The reason I was asking was -- the guide has you load a "combo" .bin file but there is no step to then load the update_bootlader so it just loads a slightly older version
#circuitpython-dev
solar whale
tulip sleet
i'm not sure what the difference is between bootloader.bin and tinyuf2.bin
we've been using the combo to get people started; it's your choice whether to update or not. I don't know if it's necessary to get the functionality you want
solar whale
but I think you clarified -- if the loaded version is already the latest, then loading update_bootlader.uf2 will essentiall do nothing
Thanks!
tulip sleet
it would be good if the version numbers were in the file names...
solar whale
and it looks like the combo also includes an application file (in this case, the arduino demo)
tulip sleet
and it looks like the combo also includes an application file (in this case, the...
I asked hathach, who said:
- The bootloader.bin is 2nd stage bootloader,
- tinyuf2.bin is our "bootloader" that does uf2 upgrading and yes, it is 3rd stage but actually factory app https://github.com/adafruit/tinyuf2/blob/master/ports/esp32s2/README.md
- update-tinyuf2.uf2 (there is no .bin for update), is for upgrading the tinyuf2 (3rd stage), it will not able to update the 2nd stage bootloader.bin
solar whale
@tulip sleet Thanks! Still trying to understand all the pieces, but it is making more sense to me know. I had a very confused notion of what the update_bootloader was doing.
manic glacierBOT
manic glacierBOT
onyx hinge
@slender iron do you want me to look into the _transmit problem?
onyx hinge
do you remember, was it krayola or you who changed the core behavior so that this could work in the first place?
I was looking for that commit or PR to refresh my memory
slender iron
we both had a role in it I think
I think _transmit is meant to be on the subclass of the pixelbuf
so perhaps the lookup is happening on the super class incorrectly
onyx hinge
My first suspicion was that the core behavior changed here was reverted in one of your merges; I don't think there are any tests of it.
slender iron
yup, that could be the spot
idle owl
This is the last one I see, from 23 April. https://adafruit-circuit-python.s3.amazonaws.com/bin/mpy-cross/mpy-cross.static-x64-windows-6.2.0-4-g9d8545e5e.exe
idle owl
@onyx hinge Thanks for looking into it. We were, in the thread, questioning whether it's worth waiting until the merging is complete. But neopixel is a significantly used library, so even though we haven't released 7, it's probably worth doing. (Contrary to my comment. Mostly I didn't want a community member ending up in a rabbit hole that we needed to handle.)
onyx hinge
# Welcome to CircuitPython 2.2.0 :)
``` huh apparently I have never done anything with this boardl. o. l. Evidently not.
onyx hinge
hum this board is pretty unhappy with main. I get a hardfault from import _pixelbuf. ouch.
that's not the same as what users are seeing, is it?
idle owl
No
idle owl
that's not the same as what users are seeing, is it?
If you run NeoPixel code, where you specify a pixel, e.g. pixel[0] = (255, 0, 0), it fails with a _transmit not found type error.
So it's not importing _pixelbuf that's the issue. I did that successfully. As long as I didn't specify pixel[0].
I'm surprised the rainbow code works though because it iterates over pixels..... pixel[pixels] = colorwheel(pixel_index & 255)
That's under for pixels in range(1): (only one LED)
idle owl
Ok I'm running super simple code and it's not working and I am confused. Anyone have an ItsyBitsy RP2040?
onyx hinge
With "main"?
idle owl
"main"?
idle owl
Yes. with main.
It's running, but nothing is happening
is my issue.
(Limor asked me to test with latest for everything I do so I find bugs.)
onyx hinge
I bet you've found bugs ๐
idle owl
There are no comparable boards to test on. QT Py has no built-in little LED. and Feather, you can't use the BOOT button as a user-button in code.
I'll try it on 6.2.0 I guess.
Nothing.
It's not a main bug.
onyx hinge
oh really
idle owl
Apparently.
Ok reloaded the latest.
(no change, but I was not expecting one.)
the LED isn't blinking. I know it works because it was blinking with the pico-sdk demo that it shipped with.
I checked the pin....
Pretty sure I have two, I'll try another one.
Not blinking.
How did I break blink? ๐
solar whale
@idle owl I do have na itsybitsyrp2040 -- do you want me to try something? may take a few minutes find it....
idle owl
<@!330227457296957440> I do have na itsybitsyrp2040 -- do you want me to try som...
Yes please!
oh wtf.
I figured it out. The board def is wrong.
BUG FINDING.
onyx hinge
someone should fix it then ๐
solar whale
Glad I could be of assistance ๐
idle owl
Yes, on it.
idle owl
Glad I could be of assistance ๐
Thank you, Jerry! ๐
solar whale
I actually found it.
onyx hinge
sure
idle owl
Now we wait. ๐
onyx hinge
@slender iron re the above change, do we want it in 6.2.x and merge 6.2.x up to main?
slender iron
I think @tulip sleet has thought more about this. Ideally it'd be 6.2.x but we have plenty on main that we'll want in 6.2.x when we release
so I don't think it matters
onyx hinge
OK
onyx hinge
@idle owl I tagged you on https://github.com/adafruit/circuitpython/pull/4688 as you will be able to test
idle owl
Cheers.
onyx hinge
out to lunch! ๐ฎ
idle owl
YES.
Button pin was wrong, and I fixed it.
Also..... who has two thumbs and built CircuitPython on the first try?
๐ช
(Im going to verify the rest of the board def before committing again.)
jaunty juniper
so, on a QT PY M0 Haxpress, since merging 1.11 the board doesn't show up and messes with my mac's USB instead, making the keyboard and mouse stutter. The neopixel is lit up green.
and I got the same with an itsy M0
except sometimes it does show up after switching different versions of circuitpython back and forth, and it gives me a hard fault into safe mode, even with an empty code.py
ionic elk
post em
idle owl
D5 != GPIO15. Accurate?
(LMK if you need more than that section of the schematic.)
It's on 14.
ionic elk
D5 appears connected to GPIO14, yes
slender iron
except sometimes it does show up after switching different versions of circuitpy...
I think Jeff was seeing this too
idle owl
D5 appears connected to GPIO14, yes
Thanks!
ionic elk
what board is this a problem for?
idle owl
what board is this a problem for?
ItsyBitsy RP2040.
There are a few errors in the board def.
Already sorted out the LED and the BUTTON pins.
ionic elk
and the board profile does indicate it's connected to GPIO15, yeah
idle owl
and the board profile does indicate it's connected to GPIO15, yeah
I can use digitalio to toggle the pin and use a multimeter to verify it, correct?
Trying to avoid soldering headers on this board.
ionic elk
That should work just fine I would think
idle owl
Ok thank you
manic glacierBOT
ionic elk
Nothing special about GPIO14 or 15 that I know of, DigitalIO should work on both
idle owl
15 has that silicon bug, or used to anyway, not sure if this rev of the chips fixed it. But that wasn't relevant to using it as a digital pin, it was a potential-on-reset type thing.
ionic elk
Oh hmm, I'll have to check out the errata, didn't see that in the datasheet
idle owl
Also I did not know that building CircuitPython checks to see if changes were made or not. And if not, it doesn't build.
idle owl
Oh hmm, I'll have to check out the errata, didn't see that in the datasheet
Not sure if it ended up in the datasheet.
ionic elk
yah you can use V=1 to visualize the specific files that get recompiled
idle owl
Ah nice
ionic elk
it won't recompile files you didn't change in most circumstances
Unless you remove or add an entire module, or change a translated sentence, in which case the QSTRs screw everything up, so it's better to just make clean and rebuild in those circumstances.
slender iron
1.12 merge passed unix tests!
idle owl
Toggle appears to be working!
ionic elk
@idle owl wait wat
idle owl
I built it with the correct pin in the board def, and the pin is working.
ionic elk
Oh, so the board def was wrong
idle owl
If I'm doing this right anyway.
Yes ๐
The board def was wrong in 3 places.
