#circuitpython-dev
manic glacierBOT
tidal kiln
@tulip sleet it looks like latest version of thonny handles the escape stuff OK now? ip addr shows up in window banner, etc
tulip sleet
which is latest version? 4.0.0 handled it at least to the extend of discarding it; I can' tremember if it showed up in the window banner
tulip sleet
I did say that it works ok in one or more guides or the release notes.
so new screenshots is better anyway, for that reason too.
sounds good
tidal kiln
current guide is for old version. output ended up in window text, not banner
yep. also changing guide text.
tidal kiln
fyi - the snake emoji is a potential source of grief. getting some Tk related dumps
tulip sleet
crashes, you mean?
tidal kiln
X Error of failed request: BadLength (poly request too large or internal Xlib length error)
yep
don't have time to go down that rabbit hole
tulip sleet
ugh, which tool is throwing that?
tidal kiln
working around it for now
happens when launching thonny
with board running web workflow connected
disable web work and it launches ok
some googling error points to color emoji support issue in Tk
onyx hinge
tulip sleet
i didn't know thonny used tk
onyx hinge
@tidal kiln what are the details of your linux distribution (ubuntu version etc)
tidal kiln
@onyx hinge 20.04.5 LTS
found that same nominal solution. but also some two comments later that say "did not work for me"etc.
i'm getting thonny running enough for screen caps
so just punching thru for now
onyx hinge
with any luck it doesn't affect win/mac users
tulip sleet
did you try thonny on windows?
tidal kiln
nope. would have to dig out the windows tester box.
tulip sleet
ah, I thought your putty screenshots were windows
tidal kiln
i think limor contributed the windows stuff
tidal kiln
yah - it could just be me. most likely is.
and probably fixable. but just don't want blindly start sudo apt installing stuff. for now.
tulip sleet
whenever you get around to upgrading...
onyx hinge
on debian stable I get a black & white snake in a simple test program, don't have thonny and a board handy to test with.
tidal kiln
yah. i'm weird. i generally don't upgrade. instead i'll do a clean install with latest. and work forward from there.
very much not efficient. but upgrades always seem to only 90% work.
onyx hinge
I just buy a new computer and put it next to all the old computers. less downtime.
tulip sleet
i did too. backed up my homedir, clean install, and put back homedir. I think I removed .gconf and redid all the desktop customizations
tidal kiln
yep. or that.
onyx hinge
in thonny I get a rectangle instead of a snake
on debian stable
but I don't really care/mind
tidal kiln
@tulip sleet screenshots updated. please take a look.
tulip sleet
<@329766224093249548> screenshots updated. please take a look.
Looks great! I deleted the orange warning alerts as I checked each page.
thankyouverymuch
tidal kiln
Win 10 no CIRCUITPY folder:
https://forums.adafruit.com/viewtopic.php?t=197195
can see folder contents. just no CIRCUITPY. and Mu cant connect.
nothing in troubleshooting helps.
jaunty juniper
Win 10 no CIRCUITPY folder:
<https://forums.adafruit.com/viewtopic.php?t=197195>...
so it's just the wrong name, that can happen with a corrupted drive, if they can't access the REPL they can still put the erase code in code.py, it should execute
brazen hatch
or.. just do it from the repl..
far quicker honestly
unplugging stuff abruptly in windows is a recipe for disaster
tidal kiln
i doubt they'll be able to figure out accessing repl in their situation
jaunty juniper
it should still work by pressing "Serial"
brazen hatch
nuke.uf2 ?
tidal kiln
so just ignore the "cant find" pop up in Mu?
jaunty juniper
there should be an erase for the Feather M0 Express
I always get the message that Mu can't find the board because I never name my drives "CIRCUITPY"
but the Serial button works
(I don't use Mu for editing, only for testing Mu stuff)
jaunty juniper
by the way, I have not been on the forums since september because I discovered my company registration situation was wrong, and it took that long to figure out the problem, I should be back soon
brazen hatch
used mu for transferring stuff to micropy
the reason I stopped using micropy
tidal kiln
the nukes are good too. just that theyre specific and there never seems to be the exact one you need for a given situation. or a decoder for what alts work for what boards etc.
brazen hatch
the toml library is coming out nicely, reading (with subtables) is done
I just need to test it lots
orchid basinBOT
tidal kiln
@jaunty juniper fs clean up fixed it. thanks for pointing out that mu could still access repl despite the warning dialog.
