what to do? hmmm microcenter has feather m0 adaloggers in stock.......π
idle owl
Good problems to have? π
idle owl
@cascade#3765 No, it's using boot.py to set the flash as read-only to either CircuitPython or the system it's hooked up to via USB. Only one at a time can have write.
bronze geyser
@tulip sleet Please lemme know if I can answer any ?
@tulip sleet oops...regarding mqtt on circuit python (the port i did)...it was simple...
idle owl
@bronze geyser You could check out the forum post and respond there if you have anything to add π
bronze geyser
@tulip sleet ...of course, now - greedy thingy that i am - I want interrupts so I can do LoRa...or I have to think about a LoRa w/ UART.
idle owl
Getting fancy!
bronze geyser
@idle owl oh righto. I will go there now. Usually I don't have anything meaningful to add. However, this doesn't seem to dissuade me from well...just rambling on and on and on and on...
idle owl
@bronze geyser And yet you always end up explaining something useful. π
bronze geyser
My you are kind @idle owl. Thank you.
idle owl
@bronze geyser The original post was rather simple, not a lot of detail to the question, but if you get on the thread you can have it notify you when someone posts to it. Then you would be able to answer if they have more questions π
bronze geyser
@idle owl -> again, many thanks.
idle owl
You're always welcome!
wraith tiger
I appear to have a not fully functional Feather M0 express. It appears to be running a blinky type program, cycling through the colors on the neopixel, but the block device portion doesn't seem to be working properly. It doesn't show up on windows 7, windows 10, linux or macOS.
Are there any tricks that might rescue it or should I just contact customer support about a replacement?
idle owl
It's likely that the file system is corrupted. Does it show up as FEATHERBOOT if you tap the reset button twice?
wraith tiger
IT just came up with "Installing device driver software".
idle owl
What OS are you currently on?
wraith tiger
Windows 7
idle owl
Did it ever show up on Win7 for you? Or is this your initial attempt to use it.
wraith tiger
Threee of the four show up as "Ready to use".
The M0 is still searching Windows update.
Now FEATHERBOOT shows up.
idle owl
Ok that's good!
wraith tiger
All the driver installs competed.
idle owl
You're trying to work with CircuitPython, correct?
wraith tiger
I tried to use it last night and this morning and the block device never showed up.
idle owl
Ok, then let's get CP flashed to your board. If that fails, it's a file system issue. It might simply have needed the drivers installed. Have you downloaded the latest version of CircuitPython for the Feather M0 Express?
wraith tiger
Yes, I'll try circuit python first but I also updated the Arduino IDE .
idle owl
Nice!
wraith tiger
I've messed about a bit with an Arduino Uno previously.
idle owl
Ok, once CP is installed, you can simply load an Arduino sketch to get started with Arduino, and then follow the steps we're about to do to get CP back on if you want to try it again.
So, have you downloaded the latest CircuitPython? I can link it if not
wraith tiger
I'm downloading the Feather m0 express uf2 file from github.
idle owl
Excellent!
Once it's downloaded, drag it to FEATHERBOOT. It'll flash a bit, and then it should disconnect on it's own, reset and then mount as CIRCUITPY.
wraith tiger
OK, done. Now CIRCUITPY shows up.
idle owl
Nice!
wraith tiger
OK, the blinky from the guide appears to be running properly.
idle owl
That's great! Nice job π
wraith tiger
Thanks for your help.
idle owl
Yeah for sure! You're welcome
wraith tiger
I'm kicking myself for not realizing that I should have bought some female headers to connect the joy featherwing & feather.
I'll just have to mess with breadboard jumper for now.
idle owl
I forgot the first time I ordered a featherwing too. Ended up ordering a Feather Doubler eventually. But it was breadboard all the way at first. π
wraith tiger
Yeah, I hope to end up with a semi compact device that I'll try to design a case to 3d print for.
idle owl
Ah then headers will indeed be your friend there.
wraith tiger
I'll be experimenting with truning it into a HID.
idle owl
Oh nice
wraith tiger
I'm wondering if it would be possible for one device to be a keyboard and mouse at the same time.
idle owl
That one I'm not sure about.
wraith tiger
Hopefully either that or toggling back and forth in software will work.
Well, I'm off to lunch. Thanks again.
idle owl
You're welcome! Have a good one!
glacial bronze
You'd have to build a custom USB descriptor for a composite device, but it could be done
For example, there's iStickToIt for the Pi 0 http://isticktoit.net/?p=1383
But besides being tied to a Linux library, the resulting device will have no Windows drivers and may not be usable outside of a *nix environment
So I'd suggest actually implementing a USB hub and separate mouse and keyboard controllers into the project.
The FE1.1s is a popular and easy to use USB hub IC that's gotten a lot of use and open source board designs in the Raspberry Pi community that you could adapt or buy outright
Adafruit even stocks one from Pimoroni I think
tulip sleet
<@&356864093652516868> who are Atom editor users: simple experimental package that does the equivalent of Eject on Windows or sync on Linux to ensure that the file you write to CIRCUITPY is completely written immediately. Go to Atom Preferences->Install->Packages and search for circuitpython-force-to-drive. https://atom.io/packages/circuitpython-force-to-drive
tidal kiln
@slender iron just ran cookiecutter to start a new lib. said 'no' to depends on register. but it still put it in the README. expected?
slender iron
@wraith tiger @glacial bronze CircuitPython is set up as a mouse and keyboard by default now.
@tidal kiln yeah, because empty is false and anything non-empty is true
slender iron
@glacial bronze yes, with a composite descriptor
@tidal kiln I should switch it to a choice
glacial bronze
Neat
tidal kiln
ah. the inputs are simple. if input then foo
slender iron
yeah
the choice option doesn't look much better: https://cookiecutter.readthedocs.io/en/latest/advanced/choice_variables.html
tidal kiln
yes and no are both true π
slender iron
yeah unfortunately
@tidal kiln if you are just cookiecuttering hold off a bit
I have an update PR for it
tidal kiln
ok. a few steps in, but not too much if you want me to start over.
tulip sleet
@slender iron I had to fix a library with non-standard capitalization. Is it worth say the library folder name is always lowercase, but the library name can be properly capitalized? Or do you think that's obvious.
slender iron
package (folder) and module names (files) should be lower case
@tidal kiln I wouldn't bother having properties for one wire address
idle owl
@tulip sleet Nice with the package, I'll try to check it out on my windows machine. I switched to PyCharm and am primarily on my Mac anyway. So I haven't done anything with Windows in a while.
tidal kiln
@slender iron then how to access?
slender iron
I was thinking just set them directly self.rom = rom
but now that I look closer I'd just save _rom and make the other properties return the value from rom
so
@property
def crc(self):
return self._rom[7]
I'm not sure what you mean by data encapsulation
I think it'd be ok if rom was writeable if th erest were dependent on it
tidal kiln
doesn't make sense to change anything after it's instatiated
and idea is to pass in rom when creating the object
tidal kiln
but also, as you can see, there's not much to it, kind of why i was starting to question if it's even needed
slender iron
can we consolidate writebit, writebyte and write together?