LED, BUTTON and D5.
ionic elk
I thought you meant the default was working and was like how could that happen
ionic elk
Nice! good debug
idle owl
Thanks!
ionic elk
I was all prepared to be embarrassed at somehow blatantly misinterpreting a schematic again after last time
idle owl
Aww. ๐ Nah, you were right.
ionic elk
I am glad to see that my brain is still holding together lol
idle owl
kattni dismissed jeplerโs stale review via 4d97680 1 minute ago sounds like it's on the edge of being rude. "I DISMISS YOUR STALE REVIEW, SIR.". ๐
Thanks for the review, @ionic elk!
idle owl
AFK for an hour. Back after.
manic glacierBOT
onyx hinge
@idle owl I think it was Billy Joel who said, "don't be afraid to try again / everyone goes stale / every now and then / you've done it / why can't someone else"
idle owl
@onyx hinge Some of the checks failed on the objtype PR.
onyx hinge
boo! I'll look.
253704 bytes used, -8 bytes free in flash firmware space out of 253696 bytes (247.75kB).
yow
idle owl
Ok, I'm unclear - what's the plan for a 6.2.x release? Waiting until everything is merged? I'm asking because aside from the example code not quite working right without the board def fix I merged, folks are going to run into issues with using the LED, pin D5 and the button on the Itsy RP2040 unless they're running the absolute latest.
Trying to decide how to handle this.
manic glacierBOT
ionic elk
I have a weird compilation issue that only happens on CI, not my local builds
/usr/bin/../lib/gcc/arm-none-eabi/10.2.1/../../../../arm-none-eabi/bin/ld: build-feather_stm32f405_express/supervisor/port.o:/home/runner/work/circuitpython/circuitpython/ports/stm/./common-hal/alarm/__init__.h:32: multiple definition of `alarm_sleep_memory_obj'; build-feather_stm32f405_express/main.o:/home/runner/work/circuitpython/circuitpython/ports/stm/./common-hal/alarm/__init__.h:32: first defined here
This seems almost like the CI compilation is not recognizing my compilation ifdef?
#ifndef MICROPY_INCLUDED_STM32_COMMON_HAL_ALARM__INIT__H
#define MICROPY_INCLUDED_STM32_COMMON_HAL_ALARM__INIT__H
it's definitely there though
onyx hinge
@ionic elk can you link to the full file content on github?
I have a gut feeling but I'd like to verify it before I open my mouth
ionic elk
yeah that
onyx hinge
the line that's written there is a definition of alarm_sleep_memory_obj, so each time the file is included it creates a definition. It's included from (at least) supervisor/port.c and main.c
adding extern turns it from a definition into a declaration
if you don't see it locally, it may be because you have an older gcc
ionic elk
right, but isn't this supposed to stop that from happening?
#ifndef MICROPY_INCLUDED_STM32_COMMON_HAL_ALARM__INIT__H
#define MICROPY_INCLUDED_STM32_COMMON_HAL_ALARM__INIT__H
not really. That prevents the text of __init__.h from being added twice if it's #included twice from the same top level file, but each top-level C file that includes it still sees its own copy of those lines.
river quest
Happy 8th Birthday MicroPython!
https://blog.adafruit.com/2021/04/29/happy-8th-birthday-micropython-micropython-micropython-and-adafruit-sponsoring-join-in/
ionic elk
@onyx hinge then, why doesn't this happen on the other ports? ESP32-S2, NRF and RP2040 also don't use extern, but they don't muck up the CI
onyx hinge
esp32s2 uses a different compiler version
ionic elk
that said, you're probably right, my GCC is still 9.2.1
onyx hinge
nrf uses extern in the header file as I suggested ports/nrf/common-hal/alarm/__init__.h:extern const alarm_sleep_memory_obj_t alarm_sleep_memory_obj;
ionic elk
oh right ofc re: esp32
Ok, I missed that edit. I will update my gcc and add extern. Thanks Jeff!
onyx hinge
Glad I could help
idle owl
Hooray, the latest S3 build of CircuitPython has my fixes and it works!
ionic elk
what's the best GCC to use right now, there isn't a bug on the latest or anything, right?
idle owl
(Which is to say the code I was running on the custom firmware works on an actual build.)
@onyx hinge You asked earlier about merging things into 6.2.x, but I was unclear on the answer. I need to add warnings to this guide I think. Until we have a release of some sort with this fix.
onyx hinge
@idle owl yeah sounded like dan was going to do a sweep of changes on "main" that we want in 6.2.1 and gather tham all. so until then, the fix won't be anywhere but the latest binaries, and those are potentially beset by other problems
idle owl
Oi ok.
I'll verify with Limor on how to handle the guide and add warnings and hopefully remember to go back and remove them once we have releases.
Your reference to "main" might have been less confusing if you did main. ๐
simple pulsar
I just put absolute latest on Pico and got a MpyError: Incompatible .mpy file. - is that because it's going to be a 7.x release?
idle owl
The mpy format changed with 7. You can download mpy-cross and make your own, or use the .py versions until there's a 7.0 release.
onyx hinge
With circup, use circup install --py for it to install the non-compiled files, it'll be the best option for awhile
@idle owl glad you figured out main, main main main or whatever it should be.
idle owl
<@!330227457296957440> glad you figured out main, **main** `main` _main_ or what...
Well I had to ask earlier. You explained it.
ionic elk
Oh super thank you Apple please do mark every part of the arm compiler as malware and force me to manually allow each one from system preferences
idle owl
Oh super thank you Apple please do mark every part of the arm compiler as malwar...
Oh, yeah. That.
ionic elk
that's absolutely what I wanted to spend my afternoon on
idle owl
that's absolutely what I wanted to spend my afternoon on
I mean.... who doesn't...... ๐
jaunty juniper
isn't there a recursive command line that does that ?
i know xattr -r -d com.apple.quarantine filename or is it a different issue ?
ionic elk
If you know a way around this or an article that discusses it, please lord tell me
I'm on approval number 12 now, this might never end
idle owl
Wait, give me a minute, my partner figured out something if I remember correctly.
ionic elk
I swear Apple degrades even more every year
idle owl
But she's in a meeting, so let me see if she responds and remembers.
simple pulsar
The mpy format changed with 7. You can download mpy-cross and make your own, or ...
thanks
ionic elk
@idle owl I think I actually got through them all
but I'd still def appreciate anything, it might pop up again
idle owl
She doesn't remember without googling. But there was some command line thing iirc. That forces permissions.
Worth at least remembering there's a workaround, and looking for it next time.
hollow token
Could just pipe data from a find
find ./ -type f | xargs xattr -d com.apple.quarantine
Just cd to the directory first though
jaunty juniper
xattr -r is recursive, I don't know if it solves that issue though
fossil gorge
isn't there a recursive command line that does that ?
i know `xattr -r -d com.ap...
Don't know if that's the one for this situation, but I use this EVERY time there's an app update on Homebrew. "Looks like you downloaded something from the BIG BAD INTERNETS! Are you SURE you want to run that program? Did the big bad wolf give it to you?"
Actually, I do xattr -rc <path> because nothing works better than a hammer sometimes!
jaunty juniper
I've been using an applescript droplet for downloaded folders of stuff too, because it complained when opening a php file in a code editor... like that makes sense... I rarely need it though https://gist.github.com/Neradoc/6bc4ddcbd492d7a65a2dddb2e151c3ac
ornate breach
could someone start the CI for my pr whenever you sweep past it ๐
just need to see if i fixed it.
analog bridge
Anyone available to merge #4685... its already approved.
blissful pollen
Anyone available to merge #4685... its already approved.
