I know I don't write great copy but I hadn't even thought about your main point
#circuitpython-dev
onyx hinge
idle owl
Need to tag Justin on the PR too to make sure I didn't break any HTML.
That was my first thought upon finding it in /contributing ๐
onyx hinge
OK I won't merge it right away then
It is a kind of contributing that someone can do even if they're not ready to put themselves forward as a contributor to the core
idle owl
Oh, I think it's a perfect place for it, it simply needs to be clearer.
idle owl
@onyx hinge ^^ https://github.com/adafruit/circuitpython-org/pull/508
silver tapir
updates de ES translation again
strings keep coming from ulab, so every week that I check there is a couple of new ones to keep up.
ionic elk
@slender iron that F1 PR is probably going to be in the works for a little while, Dahanzimin is definitely trying hard to make things work but they're not super experienced with C. I had him create the PR so I could help cut out some of the RTC stuff that was giving him issues, but flash still needs a full rework.
onyx hinge
@silver tapir thank you for keeping on top of it. Many projects find that they have to actually have a step of the development called "string freeze", so that translators have a chance to catch up and make sure all the strings are translated when release time comes. I don't think we're to that point yet (and not all langauges have translators as dedicated as you anyhow), but maybe we should consider how to approach the issue.
silver tapir
Waiting for a freeze is fine. I'm just obsessing about 6.0 in 100% es. :).
slender iron
thanks for helping them @ionic elk
lone axle
Is it expected to be possible to update a displayio.Shape by calling set_boundary() with new values inside the while True: main loop?
onyx hinge
If the display is auto-update I would have assumed "yes" @lone axle
lone axle
WIDTH = 100
HEIGHT = 100
display = board.DISPLAY
shape = displayio.Shape(WIDTH, HEIGHT, mirror_x=False, mirror_y=False)
for y in range(HEIGHT):
shape.set_boundary(y, 0, (WIDTH - 1))
# Create a TileGrid to hold the bitmap
palette = displayio.Palette(2)
palette[0] = 0x0000FF
palette[1] = 0x00FFFF
tile_grid = displayio.TileGrid(shape, pixel_shader=palette)
group = displayio.Group()
group.append(tile_grid)
display.show(group)
for y in range(HEIGHT):
shape.set_boundary(y, 0, 30)
while True:
for y in range(HEIGHT):
shape.set_boundary(y, 0, 80)
time.sleep(1)
for y in range(HEIGHT):
shape.set_boundary(y, 0, 30)
time.sleep(1)
If I understand correctly this code should be changing the shape from 80 px wide to 30px wide back and forth every 1 second.
but on the screen I end up with the shape "stuck" at 80px wide.
if I add time.sleep(1) before the while True: loop then I will have the 30px one "stuck" showing on the screen instead of the 80px one.
I'm not sure that I understand it enough to say if it's warranted, I can create an issue if this behavior seems incorrect.
onyx hinge
It seems like you gathered good information, so I'd put it on an issue. either it's a bug, or given that we're used to displayio things updating it's a place the documentation would need to be refined. assuming it doesn't address this.
it looks like the "dirty area" handling that is in the core code of TileGrid is not in Shape. If the area is not marked "dirty" it won't be redrawn
low sentinel
Enabling actions on one's forked circuitpython repo on github is fantastic. The current workflow works in other accounts just fine.
This lets you iterate quickly and just push stuff off to your own branch on your own fork and github will email you with your results.
Probably I'm the last one to figure this out but hey. It's really nice to know my next PR will probably pass CI first try.
manic glacierBOT
manic glacierBOT
tulip sleet
@onyx hinge Could you leave a review that requests changes for #3203, to ensure it's addressed?
manic glacierBOT
idle wharf
I've been doing some research about the ESP32-S2 camera's capabilities (in gist here: https://gist.github.com/askpatrickw/0179d09e74d5f2a4347d5666ea937c4d ). The camera functionality is accessed via I2S, the FeatherS2 exposes I2S as DAC1 and DAC2 (pin23,pin24) assuming all that is correct, the next question I have is about I2S support in CircuitPython.
The only mentions of that I find are in relation to audio, specifically audiobusio where I2SOut is implemented.
That leads me to think, in order to use the I2S Camera functionality of the ESp32-S2 a new library would be needed.
Does that sound right or am I way off?
Espressif IDF Docs for I2S
https://docs.espressif.com/projects/esp-idf/en/latest/esp32s2/api-reference/peripherals/i2s.html
https://github.com/espressif/esp-idf/blob/625bd5eb1806809ff3cc010ee20d1f750aa778a1/components/soc/soc/esp32s2/include/soc/i2s_struct.h#L347
thorny jay
Yeah, that's me, the user that requested GamePad in Commander board.
manic glacierBOT
@WarriorOfWire Do you understand why the wall-clock time is not different even after your latest commit, ebc1373?
https://github.com/adafruit/circuitpython/actions:
solar whale
@slender iron there were recent changes to bus_device, I think -- would they be casuing this ```Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 6.0.0-alpha.2-24-ga6e048686 on 2020-07-25; Adafruit Metro M4 Airlift Lite with samd51j19
import ahtx0_simpletest
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "ahtx0_simpletest.py", line 6, in <module>
File "adafruit_ahtx0.py", line 68, in init
File "adafruit_ahtx0.py", line 78, in reset
File "adafruit_ahtx0.py", line 78, in reset
File "adafruit_bus_device/i2c_device.py", line 104, in write
TypeError: extra keyword arguments given
just running the ahtx0_simpletest
solar whale
tried another I2C device -- same error ```Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 6.0.0-alpha.2-24-ga6e048686 on 2020-07-25; Adafruit Metro M4 Airlift Lite with samd51j19
import msa301_simpletest
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "msa301_simpletest.py", line 8, in <module>
File "adafruit_msa301.py", line 221, in init
File "adafruit_register/i2c_bit.py", line 70, in set
File "adafruit_register/i2c_bit.py", line 70, in set
File "adafruit_bus_device/i2c_device.py", line 104, in write
TypeError: extra keyword arguments given
I'll open an issue
manic glacierBOT
tranquil sun
could someone help with what I think is a very easy question
I have three trinket m0s which work and run when plugged into my computer, but dont when using an independent power supply supplying 5V.
does circuit python only work when connected to a computer?
crimson ferry
CircuitPython should run just on power, unless there is somehow a dependency on the connection. It's possible (but unlikely) there's not enough power. But what does your code do, and how do you know it's not working?
tranquil sun
its switching the dot LED to white and toggling a digital pin at 22Hz when connected to the pc, when powered standalown, the dot is purple only and no blinking dout.
I thought power might be the issue, and tried 5V 500mA, USB charging cable, and a PI power supply
crimson ferry
What device is is the dout at 22Hz? Code errors are typically flashing LEDs https://learn.adafruit.com/welcome-to-circuitpython/troubleshooting. Might help if you can post your code.
crimson ferry
no real load to speak of, 500mA should be more than adequate
tranquil sun
yeah, read up the trouble shooting page.
crimson ferry
What versions of CircuitPython and libraries? If you post the code, I can try it here.
tranquil sun
no flashing as described line errors
purple flashing said to indicated ValueError, but my purple light doesnt flash
ill send main.py, one sec
example
I thought these imports being left in should be fine
ill check versions one sec
wow preloaded read me says 3.1.1 2018-11-02
crimson ferry
you'll want to update that https://circuitpython.org/board/trinket_m0/
crimson ferry
libraries too https://circuitpython.org/libraries
tranquil sun
rookie mistake, installing right away
ivory yew
What's up with chrome marking our uf2 files as dangerous? Has anyone reached out to the Chrome team?
tranquil sun
White Led now with exernal power! I will need a little more time to hook up my scope, but as it looks, promising
onyx hinge
@ivory yew is it happening again??
@ivory yew or is this about the message shown on circuitpython.org downloads pages? It happened briefly (as far as I knew) and then stopped .. but by that time we had put the warning up and nobody's taken it down yet.
crimson ferry
@tranquil sun I suspect there was something in the old (unused) keyboard code that was dependent on the connection
onyx hinge
Re #c14: It wasn't a browser update. Chrome contacts Google's Safe Browsing service to verify that a file being downloaded is safe. There was a bug in that service that caused this issue to happen, but it has now been fixed.
tranquil sun
Makes sense, just checked the pulses and I get the 22hz ๐
thanks for your help @crimson ferry
onyx hinge
@ivory yew so as far as I know, if circuitpython.org is still showing that block warning you of chrome warning (so confuse) it would be great to PR its removal now
ivory yew
Ah yeah. We should take it down. I might have some time to send a PR later but no promises. ๐
onyx hinge
I don't see it / think about it due to not using chrome ๐
anyway the other discussion of it seems to have happened internally to adafruit so you wouldn't have seen it
not intentional, it just happened that way
tulip sleet
@onyx hinge Thanks for the BusDevice PR's So is PR #54 obsolete now that you filed #55?
onyx hinge
@tulip sleet yes if there's consensus to just drop compatibility
Feel free to close it, I'm on mobile right now
tulip sleet
np, I'll merge and make a release so it gets into the bundle. Thanks for this, since a lot of people were going to hit it.
onyx hinge
I think I scooped @crimson ferry who wanted to give a first timer a chance to work on it. Thanks for your offer to help out! It would have been fine to dive in due to the impact on users of this problem while it was broken. That said, I have read that there is huge benefit to actually reserving "good first issues" for new contributors and I should be mindful of that in the future.