I think its fine
it'll make code read better
tidal kiln
yeah, but there ends up being only a few places where you use something other than the full rom
the family code check is about it
slender iron
thats ok
tidal kiln
consolidation, and even location, of the read/writes functions is up for discussion
i kind of wanted to have all that be in the device class
instead of bus class
slender iron
but _search_rom uses them
tidal kiln
mmm. good point.
slender iron
I'd advocate for readinto and write only being public but its not clear to me when individually bit writing is needed in drivers
tidal kiln
look at wait_for_convert
specific to ds18x20, but could be something similar in other devices
just keep reading a single bit to check for when conversion is complete
slender iron
no reason you couldn't just read a byte
slightly slower but one less api
still better than redoing the select
tidal kiln
wonder if other's might have some single bit dance needed for something?
yeah, should get those, and also multiple of same to test individual addressing
slender iron
we can always add it later if we need it
tidal kiln
you mean make them private now, but change later if it comes up?
slender iron
yup
tidal kiln
ok, can try re-working with only having readinto and write available, and _ the others
slender iron
thanks! overall it looks good!
tidal kiln
cool. i kind of feel like it might be a bit of a mess, with stuff spread out too much maybe. but we'll see what happens...
should i still hold off on cookiecutter? (i was starting the apds9960 lib)
slender iron
redo it now and you should be good to go
slender iron
@tidal kiln it doesn
't feel too spread out to me
I like that it matches the other classes
tawny creek
@wraith tiger I have some experience tinkering around with both, hmu with any questions
tidal kiln
@slender iron still puts in dependency, was the change to allow for saying 'no' ?
slender iron
nope, it has the same problem
the change mostly to configure it for autobuilding the release zip files
I documented the prompts π
tidal kiln
nice
also just ran cookiecutter and gave it FOOBLAH values for inputs then grepped the results
slender iron
?
tidal kiln
just a way to see where all the input values ended up
grep FOO * in resulting directory
slender iron
it does capitalization changes
tidal kiln
oh
slender iron
so look case insensitively
tidal kiln
man grep
slender iron
or look in the repo
tidal kiln
indeed, much more with -i
@slender iron maybe add that link to the prompts here? https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/creating-a-library#cookie-cutter
timber mango
I think an earlier CircuitPython M0 target stamp was released that did not have python code the end-user could modify themselves, for a pre-loaded demo (colour swirls).
The newer one (trinket M0) seems to have modifyable code loaded in it (seems to be something that could be archived to disk, and reinstated, later).
Do I have that right? I want to use the Arduino system for immediate work with the trinket M0, but I want to easily get back to factory conditions (so that I can think about the trinket M0 in terms of how a brand-new user sees it, as shipped).
tulip sleet
@timber mango You can copy everything out of the CIRCUITPY drive and save it. There is also a .zip of what it shipped with: https://learn.adafruit.com/adafruit-trinket-m0-circuitpython-arduino/downloads
@timber mango after using it with Arduino, you just need to reload it with the circuitpython .uf2.
Another thing you can do is to double-click and copy CURRENT.UF2 out of the TRINKETBOOT drive. That includes everything, including the CIRCUITPY filesystem.
unique turret
Question for @twin mica and @fast wharf , I connected a switch to GND and EN on my Feather MO Express project as you did in your CircuitPython painter. Do I need to write any code for that?
timber mango
aha. Yeah I have never archived a CURRENT.uf2. Thanks! I think that answers it, other than "Didn't it used to be true that the stamp shipped without end-user access to this stuff?"
tulip sleet
@timber mango (Copying CURRENT.UF2 only includes CIRCUITPY on boards without a separate SPI flash chip. So that doesn't work on Circuit Playground Express, etc.)
timber mango
Very familiar with going between all three environments (Microsoft, CircuitPython, Aduino) on the Feather M0 Express.
tulip sleet
timber mango
Okay, if the CPX doesn't have external SPI flashROM that's probably what was bugging me all along. I can't keep them all straight, one to the next (I have all of them).
tulip sleet
The CPX does have a separate SPI flash chip, same chip as on Feather M0 Express and Metro M0 Express.
If there's an 8-legged chip on the board, that's the SPI flash.
SPIder flash π·
timber mango
Is there any M0 SAMD21 (E18 or G18) in current production at Adafruit that cannot be restored by the end-user to factory setup?
tulip sleet
No, though I don't know offhand where the MakeCode demo is for CPX.
timber mango
For example, playing with it loading several different environments, sequentially, then restoring it to absolute factory software. As when gifting it to an absolute beginner.
tulip sleet
You'll want to upgrade to the latest version of CircuitPython and libraries anyway.
timber mango
Right. This is my one chance to see the device as if shipped new to me, so I want to remember what it can do and so forth. I have restored CircuitPython after using the Microsoft ("Make?") and after using Arduino on it (my mainstay).
tulip sleet
so if you want to gift an express board, copy CURRENT.UF2 and CIRCUITPY. If non-Express, CURRENT.UF2 is sufficient. If CPX, CURRENT.UF2 is the Makecode demo.
timber mango
One of my goals is to keep aware of how the newcomer sees these devices.
OKAY. That helps a lot.
You get what I'm after -- I probably wanted to have the CPX Makecode demo, and found I could not do so, after Arduino-deputizing the M0.
There was a gap somewhere, which is the only reason I'm asking.
One other quick question: made in China? My new CPX is marked as that.
(just curious)
tulip sleet
CPX is very popular and some places are making bulk orders; we didn't have the production capacity for that; we try to make stuff in-house.
timber mango
Sure, makes sense. Just wondered as made in NYC seems to be a point of pride.
Thanks for your help, @tulip sleet . Very much appreciated by me.
tulip sleet
yw! wish I could drop the makecode demo to you. I thought I had an unopened CPX, but apparently not.
@timber mango Aha, I do have one. Here's the demo. Note that when this is running, you single-click to get to CPLAYBOOT, not double-click
timber mango
That's a beautiful thing. It's looking at me, right now. Made an interesting noise.
idle owl
@tulip sleet You have just made the day of SO many people. That's asked after constantly.
tulip sleet
It should go in the Downloads section of the CPX Learn Guide.
idle owl
That's why I tried to recreate it in CircuitPython. Can't quite get the tones right since cpx plays tones not the same type of beeps as in the MakeCode Demo.
@tulip sleet Turns out I can take care of that!
tulip sleet
great - that section is a draft, just needs to be published after you put it in. I guess put it in the Learn Guides repo
idle owl
Oh I hadn't thought about that. I was going to just upload the file.
tulip sleet
could do that too
tidal kiln
do that
tulip sleet
gotta step out for a while
opaque patrol
I picked up an feather m0 adalogger and followed the instructions to download bossac and flash the .bin file. It works but only 47kb of free space? Is that correct?
slender iron
@opaque patrol yeah because the total file system is 64k. it should be enough to load the SD card drivers though
opaque patrol
yeah, I was spoiled by the 2mb on the cpx
slender iron
yeah, the adalogger doesn't have the external spi flash
π¬ 49 out of 55 repos need work.
tulip sleet
oy
slender iron
its easy work though π
getting the build configured right
and it tells me what needs what
tulip sleet
could you work inside the Bundle repo? Maybe you are. Since it has everything as submodules, they're all handy.
idle owl
Where does that stat come from @slender iron?
tidal kiln
anyone a read the docs guru? having issues with connecting github/rtd.
idle owl
pfft. I don't think a RTD guru exists. It's all magic and mystery. π
Seriously though, I've never dealt with connecting them. All of the RTD changes I've made were to existing repos.
tidal kiln
i've added ReadTheDocs service on the github repo.
i've manually installed repo on Read The Docs.
RTD says "no valid webhook set up" and won't update when repo is changed
idle owl
Hmm. And there's nothing about extra things needing to be changed on GitHub?
tidal kiln
it's confusing, as github has both a "Webhooks" and an "Integration & services" section for repos
idle owl
That is confusing
tidal kiln
i'm assuming the "Integration & services" is various pre-configured integrations, including one that use webhooks
and "Webhooks" is just a general purpose / manual set up
tidal kiln
i've pretty much done this:
https://github.com/rtfd/readthedocs.org#quickstart-for-github-hosted-projects
but bob refuses to be my uncle βΉ
idle owl
And the quickstart says go to "Web Hooks & Services" which really doesn't help.
idle owl
Yeah. Oi
Did you even find the RTD built in setting in GitHub?
Or did you have to create it manually
tidal kiln
it's there, on github, under "Integrations..."
tidal kiln
yep
idle owl
I realise you can read and there's literally 8 steps to this, but I've missed instructions in less before.
tidal kiln
and a button "Update service"
idle owl
Hmm. I feel like something is missing from that list of instructions considering all the menus and buttons have slightly different names now...