Done
manic glacierBOT
tulip sleet
@onyx hinge hi - do you understand the pre-commit uncrustify formatting checks? I thought that pre-commit would do the reformatting. Those checks pass locally, but fail on GitHub. There are definitely formatting changes to be made. uncrustify leave *.uncrustify files scattered around. I thought they were the unfixed versions, but they are the fixed ones.
onyx hinge
@tulip sleet no, my impression was the same as yours
@tulip sleet what's your uncrustify --version?
manic glacierBOT
onyx hinge
because codeformat.py invokes uncrustify โฆ --no-backup I think it's supposed to modify files in place, not put .uncrustify files alongside them.
tulip sleet
<@!329766224093249548> no, my impression was the same as yours
Uncrustify-0.69.0_f
onyx hinge
BUILDING.md says you need "version 0.71 (0.72 is also tested)"
tulip sleet
oh , bleah, 0.69 is what comes with ubuntu 20.04. I will find a PPA. THanks.
onyx hinge
yeah if you're on ubuntu check the ppa that the build uses
https://packages.debian.org/bullseye/uncrustify I built bullseye's uncrustify (0.72) which I think is the version in Arch as well
tulip sleet
tulip sleet
oh, I thought that was only for py-related stuff
onyx hinge
ยฏ_(ใ)_/ยฏ
tulip sleet
indeed it has updated uncrustify. thanks!
what's weird is that 0.69 did not obey --no-backup but didn't complain about it either
analog bridge
@tulip sleet no errors is due to pre-commit... it hides the logs
tulip sleet
aggh
analog bridge
if you run the uncrustify command directly, their will be a bunch of errors
tulip sleet
tnx
onyx hinge
I think codeformat isn't running on all the files we intend
for instance, it doesn't run uncrustify on ports/raspberrypi/common-hal/pulseio/PulseIn.c
hollow token
All those poor ducks, all bread and no crust
(sorry, could not help myself)
onyx hinge
there must be some context I am missing
tulip sleet
maybe we need to add the whole port directory
onyx hinge
I think pathlib.Path.match doesn't treat "**" like we need it to
tulip sleet
@onyx hinge is it a bug in pathlib, or we are using it wrong?
I tried running a cut-down version of codeformat.py just to see what it was matching, but got nothing (I was doing it wrong)
solar whale
@slender iron thanks!
solar whale
I looked at the fix -- wow! that was obscure... glad he figured it out.
@onyx hinge Thanks for fixing both the pixelbuf and the samd21 issues -- they make 7.0 much more useable!
solar whale
@idle owl what boards can the "flash_nuke.uf2" be used on?
idle owl
RP2040 boards, afaik.
solar whale
nice -- good to know. thanks.
ornate breach
The question centers around, will it work for most kinds of SPI/I2C EPD? Iโm getting some 1.54โ 152x152 epd by GoodDisplay and I want to know if they are compatible with or the same general brand as what you use in your designs
slender iron
please don't ping here directly. more folks can answer than just her
@ornate breach I think displays are sourced from good display. support depends on the chipset used on the display
and how the chip connects to the display
ornate breach
The ones Iโm getting have both SPI and I2C directly
So it has those pins broken out directly
seems similar to the 1.54" one stocked in the shop, but not the same pixel density
solar whale
@slender iron good point. I pinged because the question was related to a guide she just published but your point is well taken.
Oops. I see from the deleted messages that you may have not been referring to me . Still a good point!
onyx hinge
@tulip sleet I think specifically that the "**" syntax of git globs is not understood by the pathlib functions that deal with globs
>>> pathlib.Path("a/b/c/d").match("a/**/d")
False
the weird thing is that it says it does: https://docs.python.org/3/library/pathlib.html#pathlib.Path.glob
maybe only .glob does, but not .match
onyx hinge
maybe?
it can also look slightly like it DOES work if you use the wrong test
>>> pathlib.Path("a/b/c/d").match("**/d")
True
it turns out the match can start "part way down the filename", or something. It is not extensively documented, so it's hard for me to know exactly what it does
tulip sleet
you did the right thing, I was just surprised, since the doc appeared like it should work
onyx hinge
>>> pathlib.Path("a/b/c/d").match("d")
True
``` well, you can just omit the `**` if it's at the beginning...I appreciate you taking the time to make sure what I did was right.
tulip sleet
@onyx hinge https://bugs.python.org/issue29249
onyx hinge
hmmm my conversion to regular expressions might not be right...
since a/**/b needs to match a/b
.. right?
tulip sleet
@onyx hinge Yes, in bash, with ** turned on:
halbert@tuna:~$ mkdir -p dir1/dir2
halbert@tuna:~$ ls -d dir1/dir2
dir1/dir2
halbert@tuna:~$ ls -d dir1/**/dir2
dir1/dir2
halbert@tuna:~$
though I don't think the patterns we use depend on that
onyx hinge
they do, tools/**/*.py should match tools/codeformat.py
jaunty juniper
ok, so I was looking into the issue I had with Mu, and it seems that just sending a lot of things to the serial makes it stop receiving, or time out. Like something like this sending data to the board while it runs anything makes it impossible to ctrl-C afterwards. (With as low as 25 on M0 boards I tested)
with serial.Serial(serial_port) as channel:
channel.write(b"." * 1000)
the issue with Mu is because when you click in the Serial window it sends as many VT100 commands to move the cursor right or left as necessary
(whether you are in the REPL or the code is running)
lone sandal
crimson ferry
7.x libs: MpyError: Incompatible .mpy file. Please update all .mpy files. See http://adafru.it/mpy-update for more info. is this because the mp merge is ahead of my 7.0.0-... build?
or will the libraries get rebuilt at some point?
jaunty juniper
ah the libs are mpy-compiled with the latest tag
not the head of main
and the latest tag in 7.x is just like the start of 7.x
onyx hinge
For now use .py files. if you use circup, you can circup install --py.
jaunty juniper
I had to compile libs myself to test a neopixel on latest on a QT PY, I had Memory errors with the .py
manic glacierBOT
manic glacierBOT
lone axle
@still zephyr @mental nexus readthedocs and PyPi are both set up now on https://github.com/circuitpython/CircuitPython_DisplayIO_Annotation ๐
still zephyr
<@!607375730610798634> <@!366059101203202052> readthedocs and PyPi are both set ...
Thank you ๐
solar whale
Is it now necessary/recommended to recompile mpy-cross when ever pulling in the latest MP merges or is it updated fully as part of the pull?
solar whale
so a prebuilt version is part of the repo - not just the source.
tulip sleet
there is no executable in the repo, but we re-build mpy-cross on every PR build, and upload it to the link above
solar whale
OK - but I should rebuild it in my clone of the repo for build things locally.
make -C /mpy-cross
tulip sleet
no, you can just download from the S3 link above if you don't want to rebuild it. That would be the same as building locally.
if there's an mpy-cross in your clone, it's just left over from the last time you built it. It's not part of the pull
just like any board build is left over
we don't commit any mpy-cross executables
solar whale
ok -- got it - that makes sense -- executables are not usually checked in - my "workflow" uses a locally built verison -- especially when building boards with "frozen libs"
tulip sleet
you could build it locally or you could download the corresponding one from S3; your choice
solar whale
I asked because the "building CP " guide has you doing the make -C mpy-cross as the first step. If you pull in a modified mpy-cross you should rebuild it as well.
tulip sleet
I asked because the "building CP " guide has you doing the `make -C mpy-cross` a...
I can add a caveat to the guide about when you might need to re-build mpy-cross
solar whale
It has not changed in a long time -- and as I understand it, it will change a few times over the course of the MP updates. A note may save some confusion -- but probably not ๐
Do you recall offhand where the frozen libs are built (is it in one of the Makefiles?) I 'm not finding it and wanted to see what it used for a path to the mpy-cross.
tulip sleet
Do you recall offhand where the frozen libs are built (is it in one of the Makef...
Look in py/mkrules.mk and py/mkenv.mk. $(MPY_CROSS) is invoked which is set to be in the tree as MPY_CROSS = $(TOP)/mpy-cross/mpy-cross
solar whale
ah -- so it does use the version in your local repo
tulip sleet
oh, sure, you have to build it or something like the CPX build will fail
solar whale
yup -- and many others
tulip sleet
(and if you forget to build it and the CPX build fails, then you need to do a clean build or else you get strange errors)
solar whale
I have been doing make -C mpy-cross/ clean; make -C mpy-cross/ as well as make ... clean; make for any build...
the annoying case is if you build it with the old mpy-cross executable, it may build, but it won't load!
tulip sleet
ideally it should rebuild mpy-cross automatically, but it's hard for mpy-cross to know what's a significant change, so it doesn't. There are many changes that would force an mpy-cross rebuild unnecessarily.
solar whale
it's not a perfect system... I raised it only to hopefully avoid some confusion that is bound to occur going forward.