tulip sleet
"Good first issue" was certainly appropriate, but it also needed to be fixed quickly since so many things were broken otherwise. So I think it's fine. The two libraries that need fixing could be "good first issue" also.
onyx hinge
OK I looked a little more cloesly at the other issues I filed, refined the initial comment, and marked them "good first issue". https://github.com/adafruit/Adafruit_CircuitPython_MLX90393/issues/18 is the one that seems to have the highest odds of just being broken with 6.x. I don't think I have an MLX90393 for testing.
crimson ferry
@onyx hinge Thanks for digging in and fixing it. It was marked "good first issue" and I didn't know the protocol / priorities, but it was affecting a wide swath of devices, so it's nice to have everyone back on track.
timber mango
Hi guys, what is the cheapest circuitpython capable board with BLE?
onyx hinge
@timber mango looks like among adafruit boards it'd be https://www.adafruit.com/product/4481
among other manufacturers https://www.nordicsemi.com/About-us/BuyOnline?search_token=nRF52840DONGLE and https://makerdiary.com/collections/frontpage/products/nrf52840-mdk-usb-dongle may come in a little lower, but I think they lack the (Q)SPI flash which means the CIRCUITPY drive is smaller
timber mango
Thanks, Jeff. I know Adafruit so I'll go with that ๐
woeful siren
What is circuitpython
mental nexus
CircuitPython is an open source derivative of the MicroPython programming language targeted towards the student and beginner. Development of CircuitPython is supported by Adafruit Industries.
woeful siren
@mental nexus can i start with CircuitPython if I have low notebook? ๐
CPU celeron
GPU Intel graphic card
RAM 4gb
lone axle
@woeful siren you don't need a particularly powerful computer to use Circuit Python but you would need to get a Circuit Python microcontroller device, or a Raspberry Pi.
lone axle
@woeful siren do you have any programming experience at all or is this your first time? Also do you have a specific project in mind or just wanting to learn?
lone axle
I think the Circuit Playground Bluefruit: https://www.adafruit.com/product/4333 is a really good 1st device to learn with. It has plenty of features built in to the board to play with while you learn. And when you're ready you can start connecting things to it with alligator clips.
It has BLE connection if that is of interest to you, but you can also ignore the BLE if you don't want to use it for anything.
woeful siren
Interesting thx
crimson ferry
^^ <@&327289013561982976>
slender iron
thanks anecdata
crimson ferry
I didn't see this in alpha-1: pystack exhausted doing a Request with 6.0.0-alpha.2. It's a big app... pushing limits of 256k on PyPortal (still have >60KB free mem before the error)
But this is right after setup, during first HTTP transaction, before any displayio shows
"Let us know if you hit a RuntimeError from an exhausted pystack." ๐
crimson ferry
Is there a way to programmatically (code.py) capture the output of help("modules")? It seems to just do its thing and return None.
solar whale
also in CPython -- nothing is returned x = help("modules") does not return anything to x
@crimson ferry FYI re:pystack https://github.com/adafruit/circuitpython/issues/3208
tulip sleet
@crimson ferry @solar whale https://stackoverflow.com/questions/11265603/how-do-i-export-the-output-of-pythons-built-in-help-function none of which would work on CircuitPython. help() just prints to stdout. it's not based on pydoc, as in CPython
crimson ferry
I was hoping to iterate over the installed modules on any board, it seems to be more definitive than the list in the matrix or circuitpython.org.
Thanks for the references
@solar whale the new network stack may help, bus I suspect we need more pystack in any case
with heap growing from one end and stack growing for the other, where is pystack?
Would it be difficult or undesirable to expose some coarse memory controls to user code? Users have good control of flash, but not memory. It would be useful if user code (maybe boot.py or precursor) could tweak heap vs stack vs pystack.
tulip sleet
@crimson ferry you can set the stack size: https://circuitpython.readthedocs.io/en/latest/shared-bindings/supervisor/index.html?highlight=stack#supervisor.set_next_stack_limit
the size of pystack was set experimentally in the alpha, and we can grow it
onyx hinge
@tulip sleet Does set_next_stack_limit set the C stack or pystack size though?
tulip sleet
it sets the C stack. Using pystack at all is brand new. previously the python stack was on the c stack
onyx hinge
Thanks, confirming what I thought. So it's not a knob that anecdata can turn now to fix this problem..
silver tapir
Here's a nice new feather esp32-s2, this time from Electronics Cats from Mexico.
https://electroniccats.com/store/bast-wifi/
tulip sleet
you can do a new build and modify the pystack size experimentally
tulip sleet
@crimson ferry CPython has sys.builtin_module_names, which we don't implement, but it might be handy. help("modules") builds a nice list internally, but then it just prints it and doesn't make it available programmatically.
you could suggest this in an issue
crimson ferry
Thanks. I may put in an issue. Part of it is the inconsistency between the textual sources of what modules are on a board. help("modules") is where the rubber meets the road. Fairly regular questions in #help-with-circuitpython on what modules are available.
tulip sleet
in the meantime, you can try stuff like:
possible_modules = ('analogio', ...)
have_modules = []
for m in possible_modules:
try:
__import__(m)
have_modules.append(m)
except ImportError:
pass
ah, from a master list... that could work
e.g., I thought I remembered gc being in readthedocs, but I don't see it (and other standard CPython modules) now, and gc.mem_alloc() seems to be upython-specific http://docs.micropython.org/en/latest/library/gc.html#module-gc
(but it is in our builds)
I guess it's not important to replicate CPython docs where we are a strict subset, More useful where CircuitPython has extensions, often inherited from MicroPython.
solar whale
@tulip sleet are there plans to increase pystack soon? we can discuss at meeting today.
where is pystack set?
crimson ferry
Would it be correct to say that memory allocated to pystack isn't available to [c]stack or heap, like it's a segregated chunk of memory?
onyx hinge
That's right, to my understanding
solar whale
@onyx hinge Do you know where pystack is configured? I was going to try changing it, but I can't find it?
onyx hinge
I was just looking for that -- I'd search around for MICROPY_ENABLE_PYSTACK, see commit 1160635608ebeaf3868a327803c0ca248f549123
+#define CIRCUITPY_PYSTACK_SIZE 1024
+#endif
+static size_t PLACE_IN_DTCM_BSS(_pystack)[CIRCUITPY_PYSTACK_SIZE / sizeof(size_t)];
+#endif
``` the way it is coded currently, it is set at compile time.Thanks -- I can play with that.
crimson ferry
Backgrounder: https://github.com/adafruit/circuitpython/pull/3160
solar whale
ah -- thats the ticket!!
py/circuitpy_mpconfig.h
I love how intuitive this all is ๐
crimson ferry
i get so lost in the layers of abstraction in the circuitpython core, hard (for me) to find the code that actually does the work. Maybe the abstraction is the work ๐
solar whale
same here
doubling PYSTACK to 2048 works on the metro_m4_airlift -- my requests to adafruit_io work now
crimson ferry
I was afraid pystack was fragmenting memory in some big way, but 2KB is nothing on M4 or better boards
manic glacierBOT
solar whale
Nice to be "back in business"
slender iron
I guess it's not important to replicate CPython docs where we are a strict subset, More useful where CircuitPython has extensions, often inherited from MicroPython.
@crimson ferry I'd like to have docs for subset modules but we inherit most of them from micropython and we haven't restructured them into shared-bindings
Would it be correct to say that memory allocated to pystack isn't available to [c]stack or heap, like it's a segregated chunk of memory?
@crimson ferry Correct. We probably want to reduce the default C stack size as we increase the pystack too. otherwise we'll shrink the heap
though strinking a bit is ok because some allocations that hit the heap now are done on the pystack
crimson ferry
thanks @slender iron , I realize these are tricky engineering trade-offs that may manifest differently in different use cases. Maybe the bottom line for users is to get to know the main constraints so they can make their own trade-offs or optimizations in their application
simple pulsar
I've not measured it yet but voice 0 appears to be louder than the rest for an audiomixer. Is this by design? Did @onyx hinge create this originally?
onyx hinge
No, I'm at least the third person to work on it. You can be the fourth ๐ ๐
simple pulsar
I'm just using 1-13 ๐
onyx hinge
the first voice that is mixed in does go through different code than the others, because it does not have to sum with a voice whose sample value has already been calculated. It's an optimization and it could well have a bug.
simple pulsar
Someone else found it too: https://github.com/adafruit/Adafruit_Learning_System_Guides/blob/master/Lucio_Blaster_2020/lucio_blaster_2020.py#L112-L113
onyx hinge
@simple pulsar in shared-module/audiomixer/Mixer.c, function mix_down_one_voice there are the two blocks of code depending on a bool ... ``` // First active voice gets copied over verbatim.
if (!voices_active) {
onyx hinge
<@&356864093652516868> The weekly meeting is in about 1 hour (2PM ET). Please add your notes to the document, and if you're going to be lurking note it there (2 places) -- it's super helpful! https://docs.google.com/document/d/1K-himsiOHsJozTNUktOK8SEmiBcRhlDkBW363jlQ2DQ/edit?usp=sharing See you in the CircuitPython audio channel at 2PM ET
ionic elk
@slender iron how would you feel about me implementing the flash module on the F1 once ESP32-S2 Pulseio is done? I think everything else in the F1 port I can guide Dahanzimin though, but flash is tricky. I'm concerned it'll stall out the port.
modern wing
Good afternoon all you wonderful folks! I'll be text-only/lurking today, and I've updated the notes doc to reflect as such. 
uncut nexus
yes
lone axle
Thank you
trim elm
lurking
turbid radish
lurking
slender iron
brb but you can start without me
turbid radish
Any Python/CircuitPython news, send soon to anneb@river quest.com to make this week's issue
tidal kiln
lurking
turbid radish
2 4K monitors rocks
gilded cradle
I'm on a 4K monitor and my 15" MacBook screen
stuck elbow
lurking
modern wing
Currently using a MBP 16" + 2x 1920x1080 + 2560x1080 --- that's my remote setup. All second-hand panels.
