You did! Thanks for the fun project 😁
#circuitpython-dev
proven garnet
thorny jay
So the result will be a bitmap to use in displayio?
proven garnet
Basically. I think right now it can give both the bitmap and palette
Up to user to do a tilegrid
jaunty juniper
no, I was not really here
idle owl
@proven garnet I have a question for you too 😄
jaunty juniper
ok yeah
proven garnet
Thanks!
orchid basinBOT
digital shoreBOT
idle owl
@tulip sleet Are you dropping an 8.0.0-alpha.0 before tomorrow as well? Or not yet?
I'm going to assume no so I can wrap up the newsletter. We'll include it next week if it ends up released.
tulip sleet
<@329766224093249548> Are you dropping an 8.0.0-alpha.0 before tomorrow as well?...
As soon as 7.3.0 final is up, yes, I can make an 8.0.0-alpha.1 (alpha.0 is already cast in stone, but not released) so i will be on circuitpython.org
but given the build backlog, it will be tonight
idle owl
I removed the "and its unstable version is foo" from the newsletter. I think it's fine to include it next week.
tulip sleet
I will let Anne know what is available if she wants to make last minute changes
idle owl
Ok sounds good.
slender iron
@tulip sleet do you know the format of .a archives?
slender iron
found it: https://en.wikipedia.org/wiki/Ar_(Unix)
jaunty juniper
I updated the frozen CPX PR with the release of the library, let's see if it passes CI
tulip sleet
Well, that is a pretty timely blog entry 🙂
thorny jay
Well, that is a pretty timely blog entry 🙂
Timely? 2021-02-28. 😉
tulip sleet
i am living in the past
stuck elbow
what is time anyways
idle owl
what is time anyways
Nothing.
orchid basinBOT
tulip sleet
pretty slow light there 🙂
ornate breach
maybe it was a millisecond
it's been at least 4 years since I was taught the definition of a second xD
apparently it's based on Cesium 133
Today, one second is defined as “9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the cesium 133 atom”
the more you know.
I also had it backwards, 1 meter is defined as 1/299792458 of a second light travel in a vacuum. lol
slender iron
local /home/tannewt/repos/circuitpython/ports/espressif/build-ai_thinker_esp32-c3s-2m/extmod/ulab/code/numpy/numpy.o:ulab_numpy_globals_table 367 31729
local /home/tannewt/repos/circuitpython/ports/espressif/build-ai_thinker_esp32-c3s-2m/extmod/ulab/code/numpy/numpy.o:mp_module_ulab_numpy_globals 368 31745
global esp_wifi_init_internal 745 46668
global esp_wifi_init 883 55760
global common_hal_wifi_init 1115 65026
local /home/tannewt/repos/circuitpython/ports/espressif/build-ai_thinker_esp32-c3s-2m/shared-bindings/wifi/__init__.o:wifi___init__ 1116 65046
local /home/tannewt/repos/circuitpython/ports/espressif/build-ai_thinker_esp32-c3s-2m/shared-bindings/wifi/__init__.o:wifi___init___obj 1117 65054
global mesh_nwk_task_main 657 68875
global esp_mesh_nwk_task_init 693 71291
local /home/tannewt/repos/circuitpython/ports/espressif/build-ai_thinker_esp32-c3s-2m/shared-bindings/wifi/__init__.o:wifi_module_globals_table 1379 76698
local /home/tannewt/repos/circuitpython/ports/espressif/build-ai_thinker_esp32-c3s-2m/shared-bindings/wifi/__init__.o:wifi_module_globals 1380 76714
global wifi_module 1381 76722
local /home/tannewt/repos/circuitpython/ports/espressif/build-ai_thinker_esp32-c3s-2m/py/objmodule.o:mp_builtin_module_table 4436 241850
esp_mesh is being brought in and is huge
nope, that's a lie
ornate breach
looks like it kind of tried
proven garnet
@thorny jay here's that's image packet branch for bluefruit connect https://github.com/tekktrik/Adafruit_CircuitPython_BluefruitConnect/tree/dev/add-image-packet
As far as I can tell the bluefruit is currently having issues picking up the image
Going to try the Arduino library to confirm the app is sending properly
thorny jay
As far as I can tell the bluefruit is currently having issues picking up the ima...
Thanks, I need to locate a Bluefruit with screen hardware. My CLUE screen is dead, I found my "TFT Gizmo" but I need to locate my "Circuit Playground Bluefruit". So you try to do this: https://learn.adafruit.com/wireless-image-transfer-with-circuit-playground-bluetooth-and-tft-gizmo but in CircuitPython. And this is more information on the protocol: https://learn.adafruit.com/bluefruit-le-connect/image-transfer . I wonder if it has been used by any other guide...
proven garnet
Yeah pretty much. I used the explanation of the protocol but I'm not able to actually "get" it on my bluefruit (I don't find the leading ! or anything) so something isn't working - almost certainly my code 😂
manic glacierBOT
tulip sleet
@slender iron I closed 7.3.0 and 7.x.x milestones, created a 7.3.x milestone, and moved all 7.x.x issues to 8.0.0 milestone for now. Could still triage, and some may deserve a backport to 7.3.x.
proven garnet
Oh @idle owl and @trim elm I meant to share with you the repository I made for the patch scripts I've used. Figured there might be something cool to glean if it was public: https://github.com/tekktrik/cpy-library-scripts
idle owl
Oh <@330227457296957440> and <@362023826105827340> I meant to share with you the...
Excellent! Thanks for sharing. I feel like we could make another directory on the Adabot repo for library scripts, or I could create a separate repo on Adafruit for them. Eva has a bunch as well. I'll see whether she thinks it makes more sense in Adabot or separate, and we'll see from there.
slender iron
sounds good @tulip sleet
proven garnet
Excellent! Thanks for sharing. I feel like we could make another directory on th...
Sounds good! 
idle owl
Sounds good! <:zoop:415996886575742977>
Ok, so here's the deal. You will need to clean up your script files if necessary, and document them very thoroughly. Once you've completed that (and you can PR it when you're ready so we can go over them together), you can add them to the directory-I-haven't-named-yet on the Adabot repo. Eva will do the same. I think it's important to share them where folks can find them, and since they either tie into Adabot, or do similar things, I feel like that's an excellent place to share them.
WRT documentation, I want it to be something where anyone could pick it up, understand what's going on, and basically be able to run it entirely based on the docs in the file.
slender iron
@tulip sleet have you built CP on your m1?
idle owl
<@329766224093249548> have you built CP on your m1?
I have.
slender iron
nice! limor is trying to get it going
idle owl
Not for ESP though. If that is what you're looking for.
slender iron
that is what limor is trying
hidden rain
Hi: Moving some code from a 16bit MSP430 to RP2040 and thought I would also go with CPy. My code sets many registers which have equivalents on RP2040. So my question is "how does one set a register"? Its obviously different since no pointers, etc.
idle owl
I never took the time because it wasn't on my official list of tasks, and it seemed like a bit of a rabbit hole.
So I thought Limor would be happier if I didn't spend the time.
slender iron
@hidden rain you'd be better off doing C for register level code
hidden rain
Yes if could do the entire app that way, however, the main part of the code will be CPy. Is there a way to mix and match, or use PIO, or create a little register getter/setter?
stuck elbow
a simple C module that just exposes some registers would be quick to write though
hidden rain
How would I "mix", C/C++ as a module? Do you have a link?
I could write the PIO assembly if it can takes arguments and return values.
idle owl
Did we decide on ignore-list for the replacement, or were we opting for allowlist and denylist?
@slender iron I feel like you were part of the discussion about this that I'm vaguely remembering. ^^
I'm not sure what the opposite of "ignore-list" would be anyway.
slender iron
I feel like we decided to go with worked best for the given situation
idle owl
Hmm. Alright. Thanks.
stuck elbow
I'm not sure what the opposite of "ignore-list" would be anyway.
watchlist ;-)
idle owl
Fair enough.
proven garnet
Ok, so here's the deal. You will need to clean up your script files if necessary...