Once the MP merges are done, it probably won't change again for a long time.
tulip sleet
solar whale
Nice! Thanks!
Now, we'll see if anyone reads it ๐
I'll admit to not having reviewed the guide for a long time and recently found it had changed significantly (all improvements!) I now try to review it more frequently and when something breaks!
fossil gorge
Now, we'll see if anyone reads it ๐
Maybe! If they read it, we won't find out about it since they won't be asking questions about it! ๐
fossil gorge
Whenever anyone with access has a chance, this issue could be labeled as "Good first issue"
https://github.com/adafruit/Adafruit_CircuitPython_ProgressBar/issues/28
manic glacierBOT
manic glacierBOT
Thanks for spotting this. The "SAMD21" and "SAMD51" got reversed in error when the comments were reformatted at a certain commit. See:
https://github.com/adafruit/circuitpython/commit/5c24023240be492ddcf0cacec21318a182933391
manic glacierBOT
main furnace
Should I write an issue to go along with my pull request?
onyx hinge
@lone axle what was the source for the folder & file images in the screenshot-maker?
or, more to the point, I want to add the proper license declaration
I guess there's a .license file, it's just not spdx-style
lone axle
I guess there's a .license file, it's just not spdx-style
I don't think I ran cookiecutter when I created it. Just did some things manually.
I found those icons from https://publicdomainvectors.org/ with the idea being that they were permissively licensed.
lone axle
I believe it was these 2: https://publicdomainvectors.org/en/free-clipart/Generic-file-icon/67197.html and https://publicdomainvectors.org/en/free-clipart/Blue-folder/38883.html and then I made a few variations of the file one for different file types.
onyx hinge
thank you!
still zephyr
@idle owl Just an example, these two learning guide have been updated the full example code with the new modified examples using board.I2C(), however not the rest of the code. https://learn.adafruit.com/adafruit-veml7700/python-circuitpython and https://learn.adafruit.com/adafruit-mcp9808-precision-i2c-temperature-sensor-guide/python-circuitpython. I am unable to be in the meeting today, so I wanted to point that out. Let me know if there is something that I could do to help. thanks
idle owl
@still zephyr Thank you for letting me know. If @lone axle is up for it, he can work with you to update guides as you update the documentation examples. Foamyguy, are you interested? It's ok if you don't have the cycles, we'll go another route in that case.
lone axle
<@!607375730610798634> Thank you for letting me know. If <@!382939733107408897> ...
Yep I can work on making those changes in the Learn Guides. I'll get those two that are linked tonight. @still zephyr feel free to ping me with any others that come up and I can fix them as they come.
still zephyr
No problem, I will go back then and review as it was originally the plan, and the make a list, if that is ok?
oh i see
idle owl
Thanks @still zephyr and @lone axle!
still zephyr
@lone axle there is some that has been done already, and other that I will revise soon
Do you prefer a complete list or as they come?
lone axle
Do you prefer a complete list or as they come?
either way is fine I think. If you want to make a gist or some other public file and keep a list there I could keep checking that and work on them as you add them?
still zephyr
@lone axle Will do a Gist tonight and share the link here with a Ping,
Thanks both of you ๐
lone axle
perfect! thank you as well!
turbid radish
?serverinfo
digital shoreBOT
adafruit
onyx hinge
@idle owl hey there seem to be no adabot "circuitypython library report"s for the last two weeks or so.. any idea what's going on? https://adafruit-circuit-python.s3.amazonaws.com/index.html?prefix=adabot/bin/reports/
weeeeird, if I scroll waaaay down the page they're there
@tulip sleet is it possible this is something to do with how your directory lister works? like, is it paginated by s3 in a different order than it's displayed?
can we fix it by moving old files to a sub-folder?
and by "we" I mean someone with s3 knowledge & permissions ๐
idle owl
I was getting it from the source because I couldn't find it.
I don't know what's going on, I forgot to look into it on Friday.
tulip sleet
<@!329766224093249548> is it possible this is something to do with how your dire...
it's not my lister, but I did modify it, though not the adabot one. Could I clean out the old reports? They're in the notes. Is that OK?
it's just alphabetical, I think
idle owl
it's not my lister, but I did modify it, though not the adabot one. Could I clea...
How old? I think they're worth having for some period of time on S3...
jaunty juniper
oh there's 2 lists and I see no rime or reason why a thing is in one and not the other
onyx hinge
s3 lists the files by an arbitrary or alphabetical order, then the lister script puts them in timestamp order. If s3 requires two API calls to list all the files, it's treated as a whole second list. At least, that's the story that the second header mid-file is making me tell myself.
idle owl
delete 2020?
I guess that's ok? I mean, the thing is, some folks don't know about the CP Weekly notes. But I guess they probably don't know about this bucket either.
tulip sleet
I will check the lister. The lister script we got from somewhere else; it is kind of a mess
idle owl
Yeah I guess that's fine.
onyx hinge
surely the 2019 and older files can be removed
I don't know who will refer back to them later
jaunty juniper
ah it hits the page limit in the middle of listing circuipython files
onyx hinge
but .. they could also be moved to an archive folder or something
tulip sleet
I could put them in an OLD/ directory
idle owl
That would also work
tulip sleet
I'll do that for now, and we can discuss whether to keep them
onyx hinge
I guess there aren't a lot of 2019 files are there, must have already culled them once
thank you Dan!
idle owl
@tulip sleet That Mu feature where you don't need VID/PID is in, correct? So I do NOT need to add something to a guide about making a PR to Mu for your board?
tulip sleet
right
tulip sleet
it will work as long as there's a CDC connection. If there isn't, it will not work
idle owl
deletes
tulip sleet
I can also make it check for MSC only, but that's a rare case
idle owl
This was someone saying we should include something in the Add your Board to CP guide about PRing to Mu when you do.
We do not need to do that at this point.
slender iron
Should I write an issue to go along with my pull request?
No need if you already have a PR
tulip sleet
@idle owl @onyx hinge adabot reports now cleaned up. All but latest April and May are now in OLD/
onyx hinge
@tulip sleet thank you!
@slender iron are there more merge-related problems you know about? I think I got the ones I know about.
slender iron
not that I'm aware of. I'm not caught up on email though
onyx hinge
okay
slender iron
I'm hoping to have the first pass merge of 1.13 today
idle owl
<@330227457296957440> <@!426200547024961539> adabot reports now cleaned up. All ...
Cheers!
onyx hinge
<@&356864093652516868> hope you can join us for the discord meeting in about an hour. Add your notes to the notes document and then join us in the CircuitPython voice channel! https://docs.google.com/document/d/1z62nKbjNNv0j6uPWeUG6Co3aSCw4OQ3wl9VYlRIeTi8/edit?usp=sharing
still zephyr
@idle owl created this PR with a checklist to verify in the libraries, could you take a look and tell me if I am missing something, just to want to make double check thanks ๐
idle owl
<@!330227457296957440> created this PR with a checklist to verify in the librar...
I think it's good? I say "think" because I believe you're more on top of this than I am. ๐ You can always add to it if you find you missed something!