https://forums.adafruit.com/memberlist.php?mode=viewprofile&u=515426
jaunty juniper
yeah that dialog means it didn't find CIRCUITPY. There is a tentative PR to find the drives with other names by looking for boot_out.txt, though i'm afraid I killed it by noting that reading boot_out.txt can cause autoreload (if that causes the "access time" to be updated, which can happen)
(discotool can find the drives but they find it has too heavy dependencies)
manic glacierBOT
manic glacierBOT
spark umbra
I just saw circuitpython:main branch is 16762 commits ahead, 1401 commits behind micropython:master.
Is CircuitPython periodically updated/rebased with upstream MicroPython? Or has it changed too much for it to become viable and just some fixes are cherry-picked as needed?
jaunty juniper
I just saw `circuitpython:main` branch is 16762 commits ahead, 1401 commits behi...
we are on MP 1.18, CP 9 should pull latest MP
we mostly share the core language implementation, and the process of updating does involve some cherry picking
brazen hatch
Honestly it will be pretty hard to top off 8.x in terms of features
spark umbra
we are on MP 1.18, CP 9 should pull latest MP
Oh Cool.
languid whale
Honestly it will be pretty hard to top off 8.x in terms of features
ESP32Sx BLE would do it.
jaunty juniper
that could be in CP 8.x
manic glacierBOT
lavish saffron
Hi, I've just raised a PR on Adafruit_CircuitPython_Wiznet5k to match Wiznet sockets to CPython sockets. There's lots more to do, but it would be helpful to merge some of the outstanding open PRs before continuing. All reviews gratefully received 
Thank you!
main furnace
@craggy plume Re: trying to build Circuitpython in a Ubuntu VM.
Which board are you trying to build? At which point does the build fail?
onyx hinge
Just a reminder, there's no meeting today or this week. We'll get together with all <@&356864093652516868> next year on January 3 (Tuesday) at the usual time.
brazen hatch
Is there a guide on how to use the updated adafruit_httpserver?
The api has severly changed and I need to reimplement my stuff with it.
Cause I mean,
https://docs.circuitpython.org/projects/httpserver/en/latest/api.html
really isn't helping when we cant get passed imports..
>>> dir(adafruit_httpserver.methods)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'module' object has no attribute 'methods'
better doc would be welcome; the examples have been reworked (I think) and may be a reasonable starting point
brazen hatch
Alright I will refer the examples, thanks!
As to if I will manage to improve the docs, idk, but I will try.
jaunty juniper
Cause I mean,
https://docs.circuitpython.org/projects/httpserver/en/latest/api.h...
that's because it's a submodule:
>>> import adafruit_httpserver.methods
>>> dir(adafruit_httpserver.methods)
['__class__', '__name__', '__file__', 'HTTPMethod']
manic glacierBOT
cobalt grail
Hey guys. Got this string to translate: "%S". 🙂
onyx hinge
@cobalt grail yeah I'm sorry about that
I realize it's nonsense to present to a translator, but because of the C API inside CircuitPython, it had to be marked as a translated string. I would love if someone could show me a way around it in this case.
cobalt grail
Also, shouldn't it be %s?
onyx hinge
no
%S is an extension to printf formatting in circuitpython, which means that the associated argument is a translated/compressed string rather than a regular string
mp_printf(&mp_plat_print, "%S", translate("I have regrets about my decisions"));
cobalt grail
Ah! Ok then. Maybe there is a way in weblate to mark it "ignore".
onyx hinge
I could go translate it in all the languages before anyone else does
cobalt grail
Well, swedish is already done.
onyx hinge
@cobalt grail I think I've marked it as "read only"
cobalt grail
Not sure what that implies though.
manic glacierBOT
manic glacierBOT
manic glacierBOT
crimson ferry
os.getenv() seems to read the file each time, so presumably the best strategy is to put more frequently-used items at the top. Is there a practical limit to settings.toml file size?