Sounds good, I can do that. Should they be more of a historical "here's what I did as a basis" or "here are more tools to use"?
idle owl
Sounds good, I can do that. Should they be more of a historical "here's what I ...
Um.... Good question. Probably "here are more tools to use".
We'll want a README in the directory I think.
proven garnet
The difference being some scripts are definitely have hyperspecific purposes, but I could refactor the one that checks that state of the latest build to be more generally useful
idle owl
With maybe the names of each of the scripts and a quick description.
idle owl
The difference being some scripts are definitely have hyperspecific purposes, bu...
Right, I don't mind if they're hyperspecific. As long as they're documented well.
proven garnet
Sweeto! I can do that, just let me know what the repo ends up being and I can PR it
idle owl
It'll be Adabot but want a clear name for the directory.
So I'll get back to you as soon as I come up with some ideas. Or an idea.
proven garnet
idle owl
<:kgsHype:609985844664598544>
I don't even know where you come up with these emojis. 😆 I get what you mean based on context, but half the time I'd otherwise have no idea what you meant. I'm getting old.
proven garnet
I don't even know where you come up with these emojis. 😆 I get what you mean b...
The perks and curse of joining too many discord servers hahaha
slender iron
just posted a small javascript project to #help-with-projects for those who can make sense of it all
I want to use the treemap to see the code size of different directories of circuitpython
tulip sleet
@stuck elbow I can ping you with stuff if it would help, e.g. see x:54 above
stuck elbow
thanks, once I know there is something to watch for, I see it, it's just the long-time things that I miss
manic glacierBOT
manic glacierBOT
For some reason, I don't have the option to allow changes to pull requests:
manic glacierBOT
onyx hinge
pre-commit.ci web service has the apparently unavoidable behavior that it opens PRs to upgrade pre-commit to "the latest tag", without taking account of whether a tag is stable under python / semver standard practices. It's frustrating, and probably enough to make me turn pre-commit.ci back off for the two repos where I enabled it. fyi, since I'd softly floated the idea of switching to it for CircuitPython libs.. https://github.com/pre-commit-ci/issues/issues/124
pre-commit's author is extremely opinionated!
hidden rain
At the risk of asking a question that must have already been answered, Should I use asyncio to catch n incomming UART action that would otherwise be an interrupt?
This is for an RP2040
tulip sleet
At the risk of asking a question that must have already been answered, Should I ...
to poll, you would check .in_waiting. So if you can make a task that checks that when it's its turn to run, that would work. But it will not be pre-emptive, since asyncio is cooperative multitasking.
hidden rain
Thats cool. The app will decide 'when' to turn on the listener, so it would it does not need to be pre-emptive! Thanks.
rough folio
Found out the adafruit_ntp lib will quite happily sit in a timeout condition for hours- worth PRing a timeout for?
one of the things I was wondering is how you would log that it had timed out- standard py I'd just throw a logging statement in there but I don't think that's the route for cirpy
slender iron
Found out the adafruit_ntp lib will quite happily sit in a timeout condition for...
Yes please. I think there is a TimeoutError or maybe it's just OSError
rough folio
Nah, just raises OSError: [Errno 116] ETIMEDOUT
slender iron
that'll work
rough folio
Right now I've just got a format_traceback with some info so it doesn't just fail silently
there's also what to do if it times out for the first invoke
maybe having some list of fallback servers?
manic glacierBOT
rough folio
ie right now it's
def datetime
if time to refetch:
adjustment = get time from ntp
return time + adjustmenet```
roughly
if I change it to
def datetime
if time to refetch:
try:
adjustment = get time from ntp
except:
pass
return time + adjustmenet```
then the first time it's invoked it will be entirely wrong if it times out then
so it needs some fallback support
(but what if all the fallbacks fail too? Should it raise?)
orchid basinBOT
slender iron
the adafruit ntp server isn't working?
rough folio
the adafruit ntp server isn't working?
it's timing out every now and then yeah
slender iron
hrm, ya feel free to improve it
rough folio
I was wondering why my code was just sitting dead after a couple minutes and it's because of NTP
karmic oyster
I am having trouble with the circuit python GPS library. The problem is in the gps.update command. On line 549 in the _parse_gsa section of the function, I am getting an error that "bytes" object has no attribute "decode" I am trying to run the "simple test" example
My setup runs the gps_echotest example fine. It outputs NMEA sentences, so I don't think it's a problem with the GPS
slender iron
Ok, I made https://treemap.dev and it's example is CircuitPython file sizes
stuck elbow
clicking on the two links on the bottom does nothing, probably because they are covered by something?
slender iron
click the squares
stuck elbow
squares?
stuck elbow
there are no squares
slender iron
rectangles
stuck elbow
oh, so it's like a directory browser
slender iron
ya
slender iron
with the rectangles sized appropriately
stuck elbow
very nice
stuck elbow
interesting, so the compiler and the strings are the biggest part of the python itself
kinda makes sense
though I would expect the vm to be bigger than the compiler
manic glacierBOT
slender iron
I think it'd be cool to have the # in the url be the root
easier to share links then
though it doesn't have the file itself...
lone axle
I think it'd be cool to have the # in the url be the root
you mean a URL something like:
https://treemap.dev/#/home/tannewt/repos/circuitpython/ports
And it will load directly to that place in the file structure?
slender iron
right
but the user would still need to provide the file which is weird
back button could work then though
lone axle
if it had a server side it could keep a copy of the JSON so different users could view the same thing. Or if the circuitpython repo is the primary intended files it could just clone it's own copy of the repo.
slender iron
it's meant to be generic and is just hosted statically with gh pages
lone axle
This is a neat visualizer. I didn't quite grok what you were talking about before I saw the example.
tulip sleet
i have used tools like that a lot to figure out what's using up all my disk space (back when I didn't have giant disks)
lone axle
For circuitpython specifically, maybe actions on the core repo could generate the JSON file and this tool could grab it from there?
manic glacierBOT
idle owl
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is happening 24 HOURS LATER on Tuesday, May 31, 2022 at 11am Pacific / 2pm Eastern here on Discord. Everyone is encouraged to attend! Please add your hug reports and status updates even if you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/13ESjBlnO5LdgZWdyrCDvklIvtYcsLqgUnyzM559at9A/edit?usp=sharing
hidden oxide
Ah, OK. That's a better time for me anyway
lone sandalBOT
manic glacierBOT
ornate breach
Here is the notes document for next Monday’s CircuitPython Weekly meeting. It is...
Is there like a joke going around about May 31st that i'm missing? lol...
I've seen it twice today
no wait
my months are off
manic glacierBOT
proven garnet
.pylintrc is patched! Here are the failures/errors: https://gist.github.com/tekktrik/d8606ef897da40d039bd9064bd5127ad
warm flame
I'm trying to solve the bootloop problem from my board. Going to add an issue on it
Tried with 2 different devices same problem
warm flame
Go figure I flash T-01C3 on the board and everything is working
I flash t-oi plus and bam bootloop
orchid basinBOT
warm flame
I litterally just added pins to the t-01C3
but still nothing
last build for today ...
orchid basinBOT
tidal kiln
tulip sleet
<@329766224093249548> <https://github.com/jgarff/rpi_ws281x/issues/326#issuecomm...
ok, great, so gpio and spi groups
tidal kiln
if use specific pin(s) it seems
tulip sleet
hmm, strange restriction. I'll research this some more, thanks, unless you want to stay on the trail.
tidal kiln
@gilded cradle hey. have you ever gone down the "how to run neopixels without sudo" rabbit hole?
gilded cradle
<@478616919721771019> hey. have you ever gone down the "how to run neopixels wit...
You mean on Raspberry Pi? I don't recall doing so, but perhaps I did a while back.
tidal kiln
yep. on pi.
gilded cradle
yep. on pi.