It has the things I can think of.
still zephyr
No problem, ๐ I would start veryfing and adding accordingly.
thorny jay
I don't know if it apply here, but when a library change something, there is the check to see if it break existing learn guide. If it is new, then it does not apply (well except if some learn guide could benefit from using that new library).
still zephyr
yes, this will apply, and Foamyguy will do the check after
I included in the cheklist to add to the list that we will have together
lone axle
I don't think there is any automated check for that but when we update libraries in a way that breaks the old API we check the learn guide repos to find instances that need to be updated.
turbid radish
lurking
sterile bronze
lurking
cobalt grail
idle owl
If I remember correctly, the original author does not agree. ๐
ionic elk
it's a holy war
slender iron
let people say it however they like
thorny jay
I say PNG...
turbid radish
๐ ๐
ionic elk
tabs vs spaces, emacs vs vim
turbid radish
No one can get out of vim
ionic elk
Jiraffe vs Giraffe
cobalt grail
I am a power user. I know how to quit VI.
turbid radish
:q! doesn't do what you think it does
ionic elk
I just know because git trapped me in it so many times that eventually I didn't have to google it anymore
cobalt grail
I just care about what :wq! does. ๐
If I don't want to save the file, I just reboot the computer,
modern wing
๐ผ
idle owl
I know!
thorny jay
Not in public, in PM.
trim elm
Oh, sorry, haven't been to one of these in a while
solar whale
OV7670 is very cool!
turbid radish
THANK YOU SO MUCH KATTNI!!
idle owl
You're entirely welcome, @turbid radish!
@mental nexus We want to know what you're up to, CircuitPython or otherwise! Doesn't have to be CP-specific.
mental nexus
Thanks Kattni. Will keep making brief updates here until I get other notice!
idle owl
Please do ๐
mental nexus
โ
slender iron
the TLF progress will be helpful for CP debugging
mental nexus
CGrover the coffee scale looks really cool.
errant grail
@mental nexus Thanks. It's on hold for a while to permit completion of other projects. Hope to wrap it up in a couple of weeks.
lone axle
@errant grail this might be of interest if you are looking for invisible buttons: https://github.com/adafruit/Adafruit_CircuitPython_DisplayIO_Layout/blob/main/adafruit_displayio_layout/widgets/control.py it's a Group subclass with click API built in.
turbid radish
๐
errant grail
<@!339922377746022402> this might be of interest if you are looking for invisibl...
Excellent! Thank you.
idle owl
@trim elm There's a bunch of reasons for the auto-reloads. What OS are you on?
thorny jay
That is the 4x + pixel doubling (there is a x5 too): https://twitter.com/DavidGlaude/status/1388993353191145472?s=20
turbid radish
Will be in the newsletter
idle owl
Ubuntu to be specific. I'm using screen for my serial console
Basically anytime there's write happening to the board, it will reload. Anytime there's local activity on your computer, not necessarily to the board, it can reload. I use MacOS, and screen, and I get it if I move files around in Finder that have nothing to do with the board.
What's the specific issue you're having?
trim elm
Hmm
Yeah, so basically I get auto reloads a lot. Seems to mostly happen in loops that are running without a time.sleep or with a really short one
idle owl
Yeah, so basically I get auto reloads a lot. Seems to mostly happen in loops tha...
Hmm. On what board? Or all boards?
trim elm
Pretty much every board. It was something I remember running into the first time I ever used CircuitPython in 2018 on a Feather M0, but I'm currently running into it on the Funhouse
idle owl
Pretty much every board. It was something I remember running into the first time...
Then I think you're running into what I was talking about. Where when there is local activity on your machine, it can trigger a reload. I mean, there are times where I'll have something last ages without a reload, but typically, they're triggered pretty regularly.
slender iron
@ionic elk it'd be nice if you could pick up https://github.com/adafruit/circuitpython/pull/3454 too. I think it's related to your main.c work
trim elm
Yeah, that's exactly it
idle owl
I think it's an issue if it's happening when the board is stand-alone, but while it's connected to your computer, it can happen.
trim elm
Yeah, it never happens when it's stand-alone
ionic elk
@slender iron that's the supervisor bug I got started on this past week ๐
idle owl
Then it's expected behavior. Unfortunately or otherwise ๐
ionic elk
I didn't get super far into it but I'm on it ๐
trim elm
I think at one point someone said that I had to run some command after copying files and that fixed it, but I do not remember what the command was
ionic elk
Oh no wait it isn't, I got mixed up. But I'll check it out, looks related?
trim elm
I think it had something to do with the files being copied and the computer doing something weird with the drive a bit after that
ionic elk
whoops
turbid radish
Windows generally doesn't touch the drive except writes. AV software perhaps or Mac trash etc
trim elm
Yeah, I don't think I've ever really used windows for programming, so I'm not sure if it happens there.
tulip sleet
Some third-party utility programs, anti-virus scanners, or innate OS indexing programs can touch the drive and do writes to it.
jaunty juniper
we've had a few issues with some type of Windows 10 anti virus scanner in #help-with-circuitpython
trim elm
Interesting
Yeah, I think it had to do with an OS indexing program, but the solution I was told was like 2 years ago and sadly I did not write it down
turbid radish
thorny jay
Yeah, so basically I get auto reloads a lot. Seems to mostly happen in loops tha...
I come from java. ๐
mental nexus
Pretty much every board. It was something I remember running into the first time...
One way to debug is to try this supervisor command to prevent auto reloading: https://circuitpython.readthedocs.io/en/latest/shared-bindings/supervisor/index.html#supervisor.disable_autoreload
To reload CircuitPython you will then need to Ctrl-C -Ctrl-D
trim elm
Ok. That'd definitely work, but I still like auto reload for when I copy new files
cobalt grail
CircuitPython_Org_Xxxxx?
thorny jay
Something that start with org ... is really java and could trigger nightmare for some that migrated to Python for a reason.
errant grail
Perhaps?
slender iron
this is where we try to compare feathers: https://learn.adafruit.com/adafruit-feather/feathers
idle owl
@lone axle You're not muted.
lone axle
Thank you
trim elm
Yeah, and it's also pretty small so the pixels look amazing
errant grail
The integral displays are usually on their own SPI or parallel ports. Helps with display performance.
thorny jay
240x240 is like on all of the other board: Gizmo + Clue + ...
modern wing
Thanks!
turbid radish
Thanks all
thorny jay
So that is easy to test on the presentation.
errant grail
Thanks!
mental nexus
Thanks all!
trim elm
Thanks!
lone axle
Thanks all ๐
idle owl
@onyx hinge I'm running next week. FYI.
modern wing
I write it down.....and I forget I wrote it down....
onyx hinge
Here is the notes document for Mondayโs CircuitPython Weekly meeting. It is at the normal time of 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if youโll be attending the meeting - itโs super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and weโll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1R2Qq9dAKqrAh0wFuu1-X7r9HgPJf1bgLz45fklJHmfo/edit?usp=sharing
slender iron
@tulip sleet what did you mean by https://github.com/adafruit/circuitpython/pull/4693/files/cd2c2f19f9eddd82aea8d47c38c29d90185d8b2f#r624277785 ?
tulip sleet
the "fix formatting"? That's a mistake on my part. The gc_make_long_lived() wrapped in the file comparator, and I thought it was on the next line.
slender iron
๐
tulip sleet
@slender iron did you see this PR from @onyx hinge to tannewt/circuitpython?
also there is a stray issue here: https://github.com/tannewt/circuitpython/issues/6
i was trying to compare your force-push old commit with the new commit to see the new changes-- haven't succeeded at doing that yet
slender iron
i was trying to compare your force-push old commit with the new commit to see th...
ya, sorry. I could just do a commit on top
tulip sleet
i figured out the changes -- github said "outdated file" and that was the clue
slender iron
that issue is very old
tulip sleet
yeah, i thought it was misplaced. i was just surprised to see issues and PR's in tannewt
it was only because I went to tannewt/circuitpython to do the compare
slender iron
I think I did the flag renumber already
slender iron
ok, pushed the last array fix
tulip sleet
ok, pushed the last array fix
ok, I will stop the previous build to save time
slender iron
I think it had finished already
tulip sleet
i clean up the jobs a lot when I'm impatient ๐
also ever since i found out it might cost $15 if we were paying ๐
slender iron
ya, I think it's a good idea if you think of it
thorny jay
Hi @onyx hinge you wanted to find the bilinear interpollation discussion... but that was in PM. So for the record, here is the fully working code that
@lapis hemlock shared with me (and I shared with @errant grail).