Primary setup at home is a MBP 16" + 2x 1920x1200 + 2560x1600 + iPad 12.9 via sidecar.
Would you believe I want/need more? ๐
slender iron
back
solar whale
Wow -- if you turn your volume off -- you can't hear anything!
thorny jay
Lurking. Unhappy about my contribution to the doc... I could not remember what I CP did this week!
slender iron
I am back if you want to kick to me
silver tapir
lurking
ionic elk
๐
idle owl
Same.
raven canopy
๐
ionic elk
wat
idle owl
I banned.
blissful pollen
fastest ban clicking on discord
modern wing
Thank you @idle owl.
ionic elk
was that just some drop-in rando?
turbid radish
Yes, random crashers
slender iron
ok, if anyone is not a circuitpythonista, then you won't be able to speak
let me know and I'll add you
uncut nexus
Can you add me? Thanks!
mental nexus
Sorry mic issues..
. Will try again at weekly reports.
blissful pollen
I do not think I am, you can add me though i'm not likely to talk anytime soon ๐
ionic elk
Is there a particular reason we're getting targeted more recently? Seems like it's been a big uptick
slender iron
we suspect it's the server directory which publicly lists us
ionic elk
Ahh that'd do it, if we're ranking highly
slender iron
I think we're on the education page...
ionic elk
I see us as number 8 on public servers home, but that might just be me
definitely seems like something Discord itself should be tackling
slender iron
yup. I'll check in with the python discord folks too
them and home assistant are probably having the same issue
blissful pollen
the Travis/git actions was confusing as someone new. I was able to see/work around it, but as someone still new if you want someone to test updated instructions let me know
ionic elk
Github in general is a heck of a learning curve for beginners so anything in this vein is great
gilded cradle
Sometimes I'll stare at the GitHub screen for a couple of minutes trying to figure out where they hid stuff.
Usually it's closer to 30 seconds though.
modern wing
oooo, +1 for the fpga board
thorny jay
I love the cat cam, but maybe a good CP project for @slender iron would be a "cat epilepsy detector" (or at least record on SD card normal cat vibration and abnormal cat vibration to feed an AI for detection).
slender iron
I've thought about that. it could be a hiss detector
though that would include false positives
turbid radish
I think someone on Twitter awhile back was working on a microcontroller pet epilepsy detector
crimson ferry
we should figure out some tech that would need a library called eieio ๐๐
ionic elk
what's the new top secret board? I have youtube blocked on my devices
thorny jay
@tulip sleet Do we need to acquire something to test HCI? Or just migrate a AirLift into a BluefruitLift by changing the firmware? I want to be ready to play with this.
tulip sleet
@thorny jay There will be updated ESP32 firmware. It will still do wifi also: the changes are just pin changes
I am using a regular Metro M4 AIrlift Lite, which needed ne firmware, but the breakout may not need new firmware
@ionic elk BLM badge
lone axle
I think this is the device from top secret:
slender iron
@ionic elk neopixel with the RMT on esp would be cool
ionic elk
@slender iron interesting
slender iron
thorny jay
@tulip sleet I have a PyBoard with AirLift build in, and the AirLift Feather. I might acquire a breakout of some kind.
ionic elk
@slender iron I'd probably want a shared RMT mediator in /peripherals then?
so the two modules don't conflict?
slender iron
ya if it's needed for other things
tulip sleet
@thorny jay either one you have now would be fine
onyx hinge
@ionic elk mute please?
crimson ferry
I think we want to keep the bar low for entry, but could we add some friction like needing an avatar or some other profile richness?
ionic elk
@onyx hinge shoot I'm sorry
slender iron
server discovery
tulip sleet
Unfortunately, plenty of the terrible people have an avatar or even an existing phone-verified account. We enabled 10-minute waiting before posting after joining, but people with phone-verified accounts apparently can bypass that.
slender iron
from "explore public servers" on the left hand side
thorny jay
I never see SPAM... but maybe because we have good moderator. So it is hard (except for what I have ear today) to figure out what it is. Does it take place in this "room"?
lone axle
It tends to only last for a few seconds before getting removed.
I have only recently actually seen it occur, because the mods are so on top of it that it's typically gone before anyone has a chance to see it. In the case I saw it was a spammed link of a photo.
raven canopy
PyDis' channel permissions, and the 'acceptance' prob blocks a lot of the spam. (as Scott just mentioned)
ionic elk
When other servers have a "introductions" channel, I wonder how much that can curb spam content?
Force all messages in one place until they get approved, or something like that?
blissful pollen
A friend's server has an intro channel and you cannot do anything else until you use a reaction to the intro message. So you have to at least read one message to know to do that
inland tusk
How about bannig UNICODE.
silver tapir
Is it possible to pause discovery while there is a response from discord?
ionic elk
@inland tusk what does banning unicode do?
silver tapir
And also, it seem like help is needed during this change. I'll be happy to moderate.
lone axle
Prevent "weird" characters from being used to get around automated bad-word filters.
ionic elk
ah
inland tusk
Okay I forgot about that.
modern wing
And to put this in perspective -- it's roughly 1% causing trouble, out of 23k members. It truly is a small amount, but unfortunately its a vocal minority.
idle owl
@silver tapir I will DM you later to discuss it.
silver tapir
Thanks.
ionic elk
I mean clearly some of this isn't just bots
Like these people screaming in the meeting, etc
Might want some human hurdles too
slender iron
I'm suspicious about that though because so many came at the same time
blissful pollen
Sadly there are discord servers dedicated to disrupting other servers
ionic elk
I feel like escalation of permissions is a good fit for human bad actors
solar whale
@slender iron is the correct way to set the pystack up and default stack down? ```
diff --git a/ports/atmel-samd/mpconfigport.h b/ports/atmel-samd/mpconfigport.h
index 540a2f77c..e25568164 100644
--- a/ports/atmel-samd/mpconfigport.h
+++ b/ports/atmel-samd/mpconfigport.h
@@ -109,7 +109,7 @@
#endif
#ifndef CIRCUITPY_DEFAULT_STACK_SIZE
-#define CIRCUITPY_DEFAULT_STACK_SIZE 4096
+#define CIRCUITPY_DEFAULT_STACK_SIZE 3072
#endif
#ifndef SAMD21_BOD33_LEVEL
@@ -134,7 +134,7 @@
#endif
#ifndef CIRCUITPY_DEFAULT_STACK_SIZE
-#define CIRCUITPY_DEFAULT_STACK_SIZE (241024)
+#define CIRCUITPY_DEFAULT_STACK_SIZE (231024)
#endif
#ifndef SAMD5x_E5x_BOD33_LEVEL
diff --git a/py/circuitpy_mpconfig.h b/py/circuitpy_mpconfig.h
index a1a7ec238..768e05af1 100644
--- a/py/circuitpy_mpconfig.h
+++ b/py/circuitpy_mpconfig.h
@@ -799,7 +799,7 @@ void supervisor_run_background_tasks_if_tick(void);
#endif
#ifndef CIRCUITPY_PYSTACK_SIZE
-#define CIRCUITPY_PYSTACK_SIZE 1024
+#define CIRCUITPY_PYSTACK_SIZE 2048
#endif
#define CIRCUITPY_BOOT_OUTPUT_FILE "/boot_out.txt"
ya, I believe so
modern wing
And to everyone here -- thank you for being awesome. You're upstanding members of the community -- a direct opposite of this issue we're facing. It's refreshing and a pleasure to have y'all here.
solar whale
OK -- works on M4 airlift -- will test CPX and then do other ports as well
idle wharf
There was a link posted on the weekend and before I could type (at) mod it was gone. Great job Mods!
idle owl
I need to remain muted beginning at 15:30. I'll stick around, but can't respond.
lapis hemlock
I've got to log off now. Good bye, everyone!
idle owl
Thanks everyone!
modern wing
Thanks everyone! Thanks for running the meeting @onyx hinge!
silver tapir
Thanks all.
idle owl
I had ๐
ionic elk
so meta
modern wing
Excellent job indeed ๐
thorny jay
Thank you @onyx hinge . If someone can prepare the next doc, that permit to update it during the week.
solar whale
@tulip sleet Is there a specific test you want me to do on the CPX with the stack changes?
tulip sleet
just importing the module blew out the stack before we grew it
it was refactored to use a superclass, and that and a fairly deep sensor initialization got too deep on the stack
solar whale
ok -- jsut the standad CPX or should I test the crikit and displayio versions as well
tulip sleet
i think they all have the same stack size, let me check
solar whale
all SAMD21 are the same
tulip sleet
circuitplayground_express_displayio/mpconfigboard.h
31:#define CIRCUITPY_DEFAULT_STACK_SIZE (4760) // divisible by 8
circuitplayground_express/mpconfigboard.h
31:#define CIRCUITPY_DEFAULT_STACK_SIZE (4760) //divisible by 8
circuitplayground_express_crickit/mpconfigboard.h
33:#define CIRCUITPY_DEFAULT_STACK_SIZE (4760) // divisible by 8
in ports/atmel-samd/boards/
solar whale
ugh -- they all need to be changed -- I only looked at the mpconfigport.h in atmel-samd
I forgot to check the individual boards.
will do that
tulip sleet
those are the only boards with a custom value, though
solar whale
whew!
tulip sleet
(grep told me)
solar whale
so they had bigger stacks to start... I'll experiment
gilded cradle
Dropping off
tulip sleet
@solar whale try something less, like 256 or 512. We didn't increase the stack a lot to begin with
solar whale
ok -- will try other values.
low sentinel
@lone axle re: https://github.com/adafruit/circuitpython/issues/3204
The issue here is that the winding number algorithm is kind of "handed." It will miss certain pixels in certain cases, and the rectangle is kind of the worst-case shape for it.