Like maybe GitHub changed something little up somewhere.
tidal kiln
yah, that's what it feels like
some under the hood changes, by one or both sides, and the plumbing is all broken
idle owl
Exactly
tidal kiln
(β―Β°β‘Β°οΌβ―οΈ΅ β»ββ»
idle owl
That's how I feel when dealing with RTD in general. Well that and Sphinx.
Is Sphinx at fault? or is that a different feature entirely.
tidal kiln
it's part of the process, but don't think has anything to do with the webhook part of it
idle owl
Yeah that seems unlikely
tidal kiln
rst -> docutils -> sphinx -> readthedocs
idle owl
And of course there's no troubleshooting stuff on RTD π
tidal kiln
those are the layers of the onion
idle owl
Good to know
idle owl
Yeah
tidal kiln
which would then kick rtd to run sphinx, etc.
wraith tiger
Well, I got them breadboarded and the joy featherwing test code working.
wraith tiger
I didn't have enought light on the subject while soldering and didn't realize how close the iron was getting to the jst connector, so it's a little worse for the wear. Dooh!
I'm seeing joystick positioning numbers show up randomly in the REPL when I'm not touching it.
idle owl
Hmm. I've never worked with it. Sounds like a sensitivity or threshold issue but that's just guessing based on general ideas.
stuck elbow
@wraith tiger how did you solve the the problem of missed button presses? are you using the interrupts somehow?
wraith tiger
I'm just running the exmple code from the guide so far.
stuck elbow
oh, I see
tidal kiln
@wraith tiger in this line:
if (abs(x - last_x) > 3) or (abs(y - last_y) > 3):
increase 3 to a higher number
idle owl
@tidal kiln Owait, I was right?
tidal kiln
prolly. i haven't worked with it yet either. but it's just an analog read, so...
idle owl
Huh. Nice.
tidal kiln
@wraith tiger to get a sense of how high it should be, just look at the delta between two succesive prints when you are not touching the joystick
wraith tiger
There's jitter in the horizontal from about 518 to 524 and vertical from about 482 to 494.
tidal kiln
way bigger than 3
wraith tiger
Oddly, though, physically shaking the unit doesn't seem to trigger it.
tidal kiln
makes some sense. it's just electrical noise. not related to any physical displacement of the joystick.
for example, try changing that line to this:
if (abs(x - last_x) > 15) or (abs(y - last_y) > 15):
Try a few breadboard sockets to see if the jitter is less -- the DC connections of breadboards are sketchy at best. Maybe an 0.1uF cap sprinkled like paprika over anything you suspect may help.
My tutor used to talk about 'oversampling' the ADC to get more resolution out of it. /not quite relevant here
wraith tiger
I added a J_F_F (jitter fudge factor) constant to adjust the tolerance.
timber mango
Time domain vs amplitude domain may obtain.
"If you see me operating the joystick actively, as evidenced by wild swings, pay a bit more attention and deliver additional accuracy."
The 15 vs 3 may be chunkier in response. AT some point it should feel wrong to the human ("this isn't responsive, then gives too much").
wraith tiger
My solder joints on SCL and SDA are a bit janky on the Feather. I'll have to touch them up.
timber mango
Pullups on SDA and SCL.
tidal kiln
@timber mango they're on the joywing already
timber mango
May be able to bias the joy to read slightly off-center on a quieter section of substrate, esp. after the center wears (unless it's optical).
stuck elbow
it's a pot
simple mechanical thing
I disassembled a bunch of very similar ones (only smd, not tht)
tidal kiln
@wraith tiger a better approach would be to filter the analog readings from the joystick, the current example codes does nothing except a simple threshold
tidal kiln
@timber lion you active? want to ask you about this:
https://github.com/adafruit/Adafruit_CircuitPython_FXAS21002C/blob/master/adafruit_fxas21002c.py#L69
ruby lake
an old but reliable analog input smoothing line would be something like x += (input.value - x) >> 2
ruby lake
by 4 gives you 75% estimated, 25% new value
stuck elbow
this is basically: x = x/3 + input.value/4
ah, so AB filter
x = 3*x/4 + input.value/4, sorry
ruby lake
hm, there is a wiki page about it, https://en.wikipedia.org/wiki/Exponential_smoothing
timber lion
@tidal kiln hey yeah i'm around but FYI that code is still being worked on, it's not done until the guide is published
it's working so far though but just a warning not totally sure it will work right now π
tidal kiln
i'm mainly just curious about the class level buffer. do you consider that an in-work feature?
timber lion
oh gotcha, yeah it's probably going to stay for now.. basically i want to reduce the need to create tuples
or other bytearray buffers
when using the SPI and I2C reads and writes they need to take in a buffer
but every time you do
foo = bytearray(x)
it's allocating a new byte array
same thing for tuples, like returning (x, y z)
in a tight loop like in user code reading acceleration, etc. it can be problematic
because each allocate requires new memory, and fetching new memory kicks in the garbage collection etc
you get a little perf hit but worse sometimes memory is fragmented and it can't find a large enough contiguous piece of memory to allocate from
so it might fail
it's rare with the memory we have on the samd21, but definitely can happen
it's why all the arduino drivers typically avoid new/malloc too
so i try to be strict in the drivers and just use one global buffer rather than allocating one each time it's needed
it's a trade off though, with a global buffer you can't have code running in parallel or being re-entrant (like multiple threads)
since one bit of code might start using the buffer while another is using it too... need locking then
which is a whole new world of complexity π
for now nothing really supports multiple threads or re-entrant code
but in the future it might change and might need to revisit, why i like to explicitly call it out in the driver
so it's easy to go back and find where those dependencies exist
reducing bytearray usage like that is pretty straight forward with a bytearray buffer at the class level
tidal kiln
is this any different than just having something like:
self._buf = bytearray(7)
in __init__?
timber lion
tuples are trickier and in some cases i live with creating them
since the readability and usability of tuples is easier vs. passing in some buffer and filling it with values
ah yeah the big diff is when it's on the class level like i have it there's only ever 1 buffer for every instance of the class
they all share the same buffer so it's basically a global variable
vs. like what you show it's an instance member so each instance of the class has its own buffer
that would totally work too and be thread safe / re-entrant
but if you have multiple sensors you have multiple buffers
for 7 bytes.. not the end of the world
but for bigger buffers could be troublesome
so yeah tradeoffs to make
no real right or wrong way to do it, just have to choose what priorities to focus on.. reduce allocations for better speed while losing some benefits like multithreading
or use a little more memory and be resilient to that
what might be confusing is that for whatever reason python has you reference class level variables with self
like self._BUFFER
just like with instance variables
but they're stored in differents places... i think there were a lot of discussions about this in python and no real happy resolution
and i think they rather prefer module level globals, like outside the class
tidal kiln
i think python.org bought stock in S, E, L, and F keys
timber lion
oh just habit, if nothing does multithreading then i figure better to be strict about memory usage
every little byte can count sometimes π
it also forces you to write the access code to use a preallocated buffer
tidal kiln
esp. here with cp
timber lion
vs. creating them all the time
tidal kiln
access code = ?
timber lion
ah all the code that interacts with spi and i2c devices, most drivers have byte read and write functions
like read_u8, write_u16 etc
those call the spi/i2c readinto and write functions which need a buffer
that's where it uses the preallocated one
vs creating a new one with bytearray
timber lion
ah no problem
tidal kiln
most of the code i've seen seems to be doing a pretty good job of using pre-allocated buffers, but at the instance level (which is how i've been doing mine as well)
timber lion
yeah no harm in doing that i think
timber mango
What is a good feather for robotics
tidal kiln
@timber mango what kind of robotics? do you want anykind of wireless control?
timber mango
Yeah
slender iron
@idle owl the stat is from a script adabot will run
@tidal kiln what repo are you trying to set up with RTD? I don't think its worth all the griping π
tidal kiln
@slender iron repo: https://github.com/caternuson/Adafruit_CircuitPython_APDS9960
rtd: https://readthedocs.org/projects/adafruit-circuitpython-apds9960/
idle owl
@slender iron That's neat
slender iron
@tidal kiln looks like I may have broken it
tidal kiln
it somehow worked once, when i first set it up for the third time, but not after that.
tidal kiln
and RTD still says there is no valid webhook
idle owl
@tidal kiln Wait that's what you're working on? I had some project ideas but no chance at porting it from Arduino myself.
tidal kiln
@idle owl just started today. if you're already working it, then no need for me to.
idle owl
@tidal kiln I am not working on it, I looked at it and realised I was completely out of my league.