This was with a lot of discussion before and optimisation after.
b = array(range(n * n)).reshape((n, n))
a = zeros((2*n-1, 2*n-1))
a[::2,::2] = b
a[1::2,::2] = 0.5 * (b[:-1,:] + b[1:, :])
a[::2,1::2] = 0.5 * (b[:,1:] + b[:,:-1])
a[::2,1::2] = 0.5 * (b[:,1:] + b[:,:-1])
a[1::2,1::2] = 0.25 * (b[:-1,1:] + b[1:,:-1] + b[1:,1:] + b[:-1,:-1])```Jeff Thanks for hosting , your voice sounded different today ๐
still zephyr
@lone axle as discussed I will start putting the learning guides to be updated here:
Thanks ๐
lone axle
perfect Thank you ๐
solar whale
Does anyone know why MISO/MOSI/SCK are present on the same Pins as D5/SDA/SCL on the Feather RP2040? Will this preclude using I2C0 and SPI0 at the same time?
The conflict with D5 and SPI0 came up in a forum discussion https://forums.adafruit.com/viewtopic.php?f=57&t=178805
Ah - I see that SDA/SCL are I2C1 not I2C0 -- the Stemma is I2C0 -- this is confusing -- but it does conflict with the use of D5....
idle owl
@solar whale I provided the pin information multiple ways. There's so much information to convey with the RP2040. Complex is kind of the only option there.
slender iron
1.12 is merged!
tidal kiln
@solar whale i think those other designations relate to the RP2040 itself, since I2C/SPI can move around. but for the feather pins, there should be no conflicts. MOSI is on physical pin 30 and a D5 is on physical pin 9.
solar whale
But there was clearly a conflict with D5 and MOSI
and it says MOSI is on the MOSI pin and on D5
tidal kiln
where?
solar whale
but as in the forum post, D5 dis not work as the CS pin
you are correct, it is not in pins.c so I am very confused....
solar whale
right -- so D5 is not available as aGPIO if SPI is in use. That is not "normal" on a feather.
it seems like MOSI is on MOSI and D5
tidal kiln
does RP2040 SPI activity show up on all pins at once?
solar whale
I have not looked. ... yet. I just found that D5 would not work as the CS pin for the rfm69 under test.
tidal kiln
same should be true for any SPI peripheral then? (don't have rfm handy, but do have BME280)
solar whale
agreed -- if you have time try it.
tidal kiln
yah. issue seems odd. i'm curious.
hollow token
I did do this on the regular pico
You can use the GPIO on a pin as long as that pin itself isn't being used in SPI/I2C operations
solar whale
that is problematic and not the same as other feathers.
tidal kiln
Adafruit CircuitPython 6.2.0 on 2021-04-05; Adafruit Feather RP2040 with rp2040
>>> import board
>>> import digitalio
>>> cs = digitalio.DigitalInOut(board.D5)
>>> import adafruit_bme280
>>> bme = adafruit_bme280.Adafruit_BME280_SPI(board.SPI(), cs)
>>> bme.pressure
1009.7
>>>
i'm gonna grab an RFM and try that as well. in case it's something specific to that.
solar whale
hmm -- the FRAM works as well -- so false alarm -- now to concetrate on why the rfm does not work.
tidal kiln
Adafruit CircuitPython 6.2.0 on 2021-04-05; Adafruit Feather RP2040 with rp2040
>>> import board
>>> import digitalio
>>> import adafruit_rfm69
>>> cs = digitalio.DigitalInOut(board.D5)
>>> reset = digitalio.DigitalInOut(board.D6)
>>> rfm69 = adafruit_rfm69.RFM69(board.SPI(), cs, reset, 915.0)
>>>
now going to try per forum posters code...
solar whale
did 5/6 work for you?
tidal kiln
yep. see above.
solar whale
oh --yuk -- I have my feather plugged into a tft feather wing -- it has the conflicts!!!!
nevermind--- D5/D6 are fine ....sigh -- sorry about that
tidal kiln
Adafruit CircuitPython 6.2.0 on 2021-04-05; Adafruit Feather RP2040 with rp2040
>>> import board
>>> import busio
>>> import digitalio
>>> import adafruit_rfm69
>>> RADIO_FREQ_MHZ = 915.0
>>> CS = digitalio.DigitalInOut(board.D5)
>>> RESET = digitalio.DigitalInOut(board.D6)
>>> spi = busio.SPI(board.SCK, MOSI=board.MOSI, MISO=board.MISO)
>>> rfm69 = adafruit_rfm69.RFM69(spi, CS, RESET, RADIO_FREQ_MHZ)
>>>
also works
no worries. just nice to know why.
i see something. i'll respond in thread.
forum thread
solar whale
Thanks!
We now return to our usual programming ... ๐คฆ
@slender iron so far have tried the MP 1.12 merge on a feather_rp2040 and on an esp32s2 Funhouse. No problems other then my confusion above....
slender iron
great!
hollow token
I should get some of my build infrastructure making me these PRs to flash onto my pico. I love the features that will/might be in 7.x
solar whale
You can get the latest build here https://adafruit-circuit-python.s3.amazonaws.com/index.html?prefix=bin/raspberry_pi_pico/
hollow token
I know, I'd rather have local builds available to me
solar whale
Me too...
hollow token
I already have a local debian x86 and x64 package mirror and some PXE images for my VMs, plus I already use GitLab mirroring so putting in some patches to allow localised CICD runs shouldn't be too hard.
AKA, I have the technology :P
solar whale
Not sure what CICD is?
solar whale
I do local builds on Raspberry Pi(4) , Linux (Ubuntu) or MacOS. I just do the build following https://learn.adafruit.com/building-circuitpython not the full CI
hollow token
I'd prefer the end-to-end job, automated tasks from a repo should be fairly portable which aids in https://reproducible-builds.org/
fossil gorge
I write it down.....and I forget I wrote it down....
Same - or worse, WHERE I wrote it down ๐คฆ
fossil gorge
Continuous development continuous integration
I've often heard the "D" for "Delivery". Ship on every build type thing
*ahem* successful builds!
solar whale
@onyx hinge FYI - I built CP for the RP2040 pico and the GCM4 with the MP 1.12 merges and tried the OV7670 examples. It works OK on the RP2040, but fails (safe mode) on the GCM4. Even the OV7670_simpletest goes to safe mode on the GCM4... I'll do more testing tomorrow and open an issue if I can't resolve it but just wanted to let you know.
hollow token
I wonder if I can get an automated testing rig to test builds on actual devices for PRs.....
onyx hinge
@solar whale yeah there's something haunted on grand central. I just don't know what. Nothing's made sense yet. It's not even the camera capture code because that doesn't get run before the hardfault occurs. I don't even think it's "due to" code changes, for me this hardfault while setting up the camera has been coming and going the whole time I've worked on the project.
solar whale
Yes - I am having issues even getting previously working builds to run now....
Time for bed... ๐ค
onyx hinge
I'm sure someone will figure it out. Doesn't seem to have affected the aduino version...
ornate breach
I've often heard the "D" for "Delivery". Ship on every build type thing
Youโre right, I had it mixed up for some reason ๐
thorny jay
this is where we try to compare feathers: https://learn.adafruit.com/adafruit-fe...
Ok, that guide is not helping me, nor "Michael (Mike) Horne" that I try to help. The guide does not know about the Feather RP2040 no the FeatherS2 so it is lacking behind. I don't find yet @idle owl buying guide that was listing various board and benefits of each. One useful information for me would be the memory available once CP run (the output of import gc gc.collect() print(gc.mem_free()) once the board is booted.
And then the same thing with the full library set and big board specific library (like for PyPortal or FunHouse or ...)