I spent a lot of time over the weekend looking for a solution and haven't landed on one I'm happy with yet.
The code on your issue should produce a 51x51 square, not 50x50. I'd welcome you (or someone else) to take a look and see about an improvement to account for line edges a little better.
onyx hinge
winding rules: https://en.wikipedia.org/wiki/Evenโodd_rule
solar whale
@tulip sleet ok -- reducing by 512 seems to work on CPX -- I'll do some more testing and probalby put in a PR tomorrow
tulip sleet
thank you!
onyx hinge
Belated reminder, we'd love your submissions to the newsletter! You can make pull requests, and the "development" of the newsletter is done in the open: https://github.com/adafruit/circuitpython-weekly-newsletter/blob/gh-pages/_drafts/2020-07-28-draft.md
slender iron
lone axle
@low sentinel whoops, you are right it should be 51x51. So it's not that it is shifted by one, but rather that it didn't fill in the first column of pixels. Neat, thank you, I was unfamiliar with this algorithm. If it were possible to change the size of a vectorio.Rectangle I would probably opt to use that instead of vectorio.Polygon for most of the use cases that I have looked at so far. Perhaps it's worth revisiting that idea if the rectangle is particularly tricky for this algorithm. I can try to look into a bit, but C is mostly outside of my skillset.
solar whale
@tulip sleet nice! the metro_m4_airlift still can use requests and access adafruit_io with only a 512 byte pystack increase
tulip sleet
ok! the 1024 was picked because one micropython port had picked that value for pystack: we didn't have a lot of data to go on
solar whale
m4 was set for a STACK of 24*1024 so I am also reducing it by 512 bytes (to 24064)
idle owl
@lapis hemlock You're still in one of the voice channels. I recommend disconnecting as someone could drop in and start talking (or screaming) unexpectedly.
onyx hinge
Here is the notes document for next Mondayโs CircuitPython Weekly meeting, August 3 2020. 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/1i_l-w0lmHZg7vSoNgAvNWDoVwK1jU5dojDuJvzQgvMQ/edit?usp=sharing
idle owl
Cheers, @onyx hinge Again, great job today!
onyx hinge
You're welcome.
We'll need to revise the opening to advise that you now MUST be a circuitpythonista to speak in the meeting.
idle owl
Accurate. I'll make a note to update the guide.
onyx hinge
This meeting typically happens on Mondays at 2PM Eastern Time, but the date occasionally varies for US holidays. We have an online calendar. If you would like to speak in the meetings,, let us know and we will add you to the โCircuitpythonistasโ role. This will also get you a small number of notifications from Discord during the week, mostly reminders about the meeting. The length of the meeting varies, but itโs often from 60 to 90 minutes long.
Here's what I wrote in my meeting intro script, does that make sense?
lone axle
Looks good to me fwiw
idle owl
Mostly makes sense. I might make it more explicitly clear that it's required to speak. I'll let you know when I update the guide, and you can adapt whatever I end up adding to fit your wording.
tulip sleet
@lapis hemlock I disconnected you from the voice chat
lapis hemlock
@idle owl @tulip sleet Thanks, I logged off in a hurry. ๐ณ
idle owl
@lapis hemlock No worries! Simply looking out for you.
tidal kiln
just checking - the UF2 bootloader updaters (from cp.org) will set the write protection bits after updating?
tulip sleet
@tidal kiln, yes that's right, it's an easy to way to make sure the BOOTPROT fuses are set properly. You can update to the same version
tidal kiln
oh. hey! a wild danh appears. excellent.
thanks. would you mind responding here:
https://forums.adafruit.com/viewtopic.php?f=57&t=167795
tulip sleet
sure
idle wharf
I've been doing some research about the ESP32-S2 camera's capabilities (in gist here: https://gist.github.com/askpatrickw/0179d09e74d5f2a4347d5666ea937c4d ). The camera functionality is accessed via I2S, the FeatherS2 exposes I2S as DAC1 and DAC2 (pin23,pin24) assuming all that is correct, the next question I have is about I2S support in CircuitPython.
The only mentions of that I find are in relation to audio, specifically audiobusio where I2SOut is implemented.
That leads me to think, in order to use the I2S Camera functionality of the ESp32-S2 a new library would be needed.
Does that sound right or am I way off?Espressif IDF Docs for I2S
https://docs.espressif.com/projects/esp-idf/en/latest/esp32s2/api-reference/peripherals/i2s.html
https://github.com/espressif/esp-idf/blob/625bd5eb1806809ff3cc010ee20d1f750aa778a1/components/soc/soc/esp32s2/include/soc/i2s_struct.h#L347
I probably should have waited to post this on a weekday. I was curious what the team thinks about the ESP32-S2 Camera support... (and if I'm understanding it correctly)
slender iron
@idle wharf I think it's on @onyx hinge's radar
onyx hinge
I don't know anything specific about it but adafruit may ask me to work on it at some point in the future
idle wharf
๐ got it
solar whale
@slender iron do you want the adjustment for the increased pystack to be made for the CIRCUITPY_DEFAULT_STACK_SIZE for all ports or just for the atmel-samd (SAMD21/51). in particular for the cxd56,stm,esp32s2,litex ?
solar whale
ok -- should I just do it there an leave the others alone
slender iron
either way ๐
orchid basinBOT
solar whale
ok -- I'll put an a PR with only the SAMD21 and we can discuss it there.
slender iron
k, thanks!
solar whale
@slender iron I am working with the the tip of main and now the frozen bus_device module is out of date and the CPX library fails.. how to I update the frozen module in my repo.
slender iron
@solar whale find the module in the frozen folder and update it like you would in git normally
actually, I think there is a make target to do it
at the top level
manic glacierBOT
low sentinel
winding rules: https://en.wikipedia.org/wiki/Evenโodd_rule
@onyx hinge yeah, vectorio.Polygon uses the non-zero winding idea like standard SVG (not even-odd). It has a fixed ray direction though, and per w3.org:
The above descriptions do not specify what to do if a path segment coincides with or is tangent to the ray. Since any ray will do, one may simply choose a different ray that does not have such problem intersections.
I suspect that if I were to choose a different ray direction, I could fix it for rectangles - but any fixed ray angle will be susceptible to ill-defined behavior rising from a colinear ray and segment path.
Might help to do the calculation in floating point and have some heuristic for nudging values by an epsilon in the case of coincidence - but ideas there I think would tend to produce a different artifact along segments (like shape -> 1 empty pixel -> 1 ghost pixel from epsilon) at certain angles and that would slow the calculation tremendously on uc's without fpu's.
or, if you prefer meme form, one does not simply choose a different ray :-p
manic glacierBOT
onyx hinge
@low sentinel using the top-left rule https://docs.microsoft.com/en-us/windows/win32/direct3d11/d3d10-graphics-programming-guide-rasterizer-stage-rules will give consistent results, but that assumes you reduce your polygon to triangles. When rasterizing arbitrary shapes, you can apply a top-left-inspired rule. Say that you approach rasterization by thinking "across" rows. You just need to know the segments of the shape that cross that row (a fully horizontal line NEVER crosses a row), the X-coordinates of their intersections, and then you can apply your fill rule.
sweep line algorithm will let you efficiently know what segments cross the row you are dealing with, and you also have to sort the intersections within a row.
https://github.com/jepler/dashing/blob/master/dashing.hh#L130 is code I wrote in the fairly distant past that is vaguely related to this, though I notice it works on doubles and not integers
the top-left rule or variants are nice because when you put shapes next to each other that share endpoints, you know the rasterization will be without gaps and without double-covering pixels
(say you wanted a spinning beach ball animation using shapes...)
manic glacierBOT
low sentinel
@onyx hinge Thanks, I hadn't seen the top-left rule. In effect, vectorio.Polygon does a bottom-right currently. I think these rules produce shocking results, but I'm not an expert in rasterization
this 3x3 shape produces a 2x2 rectangle
onyx hinge
It's true, DirectX and OpenGL is probably not designed so that beginning users get the results they want, but so that advanced users can get the results they want with care
If [1:3] is a slice of 2 items in Python, then a rectangle where low_x = 1 and high_x = 3 being 2 pixels wide is in some sense just as defensible
low sentinel
sure but what is your shape going to be like when you have something more interesting than a rectangle
weirdly shifted by 1 pixel it seems
in python am array starts with index 0, and ends at index len()-1. That behavior seems most natural to expect
onyx hinge
I'm happy to leave the actual decisions with you ๐ but this info about how people thought about rasterization in the 80s and 90s is getting harder to find and I wanted to pass it on in case it was missing knowledge for you.