@tidal kiln I don't know Arduino at all.
tidal kiln
@slender iron i had similar grief getting github and RTD hooked up with this:
repo: https://github.com/caternuson/readthefoo
rtd: https://readthedocs.org/projects/read-the-foo/
slender iron
looks like its going now
tidal kiln
after a full day of messing around
slender iron
try triggering a rebuild for this: https://readthedocs.org/projects/adafruit-circuitpython-apds9960/
maybe they just had a minor outage
tidal kiln
i had to setup a webhook on the github side, not use the "Integration..." one
slender iron
maybe
tidal kiln
simple as pushing a commit to trigger a rebuild?
slender iron
that works or manually through their interface
tidal kiln
oh. right. the "Update service" button under settings...hold.
github: " Okay, the hook was successfully updated."
slender iron
why are you doing that?
tidal kiln
won't that effectively make RTD do a build?
slender iron
I guess manually through the interface was ambiguous. I meant the "Build a version" dialog in a project over view on the RTD project page
slender iron
yeah, I'm fixing it
tidal kiln
@timber mango checkout the hardware used in AdaBox 002:
https://learn.adafruit.com/adabox002
tidal kiln
done
slender iron
I believe that triggers a rebuild and it should work
slender iron
looks like I can add something to readthedocs.yml to do it too
tidal kiln
what about on RTD under Admin -> Intergrations
should anything be added there?
slender iron
no, I add the integration on the github side
tidal kiln
that part still doesn't work. just did a quick commit to repo. nothing on RTD.
slender iron
which repo are you trying with?
tidal kiln
that APDS9970 one:
https://github.com/caternuson/Adafruit_CircuitPython_APDS9960/commits/master
@idle owl so you're not working on driver, just want one for projects?
slender iron
I can't see your settings on either side
adding the readthedocs integration on the github side should be enough
tidal kiln
slender iron
my github account is connected to my RTD
tidal kiln
that seems to be the main difference
idle owl
@tidal kiln Correct. I thought about trying to work on it since there wasn't one, but it's way outside my knowledge.
tidal kiln
@idle owl ok. stay tuned then, i guess.
slender iron
and I do see something listed under the RTD integrations tab
idle owl
@tidal kiln Will do
tidal kiln
slender iron
right, I do have something listed there
maybe its added automatically because my accounts are linked
tidal kiln
"Github incoming webhook" ?
slender iron
yup
tidal kiln
i think i tried adding that earlier and it still didn't work. let me try again. maybe it was the cpython2.x thing getting in the way...
tidal kiln
that's a drop down, can't screen grab it easily, goes away when focus changes
where it says "No integrations are currently configured"
slender iron
ah
I can't help you with that. why not link your accounts and re-add the integration from github?
tidal kiln
guess i could, but nothing says thats necessary to get things to work
just makes it easier to config
slender iron
docs aren't always complete or correct π
tidal kiln
even Read The Docs docs?
https://github.com/rtfd/readthedocs.org#quickstart-for-github-hosted-projects
tidal kiln
on the github side, you just have the entry under Integrations? nothing in Webhooks?
slender iron
correct
opaque patrol
I am having trouble reading the sd card. The only library bundle I could find for download was the latest build...
raceback (most recent call last):
File "code.py", line 8, in <module>
File "/home/travis/build/adafruit/Adafruit_CircuitPython_Bundle/libraries/drivers/sdcard/adafruit_sdcard.py", line 75, in __init__
File "/home/travis/build/adafruit/Adafruit_CircuitPython_Bundle/libraries/helpers/bus_device/adafruit_bus_device/spi_device.py", line 61, in __init__
AttributeError: 'Pin' object has no attribute 'switch_to_output'
Should /home/travis be in the path?
slender iron
yeah, the path is included in the mpy when its built
pass in digitalio.DigitalInOut instead of pin
tidal kiln
@slender iron thanks for the help. feel like this worked first time i went through your guide, but not since then.
slender iron
no problem, maybe they changed things
tidal kiln
yah. that's what it seems like.
pastel panther
The plot thickens!
meager fog
:>
glacial bronze
Plotter? Darn near killed'er!
meager fog
so far so good, tomorrow will try integrating it into the mu editor!
sick creek
now i started to think if I can integrate mu editor to my iot app as rosie CI is put to nanocontainer and it talk to travis ci so with circuit python where that pi connect so with touchscreen the mu editor would be nice
manic glacierBOT
hollow tartan
Have you all heard that the Python beater has been born? It is called Mongoose! Can CircuitMongoose be far behind? LOL π
meager fog
@slender iron some plotzing occuring
not totally working yet but just needs a lil buffering
will keep hackin' till it mostly-works
torn grail
Any idea when RFM69 might be supported on Feather M0 in MicroPython?
signal root
I copied the Circuit Python uf2 over to my Express playground...how do I go back to MakeCode?
I tried copying the UF2 from makecode but it looks circuit python has a hold on the board π
timber mango
It should work.
signal root
I still keep getting the circuit python repl
timber mango
That would be circuit python, yes. π
signal root
so it's definately circuit python still running
timber mango
No question on that.
timber mango
Ordinarily one might drag and drop it, obviating the need to differentiate that. π
It's going to be the exact same case, always.
Filenames and paths are case-sensitive.
idle owl
Are you double-tapping the reset button to get to the bootloader?
idle owl
The lights on the board should all flash red and then green, and you should see a drive called CPLAYBOOT. That's what you copy a .uf2 file to when you want to load it.
signal root
Now I feel silly
idle owl
Don't! It's a new thing!
FYI, when it's running MakeCode, it's a single tap of the reset button to get to the bootloader.
And you'll do that when you want to load a new MakeCode file.
If CircuitPython is installed, double-tap.
signal root
Ohhhh
Good to know π
I am digging around pxt code and circuit python - I am seeing what it takes to build a runtime for these baords π UF2 seems a nice way to go
idle owl
UF2 is pretty great
signal root
so if I drag any UF2 to the board, the board will reset with that binary running
So the "CURRENT.UF2" is the firmware that's running
In the case of circuit python it has it's own file system
where you pop your main.py
idle owl
The CURRENT.UF2 is the entire file system
idle owl
If I understand your statements correctly, yes
But I think that both systems use some of that space, that some of it is allocated
so I don't think it's that full amount. I don't know the exact numbers though.
signal root
Right
Probably half of it is for doing this dance
Fantastic π
I think I have a lot clearer picture now π
idle owl
That's great!
signal root
I ordered these M0 boards a while ago - just got around to playing with them
Very excited
idle owl
Congrats!
timber mango
It's a good system and rather complete.
signal root
So, I could use platform io to build binaries, then convert to UF2 and bam!?
idle owl
@signal root I don't know platform io, so I'm not sure.
idle owl
Please do!
I'm sure there's others who would know, but sometimes it's rather quiet here over the weekend.
timber mango
https://github.com/Microsoft/uf2
as well as
---quote---
Making your own UF2 is easy! All you need is a .bin file of a program you wish to flash and the Python conversion script.
https://github.com/Microsoft/uf2/blob/master/utils/uf2conv.py
I would hasten to add that like a parrot I can speak it but I do not know what it means, with little experience to draw on. On the other hand, I did the vagrant thing and built my own flavor of Circuit Python (99.5 percent identical to what was there already).