Are there any other optimizations in Python code, or a way to read all of the values at once?
onyx hinge
That's right, each call to os.getenv reads the whole file. No, there's no way to get all values or to get a list of keys either. There may be a missing optimization to stop reading after the desired key is found and its value is read out.
lone sandalBOT
tulip sleet
`os.getenv()` seems to read the file each time, so presumably the best strategy ...
als to @onyx hinge : bill88t is working on tiny TOML parsing library. It will do sections and stuff
i don't think reading the whole file is all that expensive
also not sure what the semantics is of TOML and repeated values, whether the last one is the one that sticks. That was true for .env, so we had to read the whole file
crimson ferry
Thanks, jepler and danh, jus trying to strategize a little. I'm leaning towards keeping "standard" settings in setings.toml, along with a filename of a python settings file
tulip sleet
I'm trying to type bill88t; he's the one who was working on a TOML lib
the only strong reason to use settings.toml instead of a .py file is that the values could be read without starting up a Python VM
that was the motivation, really
crimson ferry
I don't understand what that means in the context of CircuitPython
oh, so core code could use it
tulip sleet
so the web workflow could get started up really early
or we could set a memory value that the VM needed to start
devout jolt
Is anyone else getting Github errors trying to download the CircuitPython Library Bundles? In doing a circup --verbose install [somepackage] I am getting "ERROR: 503 Server Error: Egress is over the account limit", e.g.:
12/28/2022 09:34:37 INFO: Downloading bundle: https://github.com/adafruit/CircuitPython_Community_Bundle/releases/download/20221222/circuitpython-community-bundle-7.x-mpy-20221222.zip
12/28/2022 09:34:43 WARNING: Unable to connect to https://github.com/adafruit/CircuitPython_Community_Bundle/releases/download/20221222/circuitpython-community-bundle-7.x-mpy-20221222.zip
There was a problem downloading the bundle. Please try again in a moment.
12/28/2022 09:34:43 ERROR: 503 Server Error: Egress is over the account limit. for url: https://objects.githubusercontent.com/github-production-release-asset-2e65be/98363921/76c8368a-6960-4dd2-8aac-e2d41d7b35e3?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIWNJYAX4CSVEH53A%2F20221228%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20221228T173438Z&X-Amz-Expires=300&X-Amz-Signature=f40fabc5348c03f89661bec8fae374e4d13598bcd0e80874180c6b4a5bb0a01d&X-Amz-SignedHeaders=host&actor_id=0&key_id=0&repo_id=98363921&response-content-disposition=attachment%3B%20filename%3Dcircuitpython-community-bundle-7.x-mpy-20221222.zip&response-content-type=application%2Foctet-stream
brazen hatch
all the pystack talk got me thinking about swap storage
imagine if we could swap to an sdcard. that would be great
I dont wanna imagine booting off of a spi-mode sdcard though
brazen hatch
I am having quite a lot of trouble creating the stackless builds..
../../py/compile.c: In function 'do_import_name':
../../py/compile.c:1095:27: error: 'q_ptr' may be used uninitialized [-Werror=maybe-uninitialized]
1095 | qstr q_full = qstr_from_strn(q_ptr, len);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from ../../py/obj.h:33,
from ../../py/parse.h:32,
from ../../py/scope.h:29,
from ../../py/compile.c:33:
../../py/qstr.h:84:6: note: by argument 1 of type 'const char *' to 'qstr_from_strn' declared here
84 | qstr qstr_from_strn(const char *str, size_t len);
| ^~~~~~~~~~~~~~
@wraith crow sorry to ping you, but could you please tell me how you compile your stackless builds?
tulip sleet
I am having quite a lot of trouble creating the stackless builds..
```
../../py/...
probably yes, but did you do a clean first?
brazen hatch
Remember the aliases you gave me?
I have modified them and put em on all my machines.
I always picow-cleanmake now.
tulip sleet
i will try a build
brazen hatch
I set MICROPY_STACKLESS to (1) and CIRCUITPY_PYSTACK_SIZE to (0)
It fails ~20s in.
tulip sleet
I did a build for pico that compiled
brazen hatch
Could be incorrect placement though.
tulip sleet
where did you set those?
brazen hatch
in ports/raspberrypi/boards/raspberry_pi_pico_w/mpconfigboard.h
I have git restore'd already so I may have forgotten something.
tulip sleet
what does arm-none-eabi-gcc --version say?
brazen hatch
arm-none-eabi-gcc (Arch Repository) 12.2.0
Yea, I use Arch BTW.