I think I looked into it for I2C and SPI but didn't get very far. It was more like peeking into the rabbit hole.
tidal kiln
ok. seems like we've all taken a peek 🙂
gilded cradle
🙂
tidal kiln
@tulip sleet if you're willing. i think you'd be better suited. could get into some lower level stuff you'd be a better ninja at.
jaunty juniper
I2C I was able to use without sudo by adding pi to the I2C group or something
tulip sleet
yes, that seems straightforward, it's neopixels in particular
jaunty juniper
but neopixel uses /dev/mem which is kernel protected or something
tidal kiln
the neopixel lib can use different plumbings under the hood. spi being one of them.
yah, it seems any attempt to use /dev/mem is where it becomes a bigger issue. not as simple as adding a group permission.
idle owl
@tidal kiln Ready for a shallow dive into Yet Another Schematic?
tidal kiln
sure
idle owl
I do not see a pull-up on A0. Am I missing something or is that correct?
I'm trying to work from as much of the pinouts page for the ESP32-S2 as possible, but obvs there are differences.
tidal kiln
do you have eagle installed?
idle owl
Ok I see some things highlighted
tidal kiln
ok. yah. anywhere works.
now can look around at all the highlighted wires and see if there are any pull ups
idle owl
It's so barely obvious on the schematic what's lit up. 😕 The color change isn't very ... different.
I see it on the board though, which is easier.
tidal kiln
yah...it's not a very contrasty color 😦
idle owl
Looks like a direct connection there, although it also looks like a trace is missing. Might have turned off a layer I shouldn't have.
No.... there's a layer I've never seen before that has the trace in it. Is this a multi-layer board?
It's a direct connection from the pad to the chip on the board.
So I guess no pullups.
@tidal kiln Thanks. I'll remember to do the eyeball-clicky-thing eventually before asking you. 😄
idle owl
Not sure how I've managed to get this far and never deal with a multi-layer board, or get this far having somehow missed the extra layers entirely when working with previous multi-layer boards. 😄
tidal kiln
i think multilayer is only used if necessary. most boards are 2 layer.
tidal kiln
i wish the line thickness would also increase
idle owl
Yeah agreed.
This is at least a huge improvement.
I can see things when zoomed out now
tidal kiln
that trick is handy, because it's possible to tie in something like a pull up resistor and, on the schematic, tuck it away in some odd corner that is easy to overlook with just visual scanning.
idle owl
I went to Options>Set... and checked the High Contrast box in the "Highlight" section of the window, and dragged it around until I was happy with the visual effect.
Sorry if you already knew what I did. Figured I'd share in case you weren't.
tidal kiln
huh. i had that on. but never noticed the hue slider.
can use that to dial it in even more
digital shoreBOT
slender iron
@warm flame is there a UART debug output on the board? It will tell you what the restart reason is
orchid basinBOT
warm flame
<@385537177838288898> is there a UART debug output on the board? It will tell yo...
Will check tomorrow
orchid basinBOT
slender iron
@tulip sleet micropython has C nina support: https://github.com/micropython/micropython/blob/master/drivers/ninaw10/nina_wifi_drv.h
orchid basinBOT
warm flame
<@385537177838288898> is there a UART debug output on the board? It will tell yo...
the uart pins are attached to the usb chip too. can I check via USB ?
slender iron
what esp is on the board?
warm flame
esp32-c3
slender iron
ah, so maybe the uart is already used for circuitpython
warm flame
but when I flash the t-01c3 build it works
warm flame
the t-oi-plus build don't
slender iron
you could try loading a debug build and seeing if you get any extra output
slender iron
ya, add DEBUG=1 to the make command
warm flame
ok
warm flame
ya, add `DEBUG=1` to the make command
on the board image they say gpio20 and gpio21 are UART. But when I look at the schematic they are in esp32-c3 pin 27 28
warm flame
21 for tx and 20 for rx.
slender iron
there are usually multiple uarts that are broken out
warm flame
ok
slender iron
ya, those look like the default
warm flame
slender iron
it is possible that the gpio number isn't the same as the package pin number
warm flame
ok
slender iron
I haven't done a ton with the C3
warm flame
Executing action: all (aliases: build)
Running ninja in directory /home/delarn/Documents/dev/circuitpython/ports/espressif/build
Executing "ninja all"...
[1/6] Performing build step for 'bootloader'
[1/1] cd /home/delarn/Documents/dev/circuitpython/ports/espressif/build/bootloader/esp-idf/esptool_py && /home/delarn/.espressif/python_env/idf4.4_py3.9_env/bin/python /home/delarn/Documents/dev/circuitpython/ports/espressif/esp-idf/components/partition_table/check_sizes.py --offset 0x8000 bootloader 0x1000 /home/delarn/Documents/dev/circuitpython/ports/espressif/build/bootloader/bootloader.bin
Bootloader binary size 0x62c0 bytes. 0xd40 bytes (12%) free.
[2/4] Linking CXX executable circuitpython.elf
FAILED: circuitpython.elf
maybe it doesn't help
ok how do I fix this idf.py build error
@slender iron the failed: circuitpython.elf is making me nervoous
I did a idf.py fullclean
idf.py all
just after
[698/700] Linking CXX executable circuitpython.elf
FAILED: circuitpython.elf
again
manic glacierBOT
blissful pollen
I swear whenever I take a break from any core work and then come back I mess some upgrade up and break everything in my linux environment. Lesson learned, don't stop working
warm flame
<@385537177838288898> you shouldn't use idf to build
after performing the step : source esp-idf/export.sh the script ask me to perform idf.py build
tulip sleet
after performing the step : ```source esp-idf/export.sh ``` the script ask me to...
sorry, we have no control over that message. I can add warnings to the Building CircuitPython guide
warm flame
sorry, we have no control over that message. I can add warnings to the Building ...
Thanks
What's the difference between qio and dio ?
tulip sleet
for flash chips, 4-bits-wide data is sent over qio. dio is for "dual": 2 bits wide
tulip sleet
ok it depend on the flash chips
You can often use the same flash chips in several ways: single bit SPI, dual SPI, and Quad SPI
warm flame
@tulip sleet Nice to know
Now All I need is that the build work
@tulip sleet When I flash the new firmware do I need to flash the bootloader too ?
tulip sleet
if you flash the .bin (and erase the chip in advance to remove any partitioing), then you don't need the UF2 bootloader. You can skip that for now to remove an extra possibility of problems.
do you have other ESP32-Sn boards?
warm flame
ESP32-C3 don't have uf2
tulip sleet
oh, sorry, I forgot which board you were working on
so you are just using the ROM bootloader
there is no bootloader to flash
warm flame
@tulip sleet If I want to check if the build was succesful via debian terminal, how do I do that ?
tulip sleet
I think the repl will show up on the tty port the board presents, not sure, I haven't tried it myself. The C3 support is very early.
warm flame
how do I access the port in terminal ?
tulip sleet
let me try it on a QT Py C3 and I'll get back to you. It will be /dev/ttyUSB0 or ttyACM0 or similar
warm flame
it's on ttyUSB0
tulip sleet
you need a terminal program. We recommend tio, so install that, then just do tio /dev/ttyUSB0. ctrl-T is the escape character in tio to send it commands
don't leave tio connected when you are trying to upload the .bin; it will interfere with the upload
warm flame
ok
I am on tio
:~$ tio /dev/ttyUSB0
[tio 11:56:43] tio v1.32
[tio 11:56:43] Press ctrl-t q to quit
[tio 11:56:43] Connected
can you type at it?
warm flame
and cursor is flashing
nope
on a reset : ```
ESP-ROM:esp32c3-api1-20210207
Build:Feb 7 2021
rst:0x1 (POWERON),boot:0xd (SPI_FAST_FLASH_BOOT)
SPIWP:0xee
mode:DIO, clock div:2
load:0x3fcd6100,len:0x16b8
load:0x403ce000,len:0x930
load:0x403d0000,len:0x2d28
entry 0x403ce000
I (31) boot: ESP-IDF 944c01eef 2nd stage bootloader
I (31) boot: compile time 11:25:39
I (31) boot: chip revision: 3
I (33) boot.esp32c3: SPI Speed : 40MHz
I (38) boot.esp32c3: SPI Mode : DIO
I (43) boot.esp32c3: SPI Flash Size : 4MB
I (47) boot: Enabling RNG early entropy source...