So it is instructive to build Circuit Python from source inside the virtual machine that vagrant instantiates or mediates (not sure its exact role).
signal root
Platform IO seems to be a good way for trageting several boards for the same embedded project. That's being said, I've had limited sucesses using it to run on the ESP8266. With M0's being used in many boardsI'm excited to see if I can get a javascript interpreter like JerryScript or Duktape running on them
timber mango
I think that's the older guide for building micropython (circuit python as well).
It's the only system I've personally built that outputs to .UF2 for a drag-and-drop firmware update of the Adafruit M0 boards.
So everything needed is found while building CircuitPython (the principles involved I mean; it's a complete and working example of this).
signal root
I just built a blink hello world in platform.io and converted it to uf2 from the converter script from the microsoft repo
timber mango
!
slender iron
Plotter looks great @meager fog !
raven canopy
in #help-with-projects, @cunning crypt mentioned CP doesn't support multithreading. Instead of asking over there, I figured here would be a better place. I haven't dipped into CP or MP yet, but I might give it a go on the next project. Does MP support multithreading, and is it a planned feature in CP?
cunning crypt
@raven canopy I don't belive either the M0 or M4 are capable of multithreading
raven canopy
ahh. well, that would be a good reason not to plan for it or attempt it. π
cunning crypt
I'm sure there are ways to do software multithreading but.... Honestly, it'd probably slow the device to a crawl
slender iron
@raven canopy micropython does have some threading support but its untested in circuitpython. its not a high priority for us due to its complexity
torn grail
If I'm using a Feather 8266 and trying to drive a 7segment display, should I be using Micropython or CircuitPython? I seem to be stuck getting MP to work with the 7Segement libraries, and maybe more luck with CircuitPython??
raven canopy
@slender iron cool. Thanks for the info.
slender iron
@torn grail either should work. Just make sure the library matches the core implementation you are using. IE only circuitpython libraries work with circuitpython
torn grail
I pulled the MP libraries for matrix, but can't seem to get the 7segment to load right
doesn't appear to be the docs on 7seg, as there are for matirx.
import ht16k33_seg
display = ht16k33_seg.Seg7x4(i2c, address=0x74)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "ht16k33_matrix.py", line 18, in init
AttributeError: 'Seg7x4' object has no attribute 'WIDTH'
slender iron
@stuck elbow may be able to help with micropython
otherwise try circuitpython
there is example code for it here: https://github.com/adafruit/Adafruit_CircuitPython_HT16K33/blob/master/examples/matrix.py
actually thats a matrix example, 7 segment is here: https://github.com/adafruit/Adafruit_CircuitPython_HT16K33/blob/master/examples/segments.py
torn grail
Yeah -- thats matix, not 7segment
oh, there's segments
I'll do some more reading. Thanks @slender iron
slender iron
π
stuck elbow
@torn grail you might have found a bug
@torn grail yup, I ported the matrix part to use framebuffer and forgot that the 7seg part uses it too
and my "I" I of course mean a friend who made a pull request
So I have this CircuitPlayground Expresss Developer Editior thingy, and it doesn't have UF2 bootloader on it. I'm trying to flash circuitpython with bossac on it. Double-clicking the reset gives me a pulsing red led, and then:
BOSSA-adafruit_17_10_19/bin$ ./bossac -e -w -v -R ~/Downloads/adafruit-circuitpython-circuitplayground_express-20171203-b31c2ea.bin
SAM-BA operation failed
What am I doing wrong?
torn grail
@stuck elbow , so its bug that seven segment isn't there? I thought it was just me.
stuck elbow
@torn grail it's a bug
@torn grail working on it now
@torn grail do you want to test?
torn grail
Good, because I thought I was losing my mind!! (do I get a t-shirt?) LOL
I'd love to test. Let me know when to check it out again
stuck elbow
@torn grail try this: https://github.com/deshipu/micropython-adafruit-ht16k33
torn grail
just downloaded it. now I gotta ampy and then I'll test it out
@stuck elbow -- nope
display = ht16k33_seg.Seg7x4(i2c, address=0x74)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "ht16k33_matrix.py", line 18, in init
File "ht16k33_matrix.py", line 24, in _write_cmd
OSError: [Errno 19] ENODEV
(is there a way to make a code block in discord...)
torn grail
display = ht16k33_seg.Seg7x4(i2c, address=0x74)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "ht16k33_matrix.py", line 18, in __init__
File "ht16k33_matrix.py", line 24, in _write_cmd
OSError: [Errno 19] ENODEV
yay
torn grail
it shows three of them [112, 113, 114] becuase I have them at 0x70, 0x71, 0x72
stuck elbow
but not on 0x74?
stuck elbow
thanks for finding the bug, we didn't test that part of the code
so I'm making a pull request to the adafruit's repo
torn grail
I always find broken stuff. Thats my job!! But usually its in Java. So, I'm branching out.
stuck elbow
Ouch, java
but at least we have it now tested by a professional tester π
let me know if you find any other problems
torn grail
I'm starting to port most of my old Ardino code to MP or CP, so I'll be testing out a lot of stuff!!
If anyone starts work on porting RFM69, I'd love to test that out. I have a project using it right now.
stuck elbow
I got those modules with the mind to write that library over a year ago
never got to it
it's just too complicated β it doesn't help that I never used the Arduino libraries either
torn grail
If there is anything I can do to help, let me know
stuck elbow
I guess I just need to start building something that uses it
and then I will need it
got the NRF21L01+ one working, though
torn grail
btw -- back to seven seg...
display = ht16k33_seg.Seg7x4(i2c, address=0x70)
display2 = ht16k33_seg.Seg7x4(i2c, address=0x71)
display3 = ht16k33_seg.Seg7x4(i2c, address=0x72)
that should work to define three different displays, right?
stuck elbow
yes
torn grail
The first works, but the second two are giberish on the display
stuck elbow
that is strange, because they don't share any state
do you also get gibberish if you have only one, but not on the default address?
torn grail
>>> display.fill(0)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'Seg7x4' object has no attribute 'fill'
Phil would be said if there was no fill. LOL
stuck elbow
hmm, I might need to add that fill, actually
torn grail
there it goes
stuck elbow
I'm using it in the 14seg display code
torn grail
took out the fil command, now its hapier
timber mango
@stuck elbow I don't know how to simulate your initial conditions -- the .UF2 ships installed and I don't know how you got to the state where there isn't one present.
stuck elbow
@torn grail the fill() method is back https://github.com/deshipu/micropython-adafruit-ht16k33
@timber mango no worries
I guess I need to compile and flash the bootloader with the jlink
timber mango
/home/nisnis/.arduino15/packages/arduino/tools/bossac/1.7.0/bossac -i -d --port=ttyACM0 -U true -i -e -w -v /tmp/arduino_build_837207/thisprogram.ino.bin -R
That's the command line Arduino IDE used.
worn birch
oh gosh ... just got the You are running in safe mode... speel ... hmm ... no traceback info... is that stored somewhere after one of these something really bad happened. events?
stuck elbow
BOSSA-1.8/bin$ ./bossac -i -d -e -w -v -R ~/dev/circuitpython/uf2-samd/build/circuitplay_m0/bootloader.bin
Set binary mode
version()=Adafruit CircuitPython 0.8.3-1-gefc15375-dirty on 2017-01-18; Adafruit Circuit Playground M0 (Experimental) with samd21g18
Connected at 921600 baud
readWord(addr=0)=0x30303077
readWord(addr=0xe000ed00)=0x30303030
readWord(addr=0x400e0740)=0x23342c30
readWord(addr=0x400e0744)=0x30304577
Device unsupported
I will try it on mine.
stuck elbow
same with BOSSA-adafruit_17_10_19
idle owl
@worn birch Revert your code to whatever previous state, or backup and remove the code you're working on. Then reset the board.
opaque patrol
@stuck elbow is it possibly a circuit playground classic?
stuck elbow
it has a flash chip on it
robust coral
QuickQuestion: is there a millis() function on CircuitPython? Where? I'm using a CircuitPlayGroundExpress and can't find it.