I can try on my Pi400
tulip sleet
we aren't using 12 yet; we're using 10.3
brazen hatch
arch moment
tulip sleet
it may well indicate a possible issue, since it's a compiler warning
brazen hatch
it builds with normal settings though
tulip sleet
builds fine with your placement on 10.3
brazen hatch
imma jump on my pi and brb
tulip sleet
@brazen hatch instead of setting CIRCUITPY_PYSTACK_SIZE to 0, set MICROPY_ENABLE_PYSTACK to 0
brazen hatch
oh alright, I am updating my local repo
It will take a while.
iowait is my life now
tulip sleet
the interlocking between these settings is not necessarily perfect
i think the warning is if a malloc fails and doesn't return anything, or something like that
brazen hatch
I can actively test the result once we build.
It should be no problem verifying it's limits.
It would take very little intentional bad code to send it to 10 bytes left.
It built just fine, onto testing now.
The issue is that I can't build esp on my pi400 cause I have made the unfortunate decision of compiling and installing by hanv python3.11 which doesn't work with esp-idf.
I will only test stackless on rp2 for now.
To make it clear, my changes are:
diff --git a/ports/raspberrypi/boards/raspberry_pi_pico_w/mpconfigboard.h b/ports/raspberrypi/boards/raspberry_pi_pico_w/mpconfigboard.h
index a31ee7327c..c2df2858df 100644
--- a/ports/raspberrypi/boards/raspberry_pi_pico_w/mpconfigboard.h
+++ b/ports/raspberrypi/boards/raspberry_pi_pico_w/mpconfigboard.h
@@ -1,6 +1,16 @@
#define MICROPY_HW_BOARD_NAME "Raspberry Pi Pico W"
#define MICROPY_HW_MCU_NAME "rp2040"
+#ifdef MICROPY_ENABLE_PYSTACK
+#undef MICROPY_ENABLE_PYSTACK
+#define MICROPY_ENABLE_PYSTACK (0)
+#endif
+
+#ifdef MICROPY_STACKLESS
+#undef MICROPY_STACKLESS
+#define MICROPY_STACKLESS (1)
+#endif
+
#define CIRCUITPY_DIGITALIO_HAVE_INVALID_PULL (1)
#define CIRCUITPY_DIGITALIO_HAVE_INVALID_DRIVE_MODE (1)
All I have to say is, stackless is the way to go.
At least on rp2, it's a clear win.
tulip sleet
All I have to say is, stackless is the way to go.
I think we would strongly consider it for post 8.0.0. For 8.0.0 we are trying to not make major changes now, just fix bugs.
Why don't you add some comments to the stackless issue, and i'll move it to 8.x.x?
i appreciate your experimentation
brazen hatch
I want to test this a bit more before declaring it stable, after that, I will write all my finding onto the issue.
Also, I feel like it should be tested extensively on EVERY port
It may work fine on rp2 but fail on a espressif
I can only do rp2, nrf and esp, the rest I cannot get my hands on.
tulip sleet
Also, I feel like it should be tested extensively on EVERY port
It may work fine...
yes, exactly why not to include it in 8.0.0
brazen hatch
And I managed to crash everything (including my usb ports) with a simple
def test():
test()
Recursion limit where
import time
def test(num):
print(str(num))
time.sleep(0.01)
test(num+1)
test(0)
made it to 1.9k before hanging
Along with stackless there has to be a recursion limit implemented
crimson ferry
Is there a resource (besides the code) that describes the CircuitPython memory architecture... heap, stack, long-lived, C stack, pystack, espidf memory, differences across ports, what happens when PSRAM is involved, etc.? Scott did a deep dive long ago, but I'm not sure it covers all of the angles from a high level (and may be a little out of date).
tulip sleet
Is there a resource (besides the code) that describes the CircuitPython memory a...
not really; there are some port-specific README's, etc., but they're often out of date. There is some top-level stuff in readthedocs in various places
tulip sleet
And I managed to crash everything (including my usb ports) with a simple
```pyth...
do include in the comments
brazen hatch
I will include everything dw
brazen hatch
I am done for the day, see ya all tomorrow.