I (53) boot: Partition Table:
I (56) boot: ## Label Usage Type ST Offset Length
I (64) boot: 0 nvs WiFi data 01 02 00009000 00005000
I (71) boot: 1 app factory app 00 00 00010000 00200000
I (78) boot: 2 user_fs Unknown data 01 81 00210000 001f0000
I (86) boot: End of partition table
I (90) boot_comm: chip revision: 3, min. application chip revision: 0
I (97) esp_image: segment 0: paddr=00010020 vaddr=3c050020 size=1edc0h (126400) map
I (133) esp_image: segment 1: paddr=0002ede8 vaddr=3fc88e00 size=01230h ( 4656) load
I (134) esp_image: segment 2: paddr=00030020 vaddr=42000020 size=43ac4h (277188) map
I (198) esp_image: segment 3: paddr=00073aec vaddr=3fc8a030 size=004d8h ( 1240) load
I (199) esp_image: segment 4: paddr=00073fcc vaddr=40380000 size=08d40h ( 36160) load
I (213) esp_image: segment 5: paddr=0007cd14 vaddr=50000010 size=00010h ( 16) load
I (216) boot: Loaded app from partition at offset 0x10000
I (219) boot: Disabling RNG early entropy source...
I (235) cpu_start: Pro cpu up.
I (244) cpu_start: Pro cpu start user code
I (244) cpu_start: cpu freq: 160000000
I (244) cpu_start: Application information:
I (247) cpu_start: Project name: circuitpython
I (252) cpu_start: App version: 8.0.0-alpha.0-81-g8dcbd3ab4-dir
I (259) cpu_start: Compile time: Ma 27 2022 11:25:29
I (265) cpu_start: ELF file SHA256: 47ce06cd16a1b45a...
I (271) cpu_start: ESP-IDF: 944c01eef
I (276) heap_init: Initializing. RAM available for dynamic allocation:
I (283) heap_init: At 3FC8CB50 len 000334B0 (205 KiB): DRAM
I (290) heap_init: At 3FCC0000 len 0001F060 (124 KiB): STACK/DRAM
I (296) heap_init: At 50000020 len 00001FE0 (7 KiB): RTCRAM
I (303) spi_flash: detected chip: winbond
I (308) spi_flash: flash io: dio
I (316) sleep: Configure to isolate all GPIO pins in sleep state
I (318) sleep: Enable automatic switching of GPIO sleep configuration
I (326) coexist: coexist rom version 9387209
I (330) cpu_start: Starting scheduler.
Hello, world!
can't type
tulip sleet
on a reset : ```
ESP-ROM:esp32c3-api1-20210207
Build:Feb 7 2021
rst:0x1 (POWERO...
do you see a /dev/ttyACM0?
warm flame
will check in a min ... os update
tulip sleet
it doesn't look like you've loaded CircuitPython
oh, yes, I see it there
try this build, which is showing up on ACM0 on my QT Py S3: https://circuitpython.org/board/adafruit_qtpy_esp32c3/
the pins will not be right, but it should talk to you
warm flame
let's try it
I tried to flash wit web esptool
no good
flashing in terminal
same problem
@tulip sleet Just flashed the t-01C3 build and it works
warm flame
yes
tulip sleet
so t01 works but not tOI ? (gack, their model numbers are problematic)
warm flame
But I need to access those GPIO that are not avalaible in T-01C3 ...
exactly
T-01C3 and t-oiPlus
tulip sleet
does the "Absolute Newest" t01 work on the t-oi?
I looked at the build differences and they are minor
warm flame
yes
tulip sleet
where is the schematic for the t-oi?
tulip sleet
i am at a loss, so the prebuilt t-OI build that we have does not work, you are saying? Did you try reloading it again?
warm flame
I'm remaking the build
with default_i2c_bus and default_uart_bus
it may not like the circuitpy bus
warm flame
@tulip sleet @slender iron in my build folder I got 2 firmware.bin file. one called circuitpython-firmware.bin and the other firmware.bin
tulip sleet
firmware.bin is what you want - it's combined with other stuff to make a ful .bin
tulip sleet
if you build the 01 firmware yourself, does your own build work?
So I would stay start with the 01, and start making the OI changes, one by one, but I think you are doing that
tulip sleet
it could be the never reset stuff - that is my highest suspicion, but just because it's the most complicated change
sour spoke
Has something changed about the mu-editor. I haven't used it in a few months, but IIRC I could open the serial window, do a ctrl-c, and get REPL. Now I get a flashing cursor in the serial window, but nothing else. I am using 1.0.3 with a pybadge. I know the connection work because I can save/modify code.py
sour spoke
yup
stuck elbow
make sure you have the rights to /dev/ttyACM0
you probably need to add yourself to the dialout group
sour spoke
I can get a repl with gtkterm just fine. permission should be the same
stuck elbow
do you have the gtkterm open right now? because only one program can open the serial connection at a time
sour spoke
There is a gtkterm running but it is connected to /dev/ttyUSB0. If I start a gtkterm, it will connect to /dev/ttyACM0 and display the REPL. If I kill it and restart mu-editor, the behavior is the same.
2022-05-27 18:53:48,250 - mu.modes.base:227(find_device) INFO: Found device on port: ttyACM1
2022-05-27 18:53:48,251 - mu.modes.base:228(find_device) INFO: Serial number: e00fce6898dcda8f48f21674
2022-05-27 18:53:48,257 - mu.modes.base:278(add_repl) INFO: Started REPL on port: /dev/ttyACM1
2022-05-27 18:53:48,257 - mu.modes.base:259(toggle_repl) INFO: Toggle REPL on.
It looks like mu-editor thinks it is starting REPL.
stuck elbow
but ACM1
sour spoke
just noticed that.
sour spoke
the pybadge is on acm0, I can verify that with gtkterm. Let me look at my logs and see what got ACM1.
You were rightish! The Boron, which I normally only think of in terms of /dev/ttyUSB0 claimed ACM1 which is what mu-editor is glomming onto. Unplug the boron and things work fine. Thank you!
May 24 21:52:46 alton kernel: usb 1-4.4.2.4.1: Product: Boron CDC Mode
May 24 21:52:46 alton kernel: usb 1-4.4.2.4.1: Manufacturer: Particle
May 24 21:52:46 alton kernel: usb 1-4.4.2.4.1: SerialNumber: e00fce6898dcda8f48f21674
May 24 21:52:46 alton kernel: cdc_acm 1-4.4.2.4.1:1.0: ttyACM1: USB ACM device
One of the rare times a power-only USB will be useful!
manic glacierBOT
jaunty juniper
You were rightish! The Boron, which I normally only think of in terms of /dev/tt...
that seems to be a bug, Mu normally should pick a port whose name starts with "Circuitpython CDC ", and if there are multiple it should show a menu in the bottom right (with possibly useless "circuitpython board" tags for each but still)
manic glacierBOT
warm flame
I'm reinstalling the whole toolchain
warm flame
@tulip sleet I started by changing the names
slender iron
@shell bolt do you compile your code with clang?
warm flame
in mpconfigs and sdkconfig
tulip sleet
and adding more pins should be fine
did you reinstall the toolchain with esp-idf/install.sh ? I assume so
shell bolt
<@488724449277902859> do you compile your code with clang?
For which target ?:)
slender iron
For which target ?:)
I was trying on m0+ (samd21)
shell bolt
Ah, clang support for arm isn't great, I've mainly used it with Risc-V. There's this POC from Arm, but I haven't tried it: https://github.com/ARM-software/LLVM-embedded-toolchain-for-Arm
slender iron
Ah, clang support for arm isn't great, I've mainly used it with Risc-V. There's ...
do you know how different that is compared to normal clang?
warm flame
@tulip sleet I've added the SDA SCL in the pins but not in the mpconfig
now adding the battery stuff
do I have to do a make clean??
jaunty juniper
you can do make BOARD=... clean, it can be necessary if there are new strings (which pin names are)
shell bolt
do you know how different that is compared to normal clang?