idle owl
@worn birch Every time I've run into it, it's been due to a change in my code that failed. I don't think traceback is stored anywhere.
stuck elbow
and says 'express' on the back
idle owl
@robust coral time.sleep(seconds)
stuck elbow
@edgy hatch there is time.monotonic()
idle owl
And that
robust coral
@idle owl: not sleep, just need to know the time
opaque patrol
@stuck elbow okay, its an express....I always check the simple things first
torn grail
I'm time.sleeping(500) right now, cuz I forgot it was seconds, not microseconds! LOL
idle owl
time.monotonic() is will give you that but it's relative, to when the board started I believe. So you can compare it to itself and subtract to get a period of time if that's what you're looking for.
robust coral
that's what I need. Will try Thx
timber mango
@stuck elbow Mine does the same as yours. Not broken! Arduino is using some secret sauce.
ard $ /home/nisnis/.arduino15/packages/arduino/tools/bossac/1.7.0/bossac -i -d --port=/dev/ttyACM0 -U true -i -e -w -v /home/nisnis/Downloads/CPX-MakeCode-demo.UF2 -R Set binary mode readWord(addr=0)=0x30303077 readWord(addr=0xe000ed00)=0x30303030 readWord(addr=0x400e0740)=0x23342c30 version()=wE000ED00,4#w400E0740,4#V# chipId=0x23342c30 Unsupported processor Send auto-baud Set binary mode readWord(addr=0)=0x30303077 readWord(addr=0xe000ed00)=0x30303030 readWord(addr=0x400e0740)=0x23342c30 version()=wE000ED00,4#w400E0740,4#V# chipId=0x23342c30 Unsupported processor No device found on /dev/ttyACM0 ard $
opaque patrol
@stuck elbow I think it isn't going into bootloader mode.when you double click the reset button, all neopixels should turn green
idle owl
@stuck elbow You ordered a Circuit Playground Express and it shipped without UF2 bootloader installed?
stuck elbow
@idle owl I got it before it was on sale
idle owl
Oooh.. ok. I was incredibly confused.
stuck elbow
for testing
idle owl
I was going to say regardless of the fact that you have the tools to fix it, you really need to let support know. But that's a totally different story π
stuck elbow
it has some differences, for example a transistor in place of the audio amplifier
idle owl
I'm on the same page now.
timber mango
HEY it looks like I'm in an unusual state! No program is loaded on the CPX now. Will look at it from the host operating system to see what (if anything) is in CURRENT.UF2
CPLAYBOOT is still there.
stuck elbow
got the bootloader burned \o/
the jlink couldn't see it, but my trusty stlinkv2 worked
idle owl
@stuck elbow Nice!
timber mango
I think my mistake was the file format -- used BOSSAC to do an .UF2 which may be a forbidden combination.
stuck elbow
OK, I tried the AudioIO example, and the transistor next to the speaker got hot
and I heard nothing and smelled smoke
tulip sleet
@stuck elbow The UF2 bootloader build also builds a .ino you can compile and load to install UF2. Moot point now. and the smoked transistor: sorry...
timber mango
Return the smoke to Adafruit (need an RSA number for that) and they will analyze it for future product development.
stuck elbow
I guess that version has the audio circuit organized differently
@timber mango this is a product that was never sold anyways
timber mango
@tulip sleet How hard is it to get the CPX to a state where there is no bootloader present -- aren't there guards for that? My understanding might be described as "No way the end user is going to need a JTAG connection to revive this target board"
tulip sleet
@timber mango the bootloader is protected via the BOOTPROT fuse settings. You'd have to unset those to overwrite the bootloader. Do you think you're in that state? You didn't break anything by loading a UF2 with bossac - it just won't load a usable program.
timber mango
No I'm back to the Makecode demo as if nothing had happened. I used the drag and drop interface, as usual, to restore normal operation. Double-checked: Arduino compiles and loads my usual project, no problem, after all that (see scrollback). I could't break it (well, not yet, anyway).
Like everyone else I'd like a software kit that lets me break and reload the bootloader, by setting the fuses, via the same USB interface. I can't remember what was said about that, but I thought it was a difficult thing to do (and wouldn't happen by random writes, due to the fuse mechanism being improbably (or impossibly) triggered.
In the long run I want to be able to modify the stored program on-device (to augment it). My understanding was that can only be done from behind the 'fence' between low storage memory and high storage memory (my regular 'program' would have to operate in bootloader-space).
Well, I do have a working means to store and load the interpreted program on-device (on SPI flashROM) .. to load forth programs, or save them (I can author new forth code on-device for any of the Express series M0 targets, now).
However .. the recapitulated code (stored on flashROM) is compiled into the 32k system ram address space, so there is a 32KB limit to the size of a newly-compiled program, plus all the existing stuff that lives there (probably halves the RAM available for compiled code).
I would really like to make use of the 256KB on-chip storage. π
stuck elbow
that is way over my head
by the way, I wanted to test audio on the CPX because on my own board it's very bad
the wav's I'm playing all click and buzz
and I was wondering whether the problem is with my circuit or with the software
timber mango
The Makecode demo plays a very short sound during boot.
But it's an interesting sound. Like a voice in MIDI or something.
stuck elbow
I'm generating those wav files with sfxr
idle owl
@stuck elbow I have 4 in front of me that play wavs just fine. Could be software?
stuck elbow
I suspect they don't begin/end on a neutral position, hence the clicks
robust coral
you said "all" click and buzz tho.
stuck elbow
I have a video
robust coral
is it only at start and end of a sample?
stuck elbow
idle owl
Hmm, also if the speaker is initialising weird in the code, it'll click when it initialises and when it stops.
torn grail
That's awesome!
stuck elbow
there is a "pew" sound and an explosion sound, you can hear them behind all the clicks and buzzing
idle owl
@stuck elbow That looks great! About the sound though hmm....
timber mango
ruminants
stuck elbow
I copied the audio circuit from the CPX, but I have a different speaker, because I couldn't find the same
I'm really bad at analog circuits
some of the buzzing is my finger touching the audio in
(I moved the audio circuit out the way in the next version)
timber mango
Some audio circuits have what amounts to a squelch. When they gate-on it can be painful to listen to in headphones.
(notably most modern 2-way radios have this flaw)
stuck elbow
here the amplifier is always on
timber mango
You know it's not key-clicks from actuating the 4-way controller?
Maybe bridge the momentary switch contacts with a capacitor.
stuck elbow
no, the worst clicks are actually at the beginning of the sample
probably because the sample doesn't start at 0
also, after playing a sample, the circuit will stay buzzing
as if the DAC wasn't released
timber mango
Sounds like you need a low-level authoring tool for the raw media file. Or a 'blesser' for that kind of media file.
Can you obtain a proper file that does work with the rest of the setup the same?
idle owl
There are sets of wav files for CPX projects available if you need something to test. But you said it's not quite the same as the CPX, so I'm not sure whether it'll help.
timber mango
How long of a media file can the CPX present? Three seconds of clean audio?
manic glacierBOT
worn birch
@stuck elbow Sweet ... so small ... looks cool ... can't imagine any teachers are going to be fans though.... but fun none the less!
stuck elbow
@worn birch Thanks. I have a more teacher-friendly device, with an 8x8 LED display instead of the LCD.
tulip sleet
@timber mango re your earlier message about the bootloader: the bootloader (either arduino-style or UF2) loads a program in to flash, not into ram. That's what you want, right?
idle owl
CPX can play wavs for longer than that. The limiting issue is the file size, in my experience anyway.
timber mango
@tulip sleet I want to start with an Arduino that presents a text interpreter in the serial USB, that allows compilation into 256KB flashROM space (on-chip storage). I already compile to the 32KB RAM storage space.
So I want to write to $4000 not $2000000
(never mind how many zeroes hehe)
The code fence is near $4000 I think; I'd have to (as I understand it) code for the space below that address, and essentially author a custom bootloader.