blissful pollen
@onyx hinge were you still looking for a scorpio example project? Still winter related but the Icicle project I did based on PaintYourDragon's original Arduino one may work well. Needs lots of strands, it runs pretty decently. Code would need some tweaking/cleanup
wraith crow
@brazen hatch We seem to be on inverse schedules today 🙂 sorry I missed your ping but glad you got it worked out 😁
manic glacierBOT
crimson ferry
when there is no pystack, where does stuff go that would have gone on pystack? I tried reading the code, but it gets really deep really fast
I'm guessing not the heap, so that leaves the C stack... size of which is determined by CIRCUITPY_DEFAULT_STACK_SIZE?
wraith crow
I started looking a while back until my head started to hurt, but I think there's two different things going on. The pystack and stackless operation. I could be totally off base but my impression was that when pystack was turned off, the functionality was moved to a different stack or perhaps the heap? this is all a bit beyond me... However, there's also a CPYTHON stackless version/option (https://en.wikipedia.org/wiki/Stackless_Python) which perhaps is what gets triggered when the stackless option is enabled????
crimson ferry
there could be some simplistic version of Stackless Python implemented, many of the features seem too rich and os-based for us
but if not, if stackless uses the heap we'll need to test for fragmentation, and if it's c-stack we'll need to test for stack size and C & Python not clobbering each other in odd scenarios
jaunty juniper
my understanding:
MICROPY_STACKLESScauses all allocations to take place on the heap, meaning that even functions that are originally designed to avoid heap allocation can trigger a garbage collection, or contribute to memory fragmentation.
// Avoid using C stack when making Python function calls. C stack still
// may be used if there's no free heap.
MICROPY_ENABLE_PYSTACKenables a separate "stack" from the C stack to handle python local allocations. It's called "scoped allocation" in some of the related discussions on microptyhon.
// Whether to enable a separate allocator for the Python stack.
// If enabled then the code must call mp_pystack_init before mp_init.
How they interact with each other is unclear. At first I thought that the pystack requires explicit allocation on it, but that doesn't seem to be the case (or it would be very niche).
crimson ferry
MICROPY_STACKLESS causes all allocations to take place on the heap
seems to have large implications for fragmentation
jaunty juniper
I might have gotten all of that wrong though
crimson ferry
When I read that comment in code, I didn't know if the 0 was what the comment was refering to, or the MICROPY_STACKLESS
jaunty juniper
ah I don't know
crimson ferry
we really need some docs (with graphics!) of the CP architecture, particularly about memory
jaunty juniper
we can ask in the MP discord, they certainly can clarify or point to more info. Jimmo is active over there
crimson ferry
I'll give it a go
jaunty juniper
(or over there https://github.com/orgs/micropython/discussions)
crimson ferry
Discord so much easier 😉
onyx hinge
@lone axle maybe you know what's going on with https://forums.adafruit.com/viewtopic.php?t=197317 and can help out?
it looks like a label that starts with a space ends up different than one without, or maybe generally different depending on where the first pixel of the first glyph is at?
lone axle
Interesting. I don't know the cause off the top of my head, best guess is maybe space is "special" and handled by the code by moving some pixels in X direction. If that is the case maybe it's not using the correct value to move when it finds the space.
Is space one of the glyphs generally contained within a font file? Somehow the font file must be able to convey how wide the space should be, it could just have a glyph for it with no pixels, or perhaps it's just a number somewhere in the metadata.
lone axle
looks like newline is the only special case https://github.com/adafruit/Adafruit_CircuitPython_Display_Text/blob/0b209f9ada96151f4f2faf0714cb0880b12fb4ed/adafruit_display_text/bitmap_label.py#L380-L386 so font.get_glyph() must return something for " "
I'll post a reply on the forum and dig into it a bit further in the next day or two.
manic glacierBOT
crimson ferry
With CIRCUITPY_RESERVED_PSRAM in settings.toml as a model, is it conceivable that other low-level build features could be deferred until runtime, so that very different kinds of applications could be written without a custom build? At the moment, I'm thinking mainly of other memory configuration options (pystack, etc).
brazen hatch
With `CIRCUITPY_RESERVED_PSRAM` in `settings.toml` as a model, is it conceivable...
That would be nice but once stackless rolls out it will all be deprecated documentation and wasted code.
I'm going to try stackless on esp today and see how it fairs with psram.