Just like a C64 computer would display the available memory at boot prompt on top of the screen. ๐
onyx hinge
@lone axle will you have a chance to look at https://github.com/FoamyGuy/CircuitPython_Library_Screenshot_Maker/pull/2 today? I'd like to get this merged, then move the repo into the circuitpython organization so I can continue with the steps that I need to do in the Learn repo.
solar whale
I'm sure someone will figure it out. Doesn't seem to have affected the aduino v...
The speed of the Adruino version is impressive!
onyx hinge
The speed of the Adruino version is impressive!
yeah they have at least a couple of tricks we don't
but it's the not crashing trick I'd like to figure out
lone axle
<@!382939733107408897> will you have a chance to look at https://github.com/Foam...
Yep, I can take a look now. Meant to check out last night but slipped my mind.
onyx hinge
@lone axle I appreciate it, thank you
onyx hinge
@lone axle thanks for merging that. Now can you transfer it into the circuitpython organization?
lone axle
<@!382939733107408897> thanks for merging that. Now can you transfer it into th...
on it's way now. Github reports "may take a few minutes" but I do see it on the org page already
onyx hinge
great, that lets me do my next step.
lone axle
Sorry I didn't transfer back when I merged ๐. Thanks again for all of your help on it
idle owl
Is this a valid command? pylint Adafruit_Feather_RP2040/*/*.py Will it run Pylint on all files in all subfolders of the indicated directory? Or does it only think it does. (It passed running it that way, and passed running it specifically on the individual files, so I can't really tell if it ran properly.)
onyx hinge
@idle owl yeah. You could change "pylint" to "echo" and that will show you what files are listed
idle owl
Oh neat. Thank you.
slender iron
welp, I think I messed up the git history with my merges
still zephyr
mmmm @fossil gorge ?
fossil gorge
mmmm <@!361158808501747735> ?
NEver heard of them
still zephyr
jajaja
fossil gorge
Figment of a collective society's imagination! ๐
fossil gorge
Maybe, but I don't know... Scott's quite the power-user with Git.
still zephyr
worth the try ๐
at least we chat
@idle owl I will submit I think the last update to the design guide, that includes the code usage template, do you think that we are missing something?
idle owl
<@!330227457296957440> I will submit I think the last update to the design guide...
You already covered more than I would have thought of including. We can always add more if we decide there's something else that should be added.
still zephyr
Ok, no problem just making sure, when we started you said small updates, so hope not bothering you with that granular ๐
idle owl
Not at all!
still zephyr
I would like to cover requirements vs mock the docs vs sphinx intermapping but I do not think it is the right place
idle owl
Hmm yeah. Agreed it's worth covering, but perhaps not there.
still zephyr
We could create another document like how to, to do some libraries management
pypi, releases, readthedocs, requirements, thath kind of stuff
idle owl
I mean, there's a guide for that.
still zephyr
๐ฎ
idle owl
Some of it anyway.
This only covers part of that. https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library
still zephyr
could you point me in the right direction
ahhh yes, could I add stuff in there?
that is scott guide
idle owl
I don't see why not. I can add you as a contributor to give you access.
We've all contributed to it at this point. Scott authored the initial guide.
still zephyr
good thanks Kattni ๐
idle owl
Please create new pages, and then have us review them before they are made live, since the guide is already live.
Because once we click "Publish", the page is immediately visible to the public.
idle owl
@still zephyr Wait, have you done any editing in Learn yet?
still zephyr
Nope
idle owl
Ah! Ok. That makes things a little different. I'll DM you.
still zephyr
ok
idle owl
If anyone has a moment: https://github.com/adafruit/Adafruit_Learning_System_Guides/pull/1536 It's a quick one, it's template code, so there's nothing really to review.
blissful pollen
Um I thought I approved it but not letting it merge...
idle owl
@blissful pollen Is it not letting you actually approve the PR? Hmm.
Did you click the "approve" radio dial in the dialogue box?
nm, Dan got it. But I think you got it the second time.
blissful pollen
I thought I had...
tulip sleet
that is odd; I saw that too, so I went ahead and approved it too
blissful pollen
If it happens again I'll let someone know, never had that occur before. The comment even says "gamblor21 approved these..."
idle owl
Yeah I saw that.
But it also said "left a comment" so it seemed contradictory.
Yeah let us know if it happens again.
blissful pollen
"gamblor21 approved this but git doesn't trust him so have someone else look at it" ๐
jaunty juniper
the tooltip on the grey checkbox says "Only reviews by reviewers with write access count toward mergeability"
(as opposed to dan's green)
tulip sleet
yes, I was just making a screenshot
CircuitPython Librarians doesn't have write access to the Learn Guide repo, I guess
idle owl
That's odd, because I can add them as a reviewer.
I thought they could only be added if they had access.
Hmm ok. I guess I'll keep that in mind moving forward, or look into whether we want to add write access for that team.
tulip sleet
for the Learn Guide repo, Librarians only have read access
idle owl
I'll ask Limor if that's deliberate.
slender iron
my 1.11 and 1.12 merge commits only have one parent which makes them not merges
lapis hemlock
@onyx hinge Jeff, any side effects of hooking into the attr like this: https://github.com/micropython/micropython/pull/5648#issuecomment-830869436 ?
If would be really great, if ulab behaved in the same manner under micropython as in circuitpython.
onyx hinge
@lapis hemlock I don't know, is it the same or different than how we did it in circuitpython? i.e., is it "just" back-porting our code to MP?
oh I read a bit further down
that code link is not helping me at all, it's extremely long
lapis hemlock
<@!676859171315908639> I don't know, is it the same or different than how we did...
Right, it was a backport, but hasn't been accepted. Then somewhat suggested that we could exploit the .attr method.
onyx hinge
if micropython has its own way of doing properties, we'll probably be able to use it after the related code gets merged.
lapis hemlock
micropython doesn't really work with the concept of properties, that is the snag.
onyx hinge
py/obj.h: mp_attr_fun_t attr; we do seem to have this in our type objects, you're welcome to try it
lapis hemlock
If you search for mp_lv_color32_t_attr, there is only one place in the code, where it pops up.
lapis hemlock
`py/obj.h: mp_attr_fun_t attr;` we do seem to have this in our type objects, ...
That could be, but you wrote your own properties handler, via proxies.
onyx hinge
Well, not me personally
lapis hemlock
It doesn't matter, who it was, they had a reason.
Perhaps, the fundamental question is, what attr is meant for.
onyx hinge
@slender iron I dunno if you want to drop in and supply any history of why we coded our object properties like we did
lapis hemlock
I could definitely add the pointer to .attr, and circuitpython would still have the option to do it via proxies. I wouldn't really mind.
The way it is used in the code example is really reminiscent of the concept of the proxy.
onyx hinge
My impression of the "attr" code vs the style we adopted is that our style makes properties more "table driven" and probably reduces code size when you have many of them. Plus it naturally drives you to keep each setter separate from the others (they have to be different functions), making cleaner code. It would be interesting to compare the binary size of (say) 100 CP-style properties vs a micropython .attr that supports 100 attributes
lapis hemlock
Plus it naturally drives you to keep each setter separate from the others (they have to be different functions), making cleaner code
This is a good point, and should not be overlooked.
On the other hand, I could live with the notion of a beauty plaster for micropython.
onyx hinge
py/objgenerator.c: .attr = gen_attr, circuitpython has .attr so maybe you can try it out on both
lapis hemlock
OK, will do.
onyx hinge
or maybe you could implement a ".attr" that implements the behavior you want, but still based on the class's dict.... hmmm
๐ค
it would be nice if you can find a way to get properties in MP, that'll make it more like standard Python for your users no matter which implementation (MP vs CP) they're using
lapis hemlock
Oh, you mean like a universal attr function?
onyx hinge
yes
ionic elk
I have somehow made my Circuitpython drive into a read-only filesystem
will storage.remount("/", False) have that effect?
jaunty juniper
yes that's what it does (from boot.py)
ionic elk
Isn't the second parameter for readonly: bool = False? Or am I misreading these docs?
jaunty juniper
readonly from Circuitpython
ionic elk
how does turning readonly to false turn it on?
jaunty juniper
I guess that was the question: it is read only from where ?
idle owl
False makes it readonly by you. And read/write by CircuitPython, iirc.
ionic elk
that seems a little misleading
jaunty juniper
yeah it's readonly from the point of view of the code
idle owl
I learned this because I omitted the boolean the first time around and it was doing what I wanted it to, but I didn't know why.
jaunty juniper
and yeah there could be a more explicit way to express that, nobody ever remembers which is which ๐
(on first try)
ionic elk
I just don't like functions that make me have to erase my filesystem
especially when they don't say that upfront
jaunty juniper
did it do that ?
idle owl
Can't you remount True?
ionic elk
I mean it made the whole drive read only
jaunty juniper
you can boot in safe mode or erase boot.py from the REPL
idle owl
Oh, you didn't make it do it based on a trigger. I thought you could still remount from the REPL, but maybe I'm misremembering.