Because programs running north of that address are not allowed to write to flash.
So the work-around people use is to run a very small kernel in bootloader-space that is permitted to write to the rest of the flash space (just as the bootloader has such permission).
stuck elbow
that's weird, because the non-express versions of CircuitPython have the filesystem in flash
in the on-chip flash, I mean
timber mango
Yes there is 256KB of that.
Without the SPI flashROM chip (which is an additional 2MB)
The thing is, that SPI flashROM isn't addressible (at all). It's a filesystem.
stuck elbow
and I don't think they need a special bootloader?
well, how do you write to a filesystem if you can't address it?
stuck elbow
to the internal on-chip flash?
timber mango
It's a peripheral.
The internal flash is in the memory map. It's not a peripheral (per se).
tulip sleet
@timber mango I think you may be misinformed. It's perfectly possible for a program not in the bootloader section to write to flash
as deshipu mentions, that's how we write to the top 64k of flash to implement in the in-flash filesystem for CircuitPython.
timber mango
Okay. I don't understand enough of it to have a lucid comment. I thought this was kind of cast in stone.
tulip sleet
not at all. but you have to erase a whole section at a time and then rewrite it. Are you writing your interpreter in Arduino or bare-metal?
timber mango
Arduino.
tulip sleet
I found a library for 328 flash writing. Will ook for one for SAMD flash writing
timber mango
I am himself. I have your cmaglie bookmarked since Feb 2017. I wonder why I moved past without looking at it!
tulip sleet
cmaglie's code has some functionality to act as an eeprom and write only a small area, but take a look at FlashStorage.cpp and .h for a general interface.
timber mango
If I can buffer the segment of the flash to be rewritten I don't care how large the erase is, so long as I can restore the old part to be able to write the new part.
But generally, they are tiny writes into SRAM. Just a handful of bytes at a time, as that's how a forth compiles.
tulip sleet
I think that's what cmaglie's lib does. You'll want to buffer the writes so you don't rewrite too many times and wear out the flash.
timber mango
Yeah I would probably resort either to SRAM or maybe FRAM (external SPI or i2C) for the small incrementals, and then do a write when I was satisfied.
torn grail
@stuck elbow -- Back to ht16k33 for a sec. I think I found another bug. When I do a display.number(1111), then display.show(), I see "1110" on the 7-segment display. π
strange, 1234 works
stuck elbow
I really need to dig out my modules and test this properly
stuck elbow
I will do that tomorrow, it's past midnight here
torn grail
thanks, man. nite nite
stuck elbow
just to make sure, display.text('1111') works?
stuck elbow
what do you mean by "nope" exactly? what is shown?
stuck elbow
when you use .text and not .number too?
torn grail
I can test with you tomorrow if you want. I'm in California. Pacific time.
I'm 7segment, so no alpha. just numbers
stuck elbow
I'm in CET, 9h difference
sure, but .text will still work with numbers
that's what .number uses internally
torn grail
I'll try out some more things to see what I can "break"
stuck elbow
thanks
sacred blade
@stuck elbow it's because of rounding errors with the 30bit floats. "{:f}".format(1111)gives '1110.999941' on a m0 ...
stuck elbow
ouch
stuck elbow
I didn't expect to hit that with just 4 digits
@torn grail you can use .text and format the number to a string yourself
torn grail
I'll try the .text and see how that goes
tulip sleet
that looks like a bug; {:f} should do better than that
stuck elbow
cpython has some smart rounding on display
tulip sleet
yeah but 1111 should fit in 22 bits of mantissa
stuck elbow
I think it always picks the shorter representation when there is ambiguity
tulip sleet
>>> "{:f}".format(1111)
'1110.999941'
>>> 1111.0
1111.0
something is wrong with the arithmetic in format()
stuck elbow
@torn grail you really have a talent!
sacred blade
Yes, ```
str(1111*1.0)
'1111.0'
So there is indeed something wrong in format...bleah, py/formatfloat.c normalizes the number to be between 1 and 10 by multiplying by 0.1 repeatedly, even for f vs e format. [EDIT: slightly wrong, but close to the problem]
then it multiplies it by 10.0 repeatedly as necessary.
stuck elbow
0.1 is a great choice π
tulip sleet
ugh
stuck elbow
having infinite representation in binary an all
tulip sleet
i'll see if someone filed an issue on this in MicroPython.
timber mango
I managed to not be able to reinstall CircuitPython, by attempting to update /lib which I think must be corrupt (or non-existant). Cannot get a REPL. Arduino works fine; Makecode demo .UF2 uploads and runs fine on CPX. Seems to me the system has retained the wrong things I did to the flashROM where main.py and /lib live.
Was trying to install the bundle,
adafruit-circuitpython-bundle-2.x-20171128.zip
idle owl
Try the flash eraser. https://learn.adafruit.com/adafruit-circuit-playground-express/troubleshooting#for-the-circuit-playground-express-feather-m0-express-and-metro-m0-express
timber mango
@idle owl thanks
idle owl
You should be able to drag the /lib folder to the CIRCUITPY drive to load it when CircuitPython is loaded. It's possible it gave you issues if it reset before it was done copying something.
timber mango
It tried to overwrite existing /lib earlier. The flash eraser worked and I probably have a REPL now (can see a drive for the device in the filemanager)
idle owl
Excellent
When you load a file, the board will soft reboot, which makes sense because that's how it reloads when you update code.py. However, if you're overwriting the existing /lib folder, it could lead to a soft reboot during copying which could have corrupted it.
timber mango
Everything looks good now.
idle owl
Ok good
timber mango
I did not have a /lib this time. That was the problem, a pre-existing /lib
REPL is there. But I'm python braindead so I don't know how to test it well.
idle owl
I've overwritten an entire /lib folder previously though, and not had it end up corrupted, so it's not a guarantee that it will cause issues. It does, however, increase the likelihood as it may try to reload as each file is finished copying.
timber mango
It wanted me to make a decision to replace every file.
idle owl
I'm never great with testing in the REPL unless I have code I'm trying to debug. Just to check if it's working, I'll do ```python
import board
dir(board)
That wasn't flying too well. I must've done something wrong.
idle owl
I'm on a different OS, so that could be another reason. You may not have done anything wrong.
At least I assume I'm on a different OS. What are you running?
timber mango
Debian for the amd 64 on a laptop.
``>>> import board
dir(board)
['A0', 'SPEAKER', 'A1', 'A2', 'A3', 'A4', 'SCL', 'A5', 'SDA', 'A6', 'RX', 'A7', 'TX', 'LIGHT', 'A8', 'TEMPERATURE', 'A9', 'BUTTON_A', 'D4', 'BUTTON_B', 'D5', 'SLIDE_SWITCH', 'D7', 'NEOPIXEL', 'D8', 'D13', 'REMOTEIN', 'IR_RX', 'REMOTEOUT', 'IR_TX', 'IR_PROXIMITY', 'MICROPHONE_CLOCK', 'MICROPHONE_DATA', 'ACCELEROMETER_INTERRUPT', 'ACCELEROMETER_SDA', 'ACCELEROMETER_SCL', 'SPEAKER_ENABLE', 'SCK', 'MOSI', 'MISO', 'FLASH_CS']
``
timber mango
So that proves the bundle loaded?
idle owl
That's everything you can load out of board when you import it. If you did not already know that.
Oh... no
Do um... let's see. import adafruit_dotstar
timber mango
great. yeah now we're talking.
idle owl
Then do a dir(adafruit_dotstar)
timber mango
thanks.
idle owl
If that doesn't give you a module failure, that lib is loaded correctly, and it's a safe bet the others are too. However, without testing each one, you wouldn't know for certain. It is incredibly likely they rest are fine though.
timber mango
``>>> import adafruit_dotstar
dir(adafruit_dotstar)
['time', 'digitalio', 'DotStar', 'busio', 'name']
``
No we're just looking for a very basic "yeah it looks like it loaded because dotstar won't be there if it didn't load" kind of confimations.
idle owl
Excellent, then that's exactly what you're looking for. It's a simple way to check if a module is present.