If I manage to build it that is
my archbtw has incompatible gcc
and my pi400 has python3.11 which doesnt play well with idf
I may spin something in docker
manic glacierBOT
crimson ferry
@brazen hatch my question was a little more general, could involve other aspects of memory config, even whether Blinka shows up on displayio terminals
brazen hatch
well usability would greatly improve if everything was set via settings.toml
no need for any custom builds ever
Also, my toml library is coming out soon, so it be very easy for code to save/load settings to it too
turns out though, accounting for all edge cases was quite a lot more code that I would like
hopefully I didnt miss any
crimson ferry
plenty of time for people to try out all. the. things.
brazen hatch
also fyi, building with arm-none-eabi-gcc 12.2.0, only requires a single #pragma GCC diagnostic ignored "-Wmaybe-uninitialized"
I read the code, not sure why it thinks of this
it complains about py/compile.c:1095
tulip sleet
also fyi, building with arm-none-eabi-gcc 12.2.0, only requires a single `#pragm...
The compiler is probably not wrong -- it may be uninitialized. I would not suppress the warning immediately, but
try to fix the issue
i think it is beccause the called routine might not always set the value
brazen hatch
Other than that there are no other warnings or issues.
I will give a more throughout look at the func and try to pr a fix.
left steeple
hello, I think I found a discrepancy on the circuitpython download page and I'd like to know who to take it to. Could anyone direct me where to go?
jaunty juniper
here is fine
left steeple
on the download page for the ESP32-C3-DevKitC-1-N4 the text on the page references the "DevKitC" but the download file and url reference the "DevKitM"
onyx hinge
@left steeple if you're a github user you can open an issue at https://github.com/adafruit/circuitpython-org/issues
and if not, reply to this and tag me and I'll open the issue for you
I think "DevKitM" is right straight across: https://docs.espressif.com/projects/esp-idf/en/latest/esp32c3/hw-reference/esp32c3/user-guide-devkitm-1.html
jaunty juniper
yeah that's a good client for an issue, those devkits are weird in their names
I think they can use the same build, I have both but no time
manic glacierBOT
lavish saffron
Might this be solved by refactoring Wiznet5k? At the moment, wiznet5k_socket.gethostbyname calls wiznet5k.get_host_by_name which instantiates an instance of wiznet5k-dns.DNS and calls DNS.gethostbyname which instantiates a wiznet5k_socket.socket and makes calls to socket.send and socket.recv which calls WIZNET5k.socket_write.
This could be refactored into wiznet5k_socket.gethostbyname with a couple of functions to build and parse DNS messages and calls to the WIZNET5K instance that is already linked to the sockets module. The Wiznet5k DNS module could be removed completely.
manic glacierBOT
orchid basinBOT
manic glacierBOT
manic glacierBOT
craggy plume
@ jepler- Thanks for the conformation about the minimum 1 second alarm delay. I figured the exception was in the creation of time_alarm. I was hoping I could trick the alarm into a short delay. I think I should call this a feature request instead of a bug. The bug is that alarm.time uses a datetime_t and is limited to 1 second sleeps. 😣
digital ibex
Hi, when do I know if the CircuitPython port for my OSHW board is ready to be merged into the official repo? Right now, the firmware runs beautifully but I'm still working out the userspace code that runs within CircuitPython. Should I wait until that's done, or can I open a PR?
jaunty juniper
I would say that deep sleep is not really intended for short sleeps, it's followed by a reset, so there's a lot of wasted time (and energy) in sleeping a short amount of time compared to time.sleep()
specifically one second I would say
or was it light sleep ?
ah sorry, I misunderstood the thing 😛
I should go to deep sleep, it's late 😓
craggy plume
I am looking for serious ways to reduce power in the RP2040. I am suprised there is not more intrest in power reduction. Light_sleep is disappointing if the whole app has to shut down for a second to use it. I also found that light_sleep shuts down the PWM logic. I see in the C source the bit mask
RP_LIGHTSLEEP_EN0_MASK which shuts down clocks. I can rebuild circuitpython after commenting out CLK_SYS_PWM_BITS but want to understand more about power reduction. I am still looking for the routine that uses RP_LIGHTSLEEP_EN0_MASK.
ornate breach
I am looking for serious ways to reduce power in the RP2040. I am suprised there...
It’s mostly because the RP2040 isn’t really designed for low power. There are some people who have been able to get it down to 180uA using the pick sdk, but some real finagling with shutting down RAM power.
Those things are not really as big of a priority in my own view of the RP2040 usage as it tends to be viewed as a tool for learning embedded vs optimizing for end products.