in CP the goal I think should absolutely be helping beginners get a shape that makes intuitive sense, I just wish it wasn't in tension with other things
low sentinel
80's and 90's tech is what I've been rediscovering on my last 3 projects. Thanks ๐
onyx hinge
consistently applied, bottom-left and top-right (and the other combinations) are all equally valid, but the one which is most natural seems to depend where you put your (0,0) pixel.
a mode where you draw all the edges too (using the diamond rule ๐ if we're lucky) would probably help with the expectations of a beginning user, but that may complicate the rasterization too much
I think that if you followed the rasterization of those two triangles (or equivalently the square) with drawing the 4 outside lines, you'd get 8 pixels from the border + 1 pixel of interior. 3 of those border pixels would also have been filled by the solid rasterization step.
hmmm circa 1996 there was a software opengl (1.1) rasterizer much smaller than mesa (even circa 1996 mesa). I wonder what happened to it. It would probably fit on micros now.
low sentinel
I'd really like to see an adafruit EVE2 display. I've done some good things with those chips.
if there isn't already one, it wouldn't be too difficult to add a circuitpython library. I already wrote one for the particle photon
slender iron
@tulip sleet do we have mpy-cross builds for 5 and 6?
@low sentinel there is a _eve module already
added by the gameduino dev iirc
tulip sleet
no, there is .zip of a bunch of 4.x or 3.x builds
tulip sleet
we need to add a windows CI build of mpy-cross and then actually upload them
low sentinel
there is a
_evemodule already
@slender iron oh yeah!
okay, i'm ready to give adafruit more money
slender iron
@gilded cradle you'll know the answer to: https://forums.adafruit.com/viewtopic.php?f=60&t=167826
gilded cradle
@slender iron I'll take a look
slender iron
thanks!
tidal kiln
@slender iron just came across this old post in another forum and moved it to circuitpython:
https://forums.adafruit.com/viewtopic.php?f=47&t=164569
might make for a good FAQ somewhere?
slender iron
@tidal kiln that's the first time I've seen it. they'll have to make their own terminal
the built in one with blinka is fixed
tidal kiln
ok. thanks.
onyx hinge
oh look another day where I basically didn't read this channel or check my github pulls & issues. ๐ฅ
@tulip sleet there is a build artifact called mpy-cross.static-x64-windows made by current builds. However, it is cross built from Linux using mingw. I have tried it, but only on wine, so that's maybe not the greatestest test
usage: Z:\tmp\mpy-cross.static.exe [<opts>] [-X <implopt>] <input filename>
Options:
...```oh, I'll try it on windows. I have many windows machines ๐
onyx hinge
there's also one for macos catalina, raspbian, and amd64-linux
tulip sleet
they aren't uploaded to S3 right now, so that's the last step. We could create a separate mpy-cross directory, like the language subdirs, I guess
onyx hinge
yeah I didn't know how to do the s3 things but it would be a great addition. finding things in actions artifacts is not great.
tulip sleet
@onyx hinge OK, ran that .exe on Windows, worked fine, and the compiled .mpy (Hello World) works
i can look at the upload. It should be similar to the .uf2 etc. uploads
thanks for finding that; i had no idea
onyx hinge
sure thing (I implemented it)
tulip sleet
@onyx hinge I'm working on a PR for mpy-cross upload. not hard
onyx hinge
thanks, that'll make it much more useful to people. will you cover all of the binaries we generate?
tulip sleet
yes, and the filenames will be tagged with the version
onyx hinge
aha I think this is the small opengl implementation I remember from all those years ago https://bellard.org/TinyGL/
Fabrice Bellard is a certain kind of genius
crimson ferry
Why do SAMD51 CircuitPython devices have a USB path like /dev/tty.usbmodem14701 that is consistent with the physical USB port it's plugged into, whereas ESP32-S2 CircuitPython devices have a USB path like usbmodem1337DEADBEEF1 with the MAC address + "1"(macOS)?
Maybe the better question is, why don't SAMD51 (or other non-ESP32-S2 CircuitPython) devices present some unique-to-the-device USB path?
ionic elk
@fiery silo hoping to give you a poke about setting up the Open Book with Circuitpython. I'm not seeing anything appear on the screen and I'm wondering if there's a trick I'm missing. Also ImportError: cannot import name FlashBabel with your example code, not sure what's up wth that
But I might just be being dumb and missing something
fiery silo
@ionic elk No worries, I think I refactored that class to just be called โBabelโ at some point, but would have to duck back into the code to check it out (Iโve been mostly in Arduinoland lately)
Would be happy to regroup with you tomorrow about it either here or in DM, whichever works best
ionic elk
Sure I'll send you a ping in DM
Any ideas why the display might not be working? Libraries, maybe?
fiery silo
Youโre using the latest CircuitPython build for the open book board? https://circuitpython.org/board/openbook_m4/
ionic elk
that's the one
Or rather I've tried both, the 5.3 and the 6
I tested the board with Arduino so I know it isn't the hardware
fiery silo
At least in the case where you just print things to the console and let the code finish executing, the screen should just refresh to show the console output. If youโre drawing to the screen with displayio though you will need to call display.refresh() to make it update
ionic elk
I'm not even getting that.
hmmm maybe I should load up arduino one more time and just make 100% sure it isn't hardware
fiery silo
Also just to be sure is this the Open Book board or the e-book featherwing?
fiery silo
Yea for that one it is more manual, youโll need to use the Feather M4 UF2, youโll also need the appropriate EPD libraries and then youโll have to instantiate the display in your code
ionic elk
Yep that'd do it, I should have thought of that
Do you have the display instantiation code so I don't have to port it from C?
fiery silo
All good, I need to improve the documentation on this point
Itโs just the IL0398 driver
ionic elk
I just meant the pins but it's cool shouldn't take a minute
fiery silo
Ah right, CS is 9, DC is 10, reset is A4 and busy is A3
And SPI is just the main board.SPI
ionic elk
nice, TY!
And what did you say the proper inclusion for the Babel library is? from babel.babel import Babel?
that'll work with the flash chip?
fiery silo
Let me double check on that right quick, itโs been a minute since I looked at that code
ionic elk
Sorry to bug you so late at night I can ping you tomorrow
Screen is updating, though, so that's fun!
fiery silo
Itโs all good! Easy for me to look at now (and Iโm most productive late too)
ionic elk
Here's what I've got:
https://gist.github.com/hierophect/5082f9527efa68718a42942928a46659
I see updates on reloading, but no text, it's just white
fiery silo
Ha ok so I definitely shuffled some things around in a confusing way (left it in kind of a work in progress state)
Itโs from babel.babelflash import FlashBabel
Lemme gist this
Has dependencies adafruit_bus_device, adafruit_bitmap_font and adafruit_display_text
oh also
ionic elk
hmmm I've got all of those but I see AttributeError: 'FlashBabel' object has no attribute 'info_for_replacement_character'
fiery silo
Huh. Whatโs the stack trace for that?
ionic elk
Traceback (most recent call last):
File "code.py", line 11, in <module>
File "/lib/babel/babelflash.py", line 36, in __init__
File "/lib/babel/babel.py", line 71, in __init__
File "/lib/babel/babel.py", line 101, in _fetch_glyph_basic_info
AttributeError: 'FlashBabel' object has no attribute 'info_for_replacement_character'
OH
Did you hit the step in the setup guide where you burn the Babel.bin file to the chip using an Arduino sketch?
ionic elk
yeah it should be on there
I had it displaying chinese and japanese with arduino
But I did that a while ago, maybe it's out of date?
fiery silo
Nah, the Babel.bin file hasnโt changed in ages...
ionic elk
In the short term though I'm more concerned that I can't get a terminalIO font up and running. Does my setup look ok to you?
fiery silo
Ah, try adding display.show(None) after you instantiate the display
ionic elk
I think I see a very very faint blinka up in one corner?
but it's just like a smudge of pixels? maybe a contrast issue?
fiery silo
But the contrast looks fine on Arduino?
ionic elk
yeah, had no problems there
This whole thing works on your featherwing?
oh what
uhhh
on a whim I changed the terminalIO text color from 0xFFFFFF to color=0xFF0000 and now it works
lol
fiery silo
I was actually about to type something about the highlight color
ionic elk
highlight is still 0x000000
fiery silo
But color was part of the solution at least!
ionic elk
indeed, thank you!