I get the same thing, I checked. You're good there.
timber mango
Thank you. I've rebuilt CircuitPython in vagrant session many times (earlier in the year) but I have no memory of doing much of anything in REPL itself. I modified the REPL to do what I wanted (display status messages upon entry) and maybe something else trival.
It's like I never touched it in my life. Effects of aging I suppose.
idle owl
Could be. Or it could simply be coming at it from a different angle makes it just enough different to be foreign. For me if it's only a little different, it's far more difficult than if it's completely different.
timber mango
high-contrast different. or almost-the-same different.
idle owl
When I first started coding, MakeCode was incredibly difficult for me because I knew just enough Python to have the visual codeblocks seem impossible to understand. After a few months, I went back to it when a friend bought a CPX and wanted to start programming, and it made SO much more sense at that point. I'm happy I skipped MakeCode though, it got me right into CircuitPython. But I'm really glad MakeCode exists as it's give my friend a chance to learn so much.
timber mango
I found MakeCode to be a fun diversion and a very visual way to learn it, but I didn't spend that much time on it. I do remember editing the source the drag-and-drop interface created, in a regular text editor, once I was able to understand what it was doing.
Also I made a local installation so I didn't use the Internet to code in MakeCode, after I understood how to create the same environment, locally.
idle owl
Nice
timber mango
Adafruit has rolled out CircuitPython so fast that it kind of moots what MakeCode can do for the CPX.
idle owl
I disagree. I would have agreed with you before my friend bought hers. We tried a little CircuitPython and she was completely lost. But she's really starting to understand code structure and function using MakeCode.
timber mango
Well, one is visual and drag and drop, the other is typing and reading reference literature. π
idle owl
Valid point. But for someone who has literally never programmed anything in her life, MakeCode has been a great resource to get started with.
ruby lake
is the simplest way to write 16 bits to an SPI device something like x = bytearray(2) then spi.write(x)
assuming of course the SPI is configured
timber mango
Any port in a storm. We need more approaches because once people realize how it works it doesn't matter so much what door they walked in, so long as they're now inside.
ruby lake
and you give x its actual value
idle owl
@timber mango Exactly.
timber mango
I need to go soak my head in brine. For a long time. Thank you for your help, @idle owl
idle owl
@timber mango You're welcome! I hope you get some good rest
timber mango
whoa metro m4 testers whoa. I am going going gone. Goodnight.
idle owl
Night!
@ruby lake I'm not sure. I wanted you to know you hadn't been lost in the conversation. Late night on weekends can be quiet here. I would suggest asking again tomorrow if you have the option.
ruby lake
sure, no rush π
ruby lake
aha, thats how to do it. Enumerate a list of pots to read as pot = [ analogio.AnalogIn(board.A1), analogio.AnalogIn(board.A2), analogio.AnalogIn(board.A3)]
then I can loop in pot[i].value
idle owl
Nice!
ruby lake
is there a time.millis() or similar library call? time.monotonic is not granular enough
idle owl
I think there's only time.monotonic().
torn grail
Has anyone ever used the hardware timers in MicroPython to trigger functions at a specific interval? Other that PMW...
wraith phoenix
Just checking out CircuitPython because I have never programmed in Python before.. . .
idle owl
Excellent! Welcome!
idle owl
Ooh. Do you have a particular board you're using?
wraith phoenix
Just got a free Trinket M0 w/ my Oct. Adafruit order.
idle owl
Nice! Checkout the Trinket guide! It's full of examples. https://learn.adafruit.com/adafruit-trinket-m0-circuitpython-arduino/circuitpython
wraith phoenix
Already read ladyadas's https://learn.adafruit.com/adafruit-trinket-m0-circuitpython-arduino/circuitpython-pwm?view=all
idle owl
Well done there then! That's the best place to get started. If you've already finished that, then you'll want to start looking at other projects using CircuitPython. https://learn.adafruit.com/category/circuitpython
Not all of them will work with Trinket, but you might find some good information in other projects as well.
wraith phoenix
Am getting read to install avr-gcc for linux on my PC to program both an ATMega328p 8-bit micro plus the trinket. What do I need to look for/install from my debian repo's ?
tawny creek
it mounts as a usb drive with a file system
wraith phoenix
A 32-bit micro the size of my thumb - HA!!!!
tawny creek
you dont need anything else other than a text editor π
@wraith phoenix https://playground.arduino.cc/Linux/Debian
for arduinoo
tulip sleet
@torn grail @stuck elbow re floating point formatting: https://github.com/micropython/micropython/issues/1450#issuecomment-348847212
wraith phoenix
thank you for the URL, kattni. Will look at it tomorrow.
idle owl
@wraith phoenix We're here for questions as well! You're welcome!
wraith phoenix
Thanks for the URL, cascade. Could be my secret to success π
idle owl
@wraith phoenix Also, check out the Welcome to the Community guide. Depending on your experience level, there are other resources that might be of use to you. https://learn.adafruit.com/welcome-to-the-community/welcome-to-the-community?view=all
wraith phoenix
Thanks so much, kattni. Got your URL too. I t's been a few years since I mucked w/ IRC, so please be patient with me.
idle owl
No worries!
wraith phoenix
Actually, it's half the size of my thumb π
tawny creek
@wraith phoenix π im excited to build one from scratch when i get the pcb's from osh park
tawny creek
@wraith phoenix just for practice and experience, i want to design and build my own board eventually :3
wraith phoenix
@cascaede Great! I am new to OSHW. But am a longtime linux user.
cunning crypt
@idle owl What's involved in writing a Learn article? I feel like my examples would go great alongside an article explaining just what it is that they're doing.
...That question probably would have gone better in #learn-adafruit-authors but too late now. Got CP on my mind.
idle owl
@cunning crypt Typically they'll have you write up something on Instructables or another site like that to have an idea of your writing. Then you get added to the Learn system, run through the tutorial type thing for it. Once you're in that system you can use other guides to get an idea how they look and flow, and then you write up your own.
It's all good. It'll be CP related in the end anyway. π
cunning crypt
Honestly, they'll probably just be able to look at the pile o' comments I'm leaving in this "example"
It's basically a page of a learn article by itself
idle owl
That's great! But it's still a great idea to consider writing up a guide for it, and your comments will make it easier on you later!
cunning crypt
Oh, yeah! I want to write a guide.
Basically, this example wouldn't be an example at that point. It's... too flimsy as an actual program. The other examples would each have their page, etc.
idle owl
I ended up taking out some of the comments in my code when I put it in the guide because it was too much with all the explanation around it. But I definitely wished I had commented more when I was working on the guide. So you're in a good place.
cunning crypt
I learned long ago: Comment the heck out of everything.
That way, when you pick up the project in six months, you know what you were thinking
cunning crypt
Go ahead
slender iron
@wraith phoenix For the Trinket M0 you'll need the arm gcc toolchain not avr
@ruby lake what resolution do you need that time.monotonic doesn't have? bytearray is the way to go for 16bits on SPI
ruby lake
@slender iron I've set up a variable dacData = bytearray(2) and put my value into it, but the spi.write(dacData) gives an error saying "object requires buffer protocol"
ruby lake
x = (note - 24) * 273 #Magic formula for correct semitone scale (n-t)*4095/60
dacData = bytearray(2)
dacData = (x >> 2) + ((x >> 1) & 1) #Round to nearest int for 0.5 LSB accuracy
dacData |= 0xB000 #channel 2, gain=1x, no vref buffer, output active
the SPI write examples don't show working with a variable, only literals
slender iron
treat the byte array like you would a list
dacData[0] = first byte value
dacData[1] = second byte value
the second assign is replacing the bytearray with an int
its the benefit/problem with python. variables can change type
ruby lake
@slender iron ah, so something like dacData[0] = (x & FF00) >> 8