I think it's mostly about the libraries you need to build against, like the C standard library. Here's a good article that may be helpful: https://interrupt.memfault.com/blog/arm-cortexm-with-llvm-clang.
slender iron
Yup, I followed that. Got it going but it was too large. No size optimized lto
manic glacierBOT
manic glacierBOT
wraith crow
I'm playing with a Sparkfun SAMD51 Thing Plus which I thought had 4 MB (megabytes) of SPI flash but the CIRCUITPY drive only has about 500K available space. After digging through the CP flash code for a couple days, I finally noticed that the external SPI flash on the Sparkfun board is 4 Mb (megabits) or 512 KB in size.
So this now changes my confusion about this board. Now it appears to have too much Flash. There is 1 MB (megabyte I think 😋 ) internal flash on the ATSAMD51J20 microprocessor but the mpconfigboard.mk file in ports/atmel-samd/boards/sparkfun_samd51_thing_plus does not have the INTERNAL_FLASH_FILESYSTEM flag enabled so I wouldn't expect CP to be using the flash on the controller. I think I saw somewhere that under some conditions CP will cut the available flash in half for the file system and if that's what it's done then the 500K available makes sense.
I guess I have two questions:
-
Did I miss the setting of the INTERNAL_FLASH_FILESYSTEM flag for this build somewhere or is the internal flash selected by some other means?
-
and from a what the heck perspective, why would Sparkfun put a 512KB SPI flash chip on a board with a microcontroller that has 1MB flash onboard?
jaunty juniper
I'm playing with a Sparkfun SAMD51 Thing Plus which I thought had 4 MB (megabyte...
on chips with internal flash the entire external flash is dedicated to the CIRCUITPY drive, so 500kB for the drive seems to match ?
RP2040 and ESP builds need to share the flash with the CP binary, so it's a different situation
wraith crow
on chips with internal flash the entire external flash is dedicated to the CIRCU...
Ah, never tracked down the logic that chooses the external flash if there is internal. Thanks.
jaunty juniper
as it turns out the thing plus build uses less than half the internal flash, but that just means the space is left unused, it could be filled with more builtin modules in the future
462472 bytes used, 561528 bytes free in flash firmware space out of 1024000 bytes (1000.0kB).
Yeah, it's sort of ironic that the file system would actually have move space if CP ignored the external flash chip
Just for kicks, maybe I'll take another whack at trying to get a custom build of CP to ignore the external flash, my first attempt at that didn't go to well 😆
wraith crow
Actually, if CP cuts internal flash in half, I guess the available space would be the same and maybe my earlier attempt did work and I just didn't notice because the file system is using 512KB either way. Oh well... Thanks I think I can finally put this down now.
jaunty juniper
Actually, if CP cuts internal flash in half, I guess the available space would b...
CP doesn't cut anything in half, I don't know where you got that, it just takes the space it needs
jaunty juniper
more accurately the space used by the drive when using internal flash is defined somewhere, and CP uses the rest
wraith crow
CP doesn't cut anything in half, I don't know where you got that, it just takes ...
This is what led me to that conclusion (in circuitpython/ports/atmel-samd/mpconfigport.h)
// If CIRCUITPY is internal, use half of flash for it.
#if INTERNAL_FLASH_FILESYSTEM
#ifndef CIRCUITPY_INTERNAL_FLASH_FILESYSTEM_SIZE
#define CIRCUITPY_INTERNAL_FLASH_FILESYSTEM_SIZE (FLASH_SIZE / 2)
#endif
#else
#define CIRCUITPY_INTERNAL_FLASH_FILESYSTEM_SIZE (0)
#endif
#endif // SAM_D5X_E5X
yeah ok that's the default for samd51
onyx hinge
https://github.com/agronholm/exceptiongroup just became aware of this, not sure if it helps with adopting taskgroups in CP
manic glacierBOT
manic glacierBOT
Do you not want to make the Grove connector be board.I2C() ?
Do you not want to make the Grove connector be
board.I2C()?
would be nice, but I had such an hard time making the board work. I'll plan for a future dev.
manic glacierBOT
Hi there,
I can't use I2C on the Waveshare. The board should be able to use several sets of GPIOs as I2C pins, but I can't get any of them to work:
...and I don't see any of them defined as such in pins.c.
thorny jay
Maybe this is a message for danh (no tag as this is low priority and a day off),
I would like to use keypad to simultaneously scan a KeyMatrix, and an independent Key on a GPIO. (this is the case for C64 keyboard that have a 8x8 matrix plus the RESTORE key, this is also the case for the BBQ20KBD and the to right button that is wired outside the matrix)
The idea would be for those to share an Event queue so that event stay in order and I don't need to merge two queue in userland.
I have checked those two documentations, but I don't see the way to do what I want:
- https://docs.circuitpython.org/en/latest/shared-bindings/keypad/index.html
- https://learn.adafruit.com/key-pad-matrix-scanning-in-circuitpython/overview
The kind of API I expect is to be able to add arbitrary number of matrix, GPIO and shift register, and have CP scan that and collect the event. Where should I publish that as an issue?
PS: For Kattni, maybe you could hint about the existance of keypad build-in in this guide: https://learn.adafruit.com/matrix-keypad maybe that learnguide is outdated and use old technique to scan keys?
slender iron
<@&356864093652516868> Just a reminder that our meeting is in just over 24 hours due to a US holiday. Talk with you tomorrow!
tulip sleet
Maybe this is a message for danh (no tag as this is low priority and a day off),...
You can add an issue to circuitpython. Events now have time stamps, which might help somewhat in what you want
jaunty juniper
I would do that with a python library, the events have a timestamp so having a next_event() that gets the next event of each queue, returns the oldest and caches the other should be simple enough
proven garnet
I'm starting to get errors on my own libraries using Sphinx that language = None isn't good with Sphinx 5.0.0, and my guess is it'll be an issue for the libraries too. @idle owl, just a heads up that may need to be a patch that will need to be applied for the documentation to be published properly to RTD.
The good news is spent the week developing my patch tools so if we want to move to Sphinx 5.0.0 I can fix it pretty quickly. Or pin back to Sphinx 4.0.0
thorny jay
Ok, that's an option. Right now I erronously added that GPIO to the matrix as a row(?) and a press trigger one event for each column(?) so I can just filter out and not use n-1 of them. But I felt that something was missing. I did not check KMK yet, not sure if they handle that.
proven garnet
The fix is easy, just changing to language = "en" in docs/conf.py
Yeah, I re-ran the CI on a library that previously had no issues, and it popped out there as well :/
I guess this is a problem for everywhere where Sphinx is pinned as sphinx>=4.0.0 and language is explicitly equal to None so heads up y'all!
proven garnet
Okay, fix is easy for the libraries via Adabot patching, so here's the PR for the fix! I can have this done today as well. https://github.com/adafruit/adabot/pull/266
idle owl
@proven garnet Thank you so much for catching this and the super quick fix!
proven garnet
No problem! Always easy when it's a holiday 
proven garnet
Patched! Took longer because my laptop decided to go wonky during the patch so it was harder to find failures, but I managed to get a script to do it. Between adabot and manual patching problem should be settled!
worldly escarp
Hello, @cold wharf and I just joined the server and wanted to introduce ourselves. Alex and I make cat-shaped educational boards with a focus on learning cybersecurity and ethical hacking as HakCat.com.
We’ve been creating CircuitPython labs that run on our ESP32s2 based Nugget boards with tutorial videos on the Hak5 YouTube channel. I’ve had some great conversations with CircuitPython developers, and I was encouraged to join this server while I work on new labs. If anyone mentions one of our boards, we’re happy to answer questions. I hope we can contribute here!
blissful pollen
Hello, <@460270442058481665> and I just joined the server and wanted to introduc...