I appreciate your help
just this info_for_replacement_character thing now
fiery silo
I will look into it tomorrow, itโs weird, it looks like itโs crashing when itโs trying to set that attribute which doesnโt make a lot of sense
Though, if itโs a very old babel.bin file, maybe
ionic elk
would have been from mid spring
fiery silo
Yeah the big change happened back in feb, so, hmm.
ionic elk
well thanks for your assistance, it's almost there! Appreciate you staying up late to help out
fiery silo
Anyway Iโm glad we got halfway there, happy to have helped! Yea Iโll loop back with you in DM tomorrow
ionic elk
Coolbeans. I'm using Babel to make a japanese flashcard app and it'd be neat to show it on Show and Tell tomorrow
fiery silo
Rad! ๐
ionic elk
it is you who are rad, fitting all of these glyphs on a humble nor flash ๐
low sentinel
i can't wait for my segger j-link to get here
anyone set this up yet? https://wiki.segger.com/J-Link_Visual_Studio_Code
ionic elk
๏ผ joey for when you get on tomorrow: I'm not grepping any instances of info_for_replacement_character outside babel.py - did you maybe accidentally cut them from the class?
manic glacierBOT
low sentinel
atJeffE: Not going to do the thing with exact fill. It'll be top-left, as is the done thing in other libraries.
it fills top-left whether you go clockwise or anticlockwise; it'll do fine.
fiery silo
Morning @ionic elk! FYI, I was able to replicate your issue in the end. It has to do with the babel.bin on the chip being outdated; if you erase it and burn the latest version, it should work. https://github.com/joeycastillo/babel/raw/master/babelconvert/babel.bin
basically i made some changes to the babel.bin header in feb to accommodate more Unicode planes, and the CircuitPython library isn't seeing the new values for first and last glyph.
the reason the Arduino library still works is that it still has some of that hard-coded; I hadn't gotten around to making it load those values from the Babel image instead ๐
ionic elk
@fiery silo can I load babel.bin from circuitpython? Or do I need to swap over to arduino again to do that?
fiery silo
You would need to run the BurnBabelBurn Arduino sketch with the new bin file on an SD card, yea
But, yโknow, easy to switch back with a CircuitPython uf2. The Arduino sketch wonโt touch the files on your featherโs flash so it should just go right back to working (though backing up never hurts!)
lone axle
Love the name of the sketch
ionic elk
I feel like an idiot because I did this before, but is arduino just supposed to work after the board has been programmed with circuitpython, or do I need to run burn bootloader, or put it in UF2 mode, or what?
fiery silo
Youโre all good, the bootloader is the same so you should be able to just run a sketch
tho it never hurts to double tap reset
ionic elk
@fiery silo no luck with latest babel.bin
assuming that is the latest, up on the website?
I get the same error
fiery silo
So weird. You erased, wrote and verified it?
ionic elk
erased, written, verified
here's my babel
(I renamed it before flashing)
maybe it's different than what you have?
fiery silo
HUH. Lemme take a look.
ionic elk
babelconvert/babel.bin does say it was last updated 6 months ago
last change was add first_codepoint field
could you send me the babel.bin that's working on your end?
fiery silo
Just checked with diff, theyโre identical
ionic elk
wat
fiery silo
Yeah, you have the right bin file. And youโre using the latest library files I take it?
fiery silo
This is so confusing. But maybe we should move to DM so as not to monopolize the channel
ionic elk
yeah sounds good this is getting esoteric
fiery silo
Indeed
onyx hinge
hmmmm does bitbangio actually work with adafuit_sdcard? I should check ..
no it doesn't. scratch that off the list
ionic elk
@slender iron does Label support text wrapping?
slender iron
ยฏ_(ใ)_/ยฏ
crimson ferry
@ionic elk not last I checked, but you can insert \n
lone axle
@ionic elk @crimson ferry is correct as of now no automatic wrapping in Label. There is a rudimentary wrap_nicely() function in the PyPortal library though that can be modified to suit other purposes if you do want to have a Label that wraps without having to "manually" insert the newlines . https://github.com/adafruit/Adafruit_CircuitPython_PyPortal/blob/f53287de5b0c406e05badcedb802c3871383c136/adafruit_pyportal.py#L1101
ionic elk
It's probably fine, the font I'm using is monospace
So it's easy to calculate line lengths
ionic elk
Does anyone know how to change time_to_refresh on an e-paper screen? Right now it reads as a ridiculous value of 180 seconds, but if I try to change it I get AttributeError: 'IL0398' object cannot assign attribute 'time_to_refresh'
tulip sleet
I think a higher time to refresh can damage the screen; that's why the value was picked
ionic elk
Makes sense but three minutes seems a tad high
I rewrote it to be about 5 seconds, as per @fiery silo 's recommendation
The primary issue seems to be that the value cannot be changed from the python script itself
fiery silo
I know the tri-color displays want three minutes between refreshes, but the black and white ones seem to be fine refreshing more quickly in my experience
tulip sleet
@ionic elk @fiery silo https://learn.adafruit.com/adafruit-eink-display-breakouts/usage-expectations
The bookereader displays are very different than the small ones we sell
lapis hemlock
@tulip sleet Dan, I am thinking of compiling ulab into a native module. Does that make sense at all? What should I watch out for? Are there performance issues?
tulip sleet
i'm not sure what you mean; it's already a native module, because it's in C.
do you mean shared-* vs extmod? There would be no performance difference
lapis hemlock
No, what I meant is an .mpy file that you can copy to the hardware, without having to re-compile and flash the firmware: http://docs.micropython.org/en/latest/develop/natmod.html
ionic elk
@tulip sleet interesting I didn't know that tri-colors needed that degree of caution
tulip sleet
unfortunately, we don't support that yet -- it was added after our last merge from micropython (which was a long time ago)
lapis hemlock
But you do have .mpy files to distribute, don't you? Or are those simply the outputs of mp-cross, and come from .py sources?
tulip sleet
right, the latter
lapis hemlock
OK, then I will have to dig somewhere else. Thanks!
tulip sleet
it's a very good idea, it would save a lot of juggling of what goes in what port, but we're not there yet ๐
lapis hemlock
Right, I was about to ask, whether that could simplify your whole build process.
You could probably just release the major versions, like 6.1, and the modifications/improvements would come from the .mpy files.
Is it somewhere in the pipeline, or you haven't yet seriously considered it?
uncut nexus
@onyx hinge - Do you have a couple minutes to chat sometime?
crimson ferry
@ionic elk I think the expectation is that you would init with seconds_per_frame= and not need to change it in user code, but you can check time remaining
ionic elk
@crimson ferry I missed that seconds_per_frame would dictate time_to_refresh, you're probably right
crimson ferry
fwiw, I was running my 2c & 3c ePaper FeatherWings at much faster refreshes for several weeks or months with no ill effects, longer term YMMV
ionic elk
Might be a manufacturer warning that's just super duper cautious
tulip sleet
@lapis hemlock after 6.0.0 comes out, we'll probably wait for micropython 1.13 to come out, and then we'll do a merge. But the last merge took me 2 or 3 weeks to do, and this may be more
lapis hemlock
@lapis hemlock But the last merge took me 2 or 3 weeks to do, and this may be more
@tulip sleet Yeah, I can imagine.
solar whale
@onyx hinge where is mount_sd? is it something not yet merged?
onyx hinge
@solar whale it is a file that the user creates according to instructions in the accompanying guide.
mount_sd.py in CIRCUITPY
solar whale
๐ณ oops -- I need to read the whole thing.... I was just looking at the example
onyx hinge
I don't know that I explained it in the PR
manic glacierBOT
uncut nexus
@Jeff E. - Do you have a couple minutes to chat sometime?
onyx hinge
@uncut nexus probably -- what's the topic?
uncut nexus
issue 3216 - pulseio is broken after your background task commit
onyx hinge
OK
uncut nexus
I know what the issue is - just want to discuss the best way to fix it
onyx hinge
great. do you want to stay with text or switch to voice?
uncut nexus
voice is better, I think
onyx hinge
either way I hope you'll add a brief summary to the issue once we're through.
OK
let me find my headphones
onyx hinge
@uncut nexus sorry if I started out being a little defensive, every point you made was right on.
tulip sleet
@slender iron do you want to review https://github.com/adafruit/circuitpython/pull/3220? If not I will merge and make sure the uploads really happen. Right now it's stuck in Actions, but it ran the non-board jobs fine.
slender iron
@tulip sleet nope, all good
slender iron
@ionic elk if you are using monospace then I suggest using a tilegrid instead of label. it'll use less memory
tulip sleet
there's a huge backlog of actions, i'll need to wait
slender iron
and epaper refresh time is seconds_per_frame here: https://circuitpython.readthedocs.io/en/5.3.x/shared-bindings/displayio/EPaperDisplay.html#displayio.EPaperDisplay
ionic elk
@slender iron thanks! I've swapped back to Circuitpython work for the day but I'll check that out for the future.
silver tapir
I hope to ask this in the next meeting, but in the meantime, does anyone object if I try to translate the .org site to spanish?
(want to have this ready for 9-9 
)
tulip sleet
@silver tapir I think we would want to have a single site still and have a language tab/setting. Does that make sense to you?
silver tapir
@tulip sleet Yes. If I do things right, only if you have your browser configured for "ES" content, you won't see I did anything.
tulip sleet
sounds good!
though an explicit tab might be good also
I don't know how various sites do this: some are automatic but you can still switch languages at will
silver tapir
Yes, wikipedia for example has both. I tries to detect my lang, but I can still choose on the left side what translation do I want.
And yes, it sounds better this way.
manic glacierBOT
onyx hinge
is this just some actions weirdness? must be. https://github.com/adafruit/circuitpython/runs/924407804?check_suite_focus=true
tulip sleet
I looked at the raw log, and saw this weirdness: it can't find a machine to use:
2020-07-29T17:59:31.9783608Z ##[section]Starting: Request a runner to run this job
2020-07-29T17:59:34.9155502Z Can't find any online and idle self-hosted runner in current repository that matches the required labels: 'ubuntu-18.04'
2020-07-29T17:59:34.9155612Z Can't find any online and idle self-hosted runner in current repository's account/organization that matches the required labels: 'ubuntu-18.04'
2020-07-29T17:59:34.9155689Z Can't find any online and idle hosted runner in current repository's account/organization that matches the required labels: 'ubuntu-18.04'
2020-07-29T17:59:34.9155846Z Found online and busy hosted runners in current repository's account/organization that matches the required labels: 'ubuntu-18.04'. Waiting for one of them to get assigned for this job.