Oh right! That. Rename or delete boot.py
ionic elk
yeah there's probably some tricky way to do it
is there a command for deleting boot.py despite the drive being read only?
jaunty juniper
safe mode is easy except on pico
idle owl
jaunty juniper
it's not readonly from the code, that's the point
idle owl
I think it's included elsewhere as well, but I knew I put it in the Pico guide.
os.rename("boot.py", "boot1.py")``` or ``` import os
os.remove("boot.py")```ah, ok. Handy tricks.
Maybe a note to that effect could be added to storage.remount() docs
idle owl
I don't see any reason why the docs couldn't be updated.
ionic elk
"If the remounted drive is CIRCUITPY, this will make it read-only to the host computer. Use os.remove("boot.py") to regain access to the drive, or erase the filesystem"
idle owl
"If the remounted drive is CIRCUITPY, this will make it read-only to the host co...
PR? ๐
jaunty juniper
safe mode !
ionic elk
or safe mode. I don't mess with this stuff so I don't know the tricks, sorry! This might not be necessary I'm just griping lol
idle owl
I'll include it in the Storage page for CircuitPython Essentials when I template it, so it'll start going into board guides at least.
Though I realise fully that would not have helped you here.
Simply saying I'll get it into Learn more ubiquitously.
jaunty juniper
since it's mostly used to enable to write to the board, I think it could be interesting to have the parameter name reflect that instead of readonly = False, something like host_can_write = False or reverse it code_writeable = True and force to use the argument name so all examples are self documenting (I know, that is a breaking change ๐ฆ )
idle owl
Yeah, unfortunately, I think we need to fix this one in documentation.
If it was clearer what was made "readonly", I think it would be fine.
tulip sleet
@ionic elk is it read-only from the host's point of view or the code's?
manic glacierBOT
ionic elk
@tulip sleet the default for storage.remount(), when used on /, will make the CIRCUITPY drive read-only to the host's point of view, rendering the drive unusable until you take extra steps from the REPL. The docs say readonly (bool) โ True when the filesystem should be readonly to CircuitPython. which hints that it will do this, but I think it could spell it out more clearly, as well as offer ways to regain file access.
tulip sleet
so you were using a remount in boot.py, right? I just wanted to make sure it just wasn't a corrupted filesystem, which the host can make read-only for safety reasons.
ionic elk
but ehh. it's not that important.
tulip sleet
e.g. I have seen a filesystem become read-only to the host out from under me due to corruption
ionic elk
No, there was a remount from some example code, I don't think anything is wrong here from a design perspective, I just noted there could be a docs addition
tulip sleet
got it, thanks; i read through everything but may have missed that. a docs improvement would be great
ionic elk
@slender iron do you have some LED code for me?
Fixed the supervisor thing
or I can do #3454
jaunty juniper
does the learn guide need mode info on how to recover too ?
https://learn.adafruit.com/circuitpython-essentials/circuitpython-storage
idle owl
@jaunty juniper I'm working on templating all of the CP Essentials pages. I'll include it in the new template version. But yeah, we could probably add it to that page, even though that will eventually be deprecated.
jaunty juniper
(like even following that you might put the wrong pin name and get stuck)
slender iron
or I can do #3454
Ya, pick up the next code file stuff. I think it'd be good to do in 7.x
ionic elk
That's #3454 ya
though actually going over my todo list I should probably wrap up my sleep profiling before I get fuzzy on the details
slender iron
Oh, you mean like a universal attr function?
@onyx hinge this is exactly why CP does it the way it does. It uses a standard attr and the property object used by Python properties. custom native attrs is a recipe for weird non-standard behavior
idle owl
@lone axle Do you have a ton on your plate at the moment? Wondering whether you'd be interested in updating the Storage page in the CircuitPython Essentials guide to make what storage.remount("/", False) does clearer, and also include the few different steps discussed above (I can outline them again) available to remove boot.py and regain write access to your CIRCUITPY drive. No is an entirely acceptable answer here.
simple pulsar
<@!382939733107408897> Do you have a ton on your plate at the moment? Wondering ...
Is the kwarg form for second arg the preferred form for that? It gets a mention in https://github.com/adafruit/circuitpython/issues/286
idle owl
Is the kwarg form for second arg the preferred form for that? It gets a mention ...
Probably? It would possibly be clearer.
jaunty juniper
my vote is to always include it
it doesn't help much the first time, but it helps remember it once you know what it does
(hehe)
idle owl
Sounds good. I doubt we'll go back and update all project usages of it, but moving forward, we can indicate that it should be included.
lone axle
<@!382939733107408897> Do you have a ton on your plate at the moment? Wondering ...
I'd be up for doing that
idle owl
@lone axle Excellent. I believe the options for removing/editing boot.py once you've made CIRCUITPY readonly are safe mode (at least this was mentioned, I would suggest testing), import os os.rename("boot.py", "boot1.py")
or import os os.remove("boot.py")
Those are three separate things.
The second two are done from the REPL.
idle owl
I'd be up for doing that
The clarification needs to be made that storage.remount("/", readonly=False) makes CIRCUITPY NOT readonly to itself. Therein lies the confusion - what was being made readonly. Does that make sense?
lone axle
Yep definitely. I've had the same confusion myself when I first started working with that. Ended up just trying both to figure out which is which ๐
idle owl
The example in the Essentials guide uses switch.value to determine whether to remount, so you'll have to explain that switch.value is True or False depending on whether the button is pressed when CircuitPython starts up. THEN explain the readonly thing.
Apologies ahead of time, this page will be deprecated eventually, but it's a HUGE project that will take a while, so we should still update this page for now.
lone axle
No worries there
idle owl
As in.... templates have to be created and then something like 60 guides have to be updated. ๐
lone axle
I will probably start those new explanations on a new unpublished page so I can get feedback before it goes live. and then once good I can copy it into the real page.
idle owl
That is perfect. Name the page something ridiculous that wouldn't ever be used so it doesn't eat up a URL that might be needed.
No idea what might need it, but still.
Is it accurate to say that the code in boot.py is executed when CircuitPython starts up?
simple pulsar
The example in the Essentials guide uses `switch.value` to determine whether to ...
Might be worth a comment about left vs right for switch when it's a real one like on CPX/CPB. The example still uses board.D7 rather than board.SLIDE_SWITCH
idle owl
Yeah. Like I said, heading for eventual deprecation. To be replaced by a tailored page in each board guide with code specific to that particular board, and comments like that would be included.
@lone axle If you could include something to the effect of what kjw suggested above, it would be appreciated. Thanks!
idle owl
Is it accurate to say that the code in `boot.py` is executed when CircuitPython ...
@tulip sleet Is this valid wording?
Or do you have something better, even...
Also is boot.py used for anything other than this?
Commonly used *
I'm going to continue the usage of a switch/button/grounded pin in boot.py to remount the filesystem so folks can get CIRCUITPY to the proper state and THEN set it to readonly by the host machine deliberately.
tulip sleet
@idle owl boot.py is run when CircuitPython starts from a hard reset or a power on. It is not run on soft-reload, etc.
idle owl
Ok thank you