Hey welcome! Quiet in here today cause of the holiday. I'll make a point to check out your boards later!
proven garnet
A few of the libraries seem to have new issues probably resulting from the upgrade to Sphinx 5.0.0, looking into those now
cold wharf
We’ve been creating CircuitPython labs that run on our ESP32s2 based Nugget boar...
Heyo! Im Alex, the main Nugget developer. We look forward to sharing our contributions with this community!
tulip sleet
@onyx hinge thanks for the astroid fix! I wonder why it does not seem to be actively maintained.
onyx hinge
Yeah the last release was some time ago!
manic glacierBOT
This applies the equivalent of https://github.com/readthedocs/sphinx-autoapi/pull/340/files by "monkey" patching, that is, using the dynamic nature of Python to replace the buggy function _format_args with the fixed version.
onyx hinge
Besides working on that sphinx bug today I also toured Königstein near Dresden Germany. Hope y'all are well.
ornate breach
What a view
tulip sleet
<@&356864093652516868> Weekly meeting is today at 2pm Eastern / 11am Pacific / 1800 UTC. See the pinned messages for a link to the notes doc. See you then if you can make it!
onyx hinge
@tulip sleet it would be nice if some of this stuff were better signposted!
fathom sand
@tulip sleet would you please add me to the "circuitpythonistas"
thorny jay
Sorry, skipping the meeting, no notes, not even listening LIVE. I have not done much anyway, despite an extra day.
proven garnet
I don't have access to main for neopixel so if anyone who does has a minute, please take a look at the PR patch I submitted for it. It's the same one the rest got yesterday, for what it's worth, nothing fancy.
slender iron
idle owl
I don't have access to `main` for `neopixel` so if anyone who does has a minute,...
I'll get to it today.
We had to lock it down after a broken PR got merged. So we limit who can touch it.
Since it's such a widely used library.
tulip sleet
<@&356864093652516868> Weekly meeting starting now in CircuitPython audio channel
turbid radish
Yes, please send interesting tidbits
random junco
He fixed the linter on the awesome list, too!
turbid radish
Thanks Melissa!!
slender iron
🎉
hidden oxide
Hi all - just wanted to pop in and say hi and thanks all!!
proven garnet
Thanks!
ember iris
Thanks all! Have a great week!
random junco
@proven garnet Nice work, almost every documentation issue I look at, you're already on it. 🙂
tulip sleet
Here is the notes document for next Monday’s 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 you’ll be attending the meeting - it’s super helpful! If you are unable to attend but would still like to include updates, feel free to include them in the notes and we’ll read them off during the meeting. Hope to see you there! <@&356864093652516868> https://docs.google.com/document/d/1KXxPd9X8w8UI8Cg-YgWBTlw8rMMlCBpzM2-osVtJDrg/edit?usp=sharing
idle owl
@errant grail Did your NAU module that you added to the bundle end up on PyPI?
errant grail
<@339922377746022402> Did your NAU module that you added to the bundle end up on...
Not that I know of.
idle owl
@errant grail Would you be willing to post it to PyPI?
proven garnet
<@135415127879974912> Nice work, almost every documentation issue I look at, you...
Thanks! Happened to notice on my own stuff because I basically copy paste all my documentation infrastructure from the libraries because work smarter not harder 😂
It's also about all I can currently contribute to the core haha
random junco
me too! that's why I was looking at docs. 🙂
errant grail
<@339922377746022402> Would you be willing to post it to PyPI?
Not inclined to. Navigating the community bundle submission process was too far out of my wheelhouse. However, in a discussion with Ladyada, I gave her permission to use the driver code as needed without attribution.
idle owl
Not inclined to. Navigating the community bundle submission process was too far ...
Thank you for submitting it! That's entirely fair. She requested that I ask you. I'll see what she wants to do.
errant grail
Thank you for submitting it! That's entirely fair. She requested that I ask you....
Super! I’d love to remove it from the bundle at some point so that I can declutter my personal repo.
idle owl
Super! I’d love to remove it from the bundle at some point so that I can declutt...
Can't say that will happen anytime soon. If you're really keen on us taking it over, I can talk to her and see if she wants to take it on "officially". Also, no worries on the PyPI thing.
errant grail
Can't say that will happen anytime soon. If you're really keen on us taking it o...
Yes, I’m keen on it. Thanks.
slender iron
@stuck elbow I'm thinking about CP's built in font
what do you think about one that is 6x12 and then we could add 12x12 emojis
stuck elbow
I can design anything as long as it's at least 3x5 (not counting the spacing)
slender iron
I want to make the blinka in the corner a snake emoji
stuck elbow
6x12 seems huge to me
slender iron
its 6x14 now
slender iron
ya, that's the problem if I add emoji support
it'll be harder to have different font sizes
stuck elbow
I think I'm not in the target audience for that feature
slender iron
ya, we could have the top status bar without the snake emoji
stuck elbow
by the way, I've been meaning to ask, are the escape sequences for the terminalio documented anywhere?
I wanted to make a menu in text mode
stuck elbow
I tried a couple vt100 commands and they didn't work
slender iron
here is my status bar test
I was going to implement what I need there
commands are implemented here: https://github.com/adafruit/circuitpython/blob/main/shared-module/terminalio/Terminal.c
stuck elbow
I looked at it, but it's a bit convoluted
slender iron
ya, it is
stuck elbow
so you will need an escape code for setting the scroll area now?
slender iron
yup
and saving and restoring cursor location
a status bar seems like the best way to show wifi, ble, usb and usb host status
stuck elbow
if you used a separate terminal tilegrid, you could use different fonts for them
slender iron
ya, at a flash cost though
stuck elbow
even different text orientation
I assume this is for boards like the esp32 ones that have plenty of flash
stuck elbow
I really don't see a status bar on a 128x64 display
especially if it's 12 pixels high
with the 4x6 font it just fits most exceptions in the 64 pixels
slender iron
if the screen is high res then it'd be hard to read though
we could just drop the 12x12 emoji if the font is smaller
blinka is 16x16 atm
stuck elbow
I think I had smaller blinkas somewhere
the storage for emojis is also going to eat flash
I assume they won't be stored in the font?
slender iron
ya, I'd only store the snake blinka on small builds
I think they'd need to be separate to be more colors
stuck elbow
if you want a custom font, you could have it use more colors too. In Stage I use a 4-color font, so I get to anti-alias the letters
you would then just need custom palette for the emoji
slender iron
custom font could work for english but then other languages would be left out
stuck elbow
why, you can have diacritics in a custom font
slender iron
it's just a lot to add
stuck elbow
mostly copy-paste
slender iron
ya, true
I can't imagine a 4x6 font on this esp display
6x12 is pretty small already
stuck elbow
right, I don't think one font will work for everything
also, 4 colors means it's 2x bigger in the memory
I would say we need a small font, possibly without emojis, for small screens like the OLEDs, and a big font for the larger displays
slender iron
could we antialias a 6x12 font down?
stuck elbow
scaling down doesn't work at those sizes very well, at the very least you would need pixel hinting, most fonts require manual touch up
do we need cirillics and greek too?
or just latin with diacritics?
stuck elbow
if you want to generate a font, best take a ttf font with good hinting and use that
slender iron
I'll keep using terminus
just realized I was making it 6x14 due to a misunderstanding of offset
@tulip sleet could you pull the language download stats when you have a chance?
tulip sleet
<@329766224093249548> could you pull the language download stats when you have a...
sure, do you want for all time or since 7.0.0 or something?
slender iron
something recent-ish
tulip sleet
SELECT language, count(*) FROM "default"."adafruit_circuit_python_downloads_details" where board <> '' and language <> '' and version like ('7.%')group by language order by count(*) desc
I am not sure about the en_us (as opposed to en_US) ones
maybe it's just that case gets ignored on fetches
slender iron
do we have terminal enabled for ja?
tulip sleet
SELECT lower(language), count(*) FROM "default"."adafruit_circuit_python_downloads_details" where board <> '' and language <> '' and version like ('7.%')group by lower(language) order by count(*) desc
looking...