2020-07-29T17:59:36.1963698Z ##[section]Finishing: Request a runner to run this job
we seem to have overwhelmed the actions system somehow
also I saw messages on other jobs about DBCPU limit being reached
onyx hinge
ah I didn't think to look at the raw log. There was no "opening triangle" for the failed step.
tulip sleet
Yah, the triangles aren't working right either.
onyx hinge
nice (mpy-cross uploads)
tulip sleet
there's now a central location, which is the best thing; we should also put links in circuitpython.org somewhere.
onyx hinge
I don't know how many people are going to use a commandline program to make mpy files but I think at least a few would.
on anything that's "express" it seems like just copying the py file is fine
but I haven't been pushing the limits in terms of complex programs, some people do
prime flower
@onyx hinge buuuut if there were a drag+drop frontend for making .mpy files ๐
(at least I'd use it...)
prime flower
dang!
kind river
@slender iron I added memoryio to microcontroller to match nvm. It builds for fomu, but unfortunately I don't have one to test with and the orange crab isn't added to main. https://github.com/ryang14/circuitpython/tree/raw-memory
slender iron
@kind river nice! it should be doable on non-fpga ports too
kind river
Yeah, I'm going to try testing on an m4 tomorrow. Is there a section of memory that I can be fairly sure won't be used by something else?
Or I could just use an array for testing
slender iron
on the samd51 you could poke backup ram
kind river
Okay, I'll try that. Thanks
slender iron
np, neat! I was reworking the lamdasoc stuff here: https://github.com/tannewt/systemonachip
set up sphinx on the repo last night
kind river
Cool, I'll check it out
slender iron
I'm hoping to output a ninja build file instead of calling everything from python
kind river
Neat, would a proper build system let it do stuff like only resynthesize if the generated HDL was actually different?
slender iron
ya, that's my hope
I don't know how parallel the work is though
@kind river memoryio.ByteArray doesn't need to be a singleton. I'd add a constructor that takes in the start address and length
each port can then block some addresses that we don't want to support
kind river
yeah, I still need to rip out some nvm remains
it might also make sense to make it a separate module to allow for porting between accessing the bus locally or through usb/uart/etc from linux
slender iron
yup! That's what I was thinking. You can pass in anything that acts like a bytearray
we may want to lock it down a bit for access alignment
tulip sleet
@kind river memoryio is nice! is the idea only for RAM? I had a use for flash access (readonly would be ok,) as well: I wanted people to be able to read the fuse flags to help for support issues. So I would have a use for a memoryio array based at 0x0.
slender iron
@tulip sleet we came up with the idea for fpga actually
so we can write a python class that both defines a peripheral's gates and also be the driver for it
kind river
I think it makes sense to store start addresses/sizes in microcontroller like pins so you could do something like sram_data = memoryio(microcontroller.memory.sram)
tulip sleet
๐ all i want to do is fetch 8 bytes starting at 0x804000
other values I might want to fetch are scattered in various places, and are chip-dependent.
kind river
I mean I think that should be available as a shortcut if that makes sense
tulip sleet
seemed like a low-level peek/poke interface might be helpful. I was not expecting to use this for regular use, but for debugging/support: "Please check this address and tell me what you see", etc.
otherwise I've had to write a small Arduino program and tell the user to load it, which is a lot of work and requires some extra skills
i did start working on it and cloning nvm, but i didn't finish it, due to the kind of refactoring that you are doing now
kind river
having defined regions would be potentially useful for my intended use of accessing memory mapped custom peripherals
however, this may be taken care of by the higher level drivers in that case anyway
slender iron
I'm fine having predefined regions on boards where we have the memory
tulip sleet
you could use a memoryview on a slice of the virtual bytearray representing the entire address space
slender iron
and I'd like the constructor to have some guard rails so it's hard to actually hose yourself
tulip sleet
periph = memoryview(microcontroller.address_space[0x444000:0x4444100]
yes, a constructor might be safer; i'm just thinking idly
slender iron
a constructor can then have an iknowwhatimdoing kwarg
tulip sleet
microcontroller.dangerous_raw_memory
kind river
the constructor could take either a predefined or custom region
and maybe treat custom as read only unless explicitly told otherwise or something
slender iron
maybe we shouldn't call it ByteArray because it's usually always writeable (bytes objects aren't)
wooden nest
microcontroller.dangerous_raw_memory
@tulip sleet
Sorry for intruding, but i just wanted to say that if you do something like this, you might want to make it read-only or somehow.
Or make it a function that returns an internal variable you shouldn't touch?
I should go to sleep
tulip sleet
@wooden nest thanks, yes, we want to make it obvious it's dangerous, by default, read-only, etc.
as someone in college once told me, "sleep is where it's at"
onyx hinge
@slender iron I have a request for adding a Hindi translation of CircuitPython. It looks like a "complex" script with combining characters, but it has a romanization. I assume I should add the romanization? fwiw the requesting user is weblate @spring juniper2604 who I can't directly relate to being active in CircuitPython to date .. there's not a github user with the same exact name and the weblate account is new and doesn't have activity. I'm also not sure how this person would be informed that I added a different translation (hi_Latn) instead of the one they requested (hi). Or even if they are informed that I added a requested translation.
onyx hinge
oh good!
slender iron
we could do the complex version if we disable terminal right?
onyx hinge
Yeah it's a question if whether to prioritize showing up in a proper unicode context looking like a native reader would expect (but I hear mu is not great for utf-8 support), or able to work with displayio and the on-display terminal..
same tradeoffs as zh_Latn
slender iron
I think it'd be good to get it in and then we can disable the terminal
the prints to terminals besides mu should be ok
onyx hinge
OK so go ahead and add hi, not the romanization? I can do that.
slender iron
I think so
onyx hinge
we'll rope Joey into getting us a full unicode terminal someday soon right? Anyway it's ready for @spring juniper ! https://hosted.weblate.org/projects/circuitpython/main/hi/
And thanks in advance for your contributions
slender iron
heh, ya. that's my hope
onyx hinge
light cursing the stm sdio PR is taking longer than I thought to get to the state of "actually building". Oh well.
low sentinel
@tulip sleet this lib may interest you https://github.com/WarriorOfWire/tasko
manic glacierBOT
turbid radish
?serverinfo
digital shoreBOT
adafruit
onyx hinge
do we have a graphic for 23456 members?
turbid radish
No but feel free to pat the person on the back
lone axle
Does the bot pull those numbers from a public API?
onyx hinge
lone axle
Thank you
onyx hinge
@solar whale because that's how we've done it in the past, I guess?
solar whale
Seems overly complicated to me.
manic glacierBOT
solar whale
@tulip sleet are patches to gcc10 still needed or can it be used "out of the box"
tulip sleet
@solar whale Do you mean "to gcc10" or "for gcc10"? Scott patched the CPy code so it would compile, and that has been merged. He has been testing gcc10 with the arch Linux package; I don't know how that compares with the ARM preview download.
solar whale
ah -- OK -- I was under the impression a patch was being made to the gcc10 distribution. Glad to hear it is not....
slender iron
@tulip sleet @onyx hinge think it's worth me externing things?
onyx hinge
Right this second? Maybe not.
slender iron
it shouldn't take too long to go through them all
onyx hinge
if you do, feel free to push it to that branch
fringe trench
I'm listening to the Circuit Python weekly. Do you still need reviewers?
lone axle
@fringe trench we are always happy to have new reviewers ๐
fringe trench
I need to grind rust off my code reviewing skills, so if you don't mind, I'd like to help.
lone axle
This page is a great place to get started: https://circuitpython.org/contributing it will list all of the open pull requests for the Libraries
If do take a look at some and/or do some testing of anything please feel free to post your thoughts and/or test results on those PRs in Github.
fringe trench
Will do. I'm pcurry on github, my icon is my Gravatar.
It's also a guitar I own. ๐
lone axle
Nice! Thank you for helping out. Please feel free to ask for help if you start looking into something and wind up having questions about the Circuit Python, or Github PR Reviewing etc..
fringe trench
@lone axle honestly, my first pass is just going to be formatting, whether the tests run, etc.
silver tapir
I present to you, Quetzalblinka, the mascot for CircuitPython Day Centralamerica.
onyx hinge
adorbs
I wonder whether python3's _pydecimal.py can be hacked down into something that'll fit on circuitpython. I'd like to do non-integer arithmetic with more precision than our floats
the mpy file of it is 82kB
turbid radish
?serverinfo
digital shoreBOT
adafruit
simple pulsar
@onyx hinge Is a fixed point approach using existing CP "big" integer viable for what you're doing? Might work if you only need certain math ops
onyx hinge
I think it will have to build on big integers. Goal is a 16+ digit RPN calculator with keypad and display
onyx hinge
maybe more like a non-graphing hp48?
A 2inch-ish display https://www.adafruit.com/product/1770 and about 40 "12mm" buttons, it would be physically bigger than the HP48. Maybe comically big.
solar whale
@tulip sleet FYI -- tried gcc 10 on linux (Ubuntu 20.04) -- seems OK for atmel-samd. As noted, nrf fails. have not tried others yet. I am also trying to build a toolchain for the RPi. Takes about 8 hours to build -- sorry --initially posted to wrong channel
Heatsink and fan are holding the RPi4 at a comfortable 55C...
onyx hinge
@solar whale did you try many of the created uf2s? sometimes they have had interesting things newly busted ๐
solar whale
no -- I haven't -- I'll try some -- what issues have you seen?
but that is why I want the RPi build -- still skittish about my Linux box and resetting boards...
silver tapir
@solar whale You could speed up the rpi build by using tools like distcc, that allow you to use the compiler from another (o several) boxes.
(the only recommendation is to have the same compiler version in all distcc systems)
solar whale
@silver tapir Thanks -- I'll look into that
silver tapir
Let me know if you need some help. I sometimes use a +160 xeon cores to compile slackware-riscv64 packages.
(Quite a rush vs a 4-core riscv64 box, or emulating riscv64 in qemu)
solar whale
@onyx hinge @tulip sleet When I do my usual git pull and submodule updates, the frozen/Adafruit_CircuitPython_BusDevice submodule is not getting updated. Should it? I was not sore if the PR I did for the CPX updated it or not. Or is there plan to update al the frozen submodules at some point.
Do I have to do something special to update it? For now I just did a git pull origin master in the BusDevice subdirectory and it works for me. Just tried CPX built with gcc10 -- seems to be happy.
onyx hinge
Somebody needs to pull-request the update of the submodule. Is that something you'd be comfortable doing @solar whale ? You use git commands in the submodule to update it to the latest release tag, build & test until you're happy, then "git add" and "git commit" the change in circuitpython (or the graphical equivalent) and then create a pull request in circuitpython.
it would be good to do that, it should probably be considered as milestone 6.0.0 as well if there's not an issue already
solar whale
hmm -- so PR #3213 https://github.com/adafruit/circuitpython/pull/3213/files ddi not accomplish that?
how do I update the release tag?
Several other submodules have had updates as well. Probalbly should all be updated before 6.0 release
I have to go AFK for a few hours -- I'll look into this more when I can
tulip sleet
@solar whale I have a simple shell script to update a frozen module to the latest release (not the latest commit): circuitpython/tools/git-checkout-latest-tag.sh I run this over all the frozen module dirs with a bash for loop.
slender iron
someone want to help bundle this: https://github.com/adafruit/Adafruit_CircuitPython_TC74 ?
@ionic elk from your status update Espressif has a funny habit of naming their peripherals in a way that makes them sound kind of limited, when the truth is just the opposite ๐ฏ
onyx hinge
@slender iron do you know if I can document a "protocol" (interface) in our pyi stubs?
ionic elk
@slender iron Since neopixel_write doesn't really have a "construct", do you think it would be good if I put in a flag to detect whether a neopixel write request is the first one for a reset cycle, and then reserve an RMT channel indefinitely for it?
Or should I reserve and free an RMT channel for every neopixel write request?
slender iron
@onyx hinge I imagine you can but ciscorn is probably the best person to ask
@ionic elk I'd reserve and free every request unless it's very slow
ionic elk
Ok, I haven't done any timing analysis yet but I don't expect it is
Other quick question @slender iron I notice that functions like rmt_wait_tx_done are calling stuff from FreeRTOS for timeout functions, like xSemaphoreTake. Is that all good?
or do I need to reconfigure the timing to avoid it or something
I also can't seem to get it out of GRB mode...
slender iron
freertos stuff should be ok
ionic elk
I thing the GRB thing is a library problem? If my understanding is correct, when I put in:
pixels = neopixel.NeoPixel(board.IO18, 1, pixel_order=neopixel.RGBW)
pixels[0] = (10, 0, 0, 0)
It should be moving that 10 value to the R slot, since Neopixels use GRB order. But it's still first when I receive it in Neopixel write, and I don't see that the W does anything at all. Should I open an issue or is this something that needs fixing on my end?
idle owl
@silver tapir Quetzalblinka is amazing! I love it!
onyx hinge
def __init_(self, width=64, height=32, bpp=16):
๐คฆ
manic glacierBOT
slender iron
nice @ionic elk ! will review on my stream
onyx hinge
@slender iron are there 1-bit displayio displays? or just epaper?
ah Adafruit_CircuitPython_DisplayIO_SSD1306 maybe
I had only found the "old" one at first
slender iron
Ya oleds are a few bits too.
kind river
@slender iron For memoryio, should the contructor go in init or ByteArray? Assuming we even want to keep the ByteArray behavior.
slender iron
bytearray
init is for module level stuff
(had a good conversation with whitequark on the #nmigen channel on freenode)
onyx hinge
I don't think my 1 bit display is right yet (should be a 64x32 console with blinka)
closer!
crimson ferry
where can I find out more about memoryio goals/features? Github search on circuitpython repo yields zero results
kind river
@slender iron We should probably open a github issue for memoryio to organize conversation about it. Do you want me to do that?
The main goal is to allow for pure python drivers for custom memory mapped peripherals when running on a FPGA SoC
The secondary goal is to make it also usable for debugging or accessing stuff like RTC ram on other boards
slender iron
@kind river yes please
lone sandalBOT
hidden geyser
I've recently built a port of the Qwiic Scale lib in CircuitPython, what's the recommended way to propose a new CircuitPython device driver?
silver tapir
@ionic elk I was trying your esp32-neopixel branch, and I'm getting a compile error. I don't see it in github's ci, but I'll ask anyway.
I'm getting:
$ make BOARD=espressif_saola_1_wrover
Use make V=1, make V=2 or set BUILD_VERBOSE similarly in your environment to increase build verbosity.
common-hal/neopixel_write/init.c: In function 'common_hal_neopixel_write':
common-hal/neopixel_write/init.c:105:9: error: implicit declaration of function 'mp_raise_ValueError' [-Werror=implicit-function-declaration]
mp_raise_ValueError(translate("Could not retrieve clock"));
^~~~~~~~~~~~~~~~~~~
common-hal/neopixel_write/init.c:105:9: error: nested extern declaration of 'mp_raise_ValueError' [-Werror=nested-externs]
cc1: all warnings being treated as errors
make: *** [../../py/mkrules.mk:55: build-espressif_saola_1_wrover/common-hal/neopixel_write/init.o] Error 1
I'm using xtensa-esp32s2-elf/esp-2020r2-8.2.0 which I think is ok. On ubuntu 20.04.
low sentinel
i made a mistake and now i need a segger-port-to-idc cable
tulip sleet
@hidden geyser Check out this guide: https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library. We have a community bundle of contributed libraries.
hidden geyser
@tulip sleet nice, everything I need
This is weird, I use cookiecutter, gives me a git clone error. I clone the repo it's trying to clone manually and... indeed it fails, is it just me?
Actually I know what that issue is, it's an invalid folder name on the Windows filesystem
tulip sleet
yes, it works fine here on Linux
you may want to use WSL. WSL2 is preferred: much faster disk I/O
hidden geyser
I just upgraded WSL 2 days ago, time to check out my shiny new ubuntu โจ
This page might need a warning for Windows users: https://learn.adafruit.com/creating-and-sharing-a-circuitpython-library/creating-a-library
tulip sleet
hmm, cookiecutter is supposed to work on Windows, according to the doc. we might need to change something to make that happen
hidden geyser
The templates on that repo I tried to clone uses special characters that aren't valid for folder names on Windows
Perhaps cookiecutter has an alternative syntax that could be used
It's the pipes in the filenames in the adafruit cookiecutter, I think it's an easy fix
lone axle
I think windows does not like the {{ ... }} template directory name. I was not able to successfully clone and update the repo on windows 7. But I could from Debian VM.
hidden geyser
{{ }}i s fine as of latest win 10, but | is not
tulip sleet
I am able to create filenames with curly braces and pipes on Windows in the git bash shell
lone axle
Ah, I see
hidden geyser
No, it has a silent error
lone axle
The names look like jinja2 template strings to me
tulip sleet
hidden geyser
I'm doing the modifications and I'll have a pull up soon
@tulip sleet git clone nonetheless still fails in that shell
tulip sleet
yep, I did see that. But this private variables thing seems like a much more readable way of doing this than using horrible pathnames; thanks for finding that
i have to sleep, but thanks, looking forward to your PR or at least an issue. I did a lot of googling on this, and didn't find any similar mentions of such a problem
hidden geyser
Kinda weird, I can't be the only person to be cookiecutting on windows ๐ค
Good night!
{% if cookiecutter.library_prefix %}{{ cookiecutter.library_prefix | lower }}_{% endif %}{{ cookiecutter.library_name | lower }}.py also this file ๐
hidden geyser
@tulip sleet looks like it's more complicated in 1.7 than just using private vars, created an issue here
https://github.com/adafruit/cookiecutter-adafruit-circuitpython/issues/82
solar whale
@silver tapir @ionic elk I had the same error trying to build the neopixel PR (3232) for esp32s2 -- commented in the PR
solar whale
@tulip sleet @onyx hinge FYI -- I ran into a significant issue with my GCC10 build for grandcentral_m4 -- USB is not working -- REPL or CIRCUITPY ... I have not tried other M4 boards yet. will do so later today. CPX (M0) seems to work OK.
manic glacierBOT