This is only in the atmel-samde Makefile:
# We don't have room for the fonts for terminalio for ja and ko
# so turn off terminalio, and if it's off and displayio is on,
# force a clean build.
# Note that we cannot test $(CIRCUITPY_DISPLAYIO) directly with an
# ifeq, because it's not set yet.
ifeq ($(TRANSLATION), ja)
CIRCUITPY_TERMINALIO = 0
RELEASE_NEEDS_CLEAN_BUILD = $(CIRCUITPY_DISPLAYIO)
endif
ifeq ($(TRANSLATION), ko)
CIRCUITPY_TERMINALIO = 0
RELEASE_NEEDS_CLEAN_BUILD = $(CIRCUITPY_DISPLAYIO)
endif
and only off for samd21
tulip sleet
i would like to understand why the ru build is so big. Sure, the UTF-8 chars are mostly double-wide, but they are on the same code page, so the first byte would be the same. So why doesn't the huffman handle this succinctly, with the code page encoding being very narrow?
the ru translation is only 20-25% done, or maybe even less
slender iron
ja does have terminal but the ja chars are ignored
tulip sleet
so nothing in the font?
slender iron
yup
tulip sleet
maybe ko was/is in the font??
I don't know why I would have written that. I think it was me that wrote that comment.
slender iron
ya, maybe ko is
can you see what the most popular boards for ja are?
rp2040 is my guess
tulip sleet
let me do better than that
slender iron
tulip sleet
SELECT lower(language), board, count(*) FROM "default"."adafruit_circuit_python_downloads_details" where board <> '' and language <> '' and version like ('7.%')group by lower(language), board order by lower(language), count(*) desc, board
slender iron
my font bitmap data is 3552 uint32_t long now
tulip sleet
14kB, includes ja and ko?, or just ja?
you mean now after shrinking, or this is before any work on it?
slender iron
thats just switching to ja unifont and running the regular gen display resources
I'm thinking about supporting double wide characters so that blinka can be an emoji
tulip sleet
can the kerning be set so that blinka is made of two half images?
tulip sleet
if blinka is the only one,
slender iron
that's the concept I'd need
tulip sleet
fullwidth is the concept you'd need? I was thinking of blinka-left and blinka-right
but doing the half/fullwidth would make ja more readable
slender iron
well, ideally I'd have one character for the snake unicode codepoint
then it works in your serial terminal too 🙂
tulip sleet
i was just thinking of hacking it as two characters
slender iron
ya, I think thats how I'd do it in terminalio/tilegrids
when a characters is full width it takes two spots with indices n and n+1
slender iron
right, so monospace can still handle full and half width
tulip sleet
@slender iron I was thinking you were going to conditionalize translate.c vs .h:
This isn't as good as the current setup with LTO though. Trinket M0
loses <1k with this setup.
slender iron
baby time. bye!
proven garnet
Good call on hyperspecifying the Sphinx targets, @tulip sleet!
tulip sleet
glad it worked!
proven garnet
Wordy, but it seems to work in the local docs I built
tulip sleet
you may be able to avoid the :py:meth: part, and just do ~module.Class.method
we already do that a bunch of places
or maybe you did that already
proven garnet
Honestly, it was worth it just to learn what the ~ does, I was looking for that bit of information for a while now haha
pearl basin
Hey there pythoners. I'm looking to get adafruit-circuitpython-dht installed, but piwheels, the wheel provider, has an expected outage for the next 10 hours. Does anyone know if there are mirrors available?
Twitter
proven garnet
Hey there pythoners. I'm looking to get `adafruit-circuitpython-dht` installed,...
Can you use PyPI?
pearl basin
I don't think so; raspbian os has the default of piwheels for pip installations, its probably where the arm whl files actually reside. IDK the configuration change to make pypi work correctly on raspbian.
But maybe
proven garnet
You could also download the source code and run pip install . in that directory
pearl basin
Assuming no dependencies.
proven garnet
As long as you have Blinka already you should be good, DHT doesn't seem to have any others.
pearl basin
Install running, but its on a raspberry pi 3, so it has one core cranked to 100%, but it is churning along.
requirements file might be missing packaging and/or pyparsing for the builds, had to manually install those to get it to build on my pi.
Seems to have installed though, so lets see what happens
yay worked, now i have actual python errors.
thanks
proven garnet
No worries, glad it worked!
pearl basin
Whelp, now I'm working on the DHT11, seems to work fine mostly, but about 20% of the time I get this: ```Traceback (most recent call last):
File "./src/temp.py", line 28, in <module>
print("Temperature: {:-3.1f} C".format(device.temperature))
File "/home/pi/gardenpi2/venv/lib/python3.7/site-packages/adafruit_dht.py", line 274, in temperature
self.measure()
File "/home/pi/gardenpi2/venv/lib/python3.7/site-packages/adafruit_dht.py", line 230, in measure
raise RuntimeError("A full buffer was not returned. Try again.")
RuntimeError: A full buffer was not returned. Try again.
I can just average on this and run retries so I can always get data.
I have a 6-second-delay between reads.
Might be time to move to the #help-with-circuitpython channel, I'll do that
OK, looks like it is in fact an old CircuitPython issue: https://github.com/adafruit/Adafruit_CircuitPython_DHT/issues/33
tulip sleet
Whelp, now I'm working on the DHT11, seems to work fine mostly, but about 20% of...
Maybe you have seen this already? https://learn.adafruit.com/modern-replacements-for-dht11-dht22-sensors The DHT's are ... kind of terrible
pearl basin
Seems weird that the digital message is imprecise, since other digital protocols like I2C work just fine. However I have a dozen DHT11s laying about so I'll just put these into a retry loop and do some averaging to get healthy data.
But thanks for the help folks, even if I did invade the wrong channel.
tulip sleet
Seems weird that the digital message is imprecise, since other digital protocols...
the timings are critical, and it's hard to do precise timings on an RPi. I2C is more robust in that regard
pearl basin
Yes I assume the full buffer not returned refers to a timing error getting the bits off the wire, and the related checksum mismatch error is the same issue with double-reading bits / missing bits.
proven garnet
It's late so no need to respond now but before I forget, @lone axle, I realized while fixing my own libraries that the CircuitPython_Org libraries need the Sphinx fix. Let me know if you want me to patch that directly to main on those.
pearl basin
Maybe you have seen this already? <https://learn.adafruit.com/modern-replacement...
Ah you're right, the DHT11s stink out loud. They give consistent information, but when I pointed a fan at them, even though the /usr/bin/vcgencmd measure_temp dropped 10 degrees, they reported the same temp as before. They're only like 2 inches apart from each other. They're junk.
warm stump
Have you got this yet? I've seen this in WebUSB but not in traditional desktop python before and I have to import both usb and serial (tried pygame.midi as well, the Feather didn't show up)
This just creates a mini network bus of directly attached Adafruit Devices - which removes some of the thinking behind sending messages to certain ones when USB connects/disconnects etc...
The device in question can respond only if the message contains a CUID (constant unique identifier) pre-assigned by the code.py and never changed where possible (but this demo doesn't do that)
spiral elk
rp2040 is my guess
By far pi picos would be the most common circuitpython-capable board you'd find in Japan in terms of sales as well. That number of downloads list is relatively accurate in terms of order but for sales everything past the top 3 entries is an order of magnitude lower at least.
Based on what I see in akihabara shops.
I don't understand why the Maker Pi RP2040 is so popular here, except for the fact that its a well made board at a cheap price.
manic glacierBOT
lone sandalBOT
lone axle
It's late so no need to respond now but before I forget, <@382939733107408897>, ...
Yeah, that would be awesome. Thank you.
manic glacierBOT
silk raven
I followed this guide to set up CircuitPython. https://learn.adafruit.com/adafruit-grand-central/circuitpython But I encountered the problem shown in the picture above. So I tried to update the bootloader and reset for a few times, but it still shows the same problem.
thorny jay
Hi, I wanted to ask about the "Status of the USB-Host support and best hardware to play with that". So recently Anne was saying on Pi Cast https://www.youtube.com/watch?v=OGb55VXAp4I that it was there, and maybe @slender iron was saying in the chat "well not really". So I was wondering if there was hardware on witch it is already kind of working? Is it the RP2040 with that PIO trick and you have to solder yourself a USB-A female connector? I also saw on Show and Tell https://www.youtube.com/watch?v=cTCHe8Y8yhk that Scott use that great USB-OTG from Espressif that I have been hesitating to get (also I don't know who contributed that to the supported board): https://circuitpython.org/board/espressif_esp32s3_usb_otg_n8/ . My "usecase" or area of interest is USB-HID translation/man-in-the-middle/retro-gaming. I want to read from old joystick and convert the input to something else. Maybe similar to what @tough flax need for more useful purpose. It's OK to take a more costly hardware initially for development, if later there is a more affordable Adafruit product that does not require soldering and that I can recommend. Thanks.
tough flax
Right now use use a qtpy running cp on the device side and a trinket m0 running Arduino on the host side. They communicate over UART.
I think the best hardware for usb host right now is the teensy 4
But that’s currently Arduino only
TinyUSB Does have usb-pio integrated at this point. So I’d you’re willing to use the C API for the rp2040 that’s an option
My ideal platform is an rp2040 running CP over TinyUSB with usb host on PIO. In a perfect world the usb host would run on the second core (or we can just run the host task in the main loop but that leaves the host timing at the mercy of the device side)
@thorny jay I will tell you that our biggest tip on this is to have the host side read his inputs and update its own state (which keys are pressed, how far the mouse has moved, which cells on the intellikeys are pressed). Then the device side polls and gets the full state and sends its own HID events that show the changes. Just sending through the HID events with modifications looks simple but it leaves open the possibility of keys getting stuck down etc
thorny jay
Thanks, yes sometimes I get CTRL stuck on my physical keyboard, and the computer start to act funny until you understand!
thorny jay
<@681528999520108564> I will tell you that our biggest tip on this is to have th...
I have seen your use of two board back to back with serial in between. Maybe that is the best option right now?
solar whale
The Product page for the new adafruit feather esp32s3 - 4MB Flash/2Mbyte PSRAM https://www.adafruit.com/product/5477 says it is supported by CP but I don't see it in the espressif/boards or on the Download page at circuitpyhton.org. Am I missing it or if it is just not "quite" ready yet. Not a problem, just curious.
jaunty juniper
well it came out yesterday !
it's not in the CP repo yet, I guess it was a surprise release
solar whale
🎁
tulip sleet
I will make a build for it soon.
slender iron
@thorny jay I was working on the Teensy 4 / iMX RT for the original USB host work. It doesn't actually read and write to the device yet though. I stopped when I took leave and haven't circled back to it yet. I'm working on web workflow instead
slender iron
@tulip sleet I think I may be able to speed up the translation builds by externing the compressed string structs
tulip sleet
<@329766224093249548> I think I may be able to speed up the translation builds b...
that would be great!
idle owl
@lone axle I need a favor. I figure you could probably put this together in less than 2 minutes, whereas I can not. I need a quick piece of code for the MagTag that clears the display. That's all I need it to do. Right now, it's showing the REPL because it's the default "Hello World!" that comes with CircuitPython, but I need a GIF of that board, and I need a blank display for it. Let me know if you can help!
jaunty juniper
board.DISPLAY.show(displayio.Group()) ?
lone axle
<@382939733107408897> I need a favor. I figure you could probably put this toget...
I can do that, but it will be a few minutes before I could start.
jaunty juniper
maybe refresh()
lone axle
I do think something like that would work
idle owl
I can do that, but it will be a few minutes before I could start.
No worries there at all. I'm in the middle of another thing.
lone axle
yeah refresh after setting it to the new empty group
slender iron
ok, @tulip sleet I pushed a new version. we'll see how fast it is now
generating the huffman tree is a little slow
could probably speed things up a little by parallelizing it
onyx hinge
could probably speed things up a little by parallelizing it
not sure, maybe on local but on github the CPU parallelism available is only 2...
lone axle
@idle owl I just confirmed that this will blank the display on magtag:
import board
from displayio import Group
board.DISPLAY.show(Group())
board.DISPLAY.refresh()
while True:
pass
Oh nice! Ok, thank you so much!
proven garnet
@lone axle did you clone the FONA library to CircuitPythonOrg? Or did I accidentally do that?
onyx hinge
13512328 1.549 0.000 2.023 0.000 makeqstrdata.py:354(<genexpr>)
1 1.404 1.404 8.936 8.936 makeqstrdata.py:319(compute_huffman_coding)
1752508 0.991 0.000 3.014 0.000 {built-in method builtins.sum}
1752508 0.713 0.000 5.014 0.000 makeqstrdata.py:382(est_net_savings)
```looks like finding a better way to `est_net_savings` would be beneficialor to call est_net_savings less
- ((s, -est_net_savings(s, occ)) for (s, occ) in counter.items()), key=lambda x: x[1]
+ ((s, -est_net_savings(s, occ)) for (s, occ) in counter.items() if occ > 3), key=lambda x: x[1]
)
``` for instance this reduces the number of `est_net_savings` calls to just 132240 (7% as many) by only checking whether it's better to allocate a 'word' to a character sequence that occurs at least 3 times across all messages, and cuts the time about in half. it slightly changes the results of compression, too, not sure yet if it's an improvement or a regression> 1 is almost as good in time and should make no difference in compression
onyx hinge
that's not quite true. say for some reason you had just one occurrence of 'xyzzy' and no where else do x, y, or z occur; the huffman encoding of each letter x, y, and z would likely be >8 bits so making it a 'word' (even though it occurs just once) could still save overall space, because when it's stored as a word it's stored in code points (or maybe utf8s, I've lost track) rather than as huffman-coded.
proven garnet
@tulip sleet re: using something other than try/except for typing imports: Is using sys.implementation.name sufficient? Is that "concrete" enough that it could be used for optimization?
Or does it not meet the criteria of a constant?
And I guess is that module present in all builds?
onyx hinge
to circuitpython/micropython a "constant" is pretty much only something of the form name = const(literal integer). Anything resulting from an import cannot be a const, anything that is a string, tuple, float, can't be.
proven garnet
Thanks for explaining, I meant to ask that when this came up before.
onyx hinge
take a look at tests/micropython/const*.py for what is tested about constants
and it's late here so gn! 😴
currently in Germany 🙂
proven garnet
Thanks for the walkthrough! Gute Nacht!
manic glacierBOT
idle owl
<@330227457296957440> I just confirmed that this will blank the display on magta...
I'm getting a RuntimeError: Refresh too soon error. Added a time.sleep() but it did not help. 😕
NM. A longer time.sleep() worked.
Wait....
No. It worked for a minute, reloaded, and failed with the error.
@lone axle So, not nevermind. It worked for a moment, auto-reloaded, and failed after that with the same error.
lone axle
I find it's most convenient to put time.sleep(4) or similar at the begining of the any magtag scripts to it will always wait long enough to refresh.
lone axle
you could also catch the runtime error and then do a time sleep in the except and try one more time after that.
idle owl
Hmm also good idea. Thanks!
OK, adding the try/except seems to have made it work. It tried to fail even with sleep(4), so now there's another 4 second sleep in the except, and that appears to be working.
That I would not have thought of. Thanks so much for your help!
jaunty juniper
I do that sometimes: time.sleep(board.DISPLAY.time_to_refresh + 0.1)
I think that works 🤔
idle owl
I do that sometimes: `time.sleep(board.DISPLAY.time_to_refresh + 0.1)`
Not using the MagTag lib. I wanted it to be barebones.
Because I'm doing this so a GIF doesn't have CircuitPython on the display.
Not because I need to do more code with it.
jaunty juniper
there !
idle owl
oh interesting.
jaunty juniper
but yeah, you don't need that to make a gif, I was just thinking out loud 😉
idle owl
Fair enough! I appreciate it. I was unaware of that being a thing, so I learned something new today. 🙂