It’s also been discussed at length that the trouble of reducing low power consumption in a few cases isn’t worth the potential headaches of managing peripheral state. That said, definitely open an issue if you are curious if your suggestions are worth trying out
Maybe try them out and test them too, if they work submit a PR
I’d certainly be interested if you get more working. I do a lot with the RP2040, including a little watch I’m working on.
craggy plume
@ornate breach- I am on a journy to learn how to make optimizations to circuitpython for my needs. I have circuitpython building on a Ubuntu VM but I have not tried refactoring or adding C calls yet. I am an embedded engineer but mostly code C for PICs. Rp2040 and circuitpython displayio gave me a big headstart on my "watch" project. Now I am running into the limitations and looking for ways to get around them.
What is the pick sdk? where is it?
ornate breach
What is the pick sdk? where is it?
Sorry, Pico SDK
It’s C so right up your alley I’d bet
CircuitPython’s RP2040 builds use the Pico SDK so any optimizations you make could potentially cross over
tulip sleet
Hi, when do I know if the CircuitPython port for my OSHW board is ready to be me...
if you board presents as USB we need a valid USB VID/PID. There are multiple ways to get them: pid.codes, Raspberry Pi folks (if RP2040), Espressif folks (if Espressif-based), or your own VID (expensive). You can make a PR now and mark it as draft if you're not quite ready. You will also be making a PR to adafruit/circuitpython-org with a board description as well after it's merged to adafruit/circuitpython (you could make a draft there too if you want).
digital ibex
if you board presents as USB we need a valid USB VID/PID. There are multiple way...
I've already gotten a PID from pid.codes. Sounds like making a draft PR is my best bet, thanks!
onyx hinge
https://raspberrypi.github.io/pico-sdk-doxygen/group__cyw43__driver.html#ga6d8ce9294b54153f84561add7fae2c18 in the docs I find a getter but no setter
@tulip sleet ^
tulip sleet
https://raspberrypi.github.io/pico-sdk-doxygen/group__cyw43__driver.html#ga6d8ce...
got it, so I will add some compile-time option and remove the ro_attribute stuff from the CYW43 impl.
lavish saffron
Hi, which is preferred for minimising memory use bytearray.extend(b"abcde") or bytearray += b"abcde"?
tulip sleet
Hi, which is preferred for minimising memory use `bytearray.extend(b"abcde")` or...
I don't think it will make any difference
lavish saffron
I don't think it will make any difference
OK, Thanks
random junco
Happy New Year <@&356864093652516868>! This is just a reminder there is no meeting today - we will meet tomorrow at 2pm EST due to the holiday. See you then!
brazen hatch
Created an updated set of aliases for building CP based on Dan's:
alias espload='source ~/git/esp-idf-4.4.3/export.sh'
alias cpmload='source ~/.pyenvs/circuitpy/bin/activate'
alias espunload='export PATH=$(echo "$PATH" | sed -E "s#:?${IDF_PATH}/?##g" \
| sed -E "s#:?${IDF_PATH}/tools/xtensa-esp32-elf/esp-2020r2-8.2.0/xtensa-esp32-elf/bin/?##g" \
| sed -E "s#:?${IDF_PATH}/components/esptool_py/esptool/?##g" | sed -E "s#:?${IDF_PATH}/components/espcoredump/?##g" \
| sed -E "s#:?${IDF_PATH}/components/partition_table/?##g" | sed -E "s#:?${IDF_PATH}/components/app_update/?##g" \
| sed -E "s#:?${HOME}/.espressif/tools/xtensa-esp32-elf/esp-2021r2-patch5-8.4.0/xtensa-esp32-elf/bin/?##g" \
| sed -E "s#:?${HOME}/.espressif/tools/xtensa-esp32s2-elf/esp-2021r2-patch5-8.4.0/xtensa-esp32s2-elf/bin/?##g" \
| sed -E "s#:?${HOME}/.espressif/tools/xtensa-esp32s3-elf/esp-2021r2-patch5-8.4.0/xtensa-esp32s3-elf/bin/?##g" \
| sed -E "s#:?${HOME}/.espressif/tools/riscv32-esp-elf/esp-2021r2-patch5-8.4.0/riscv32-esp-elf/bin/?##g" \
| sed -E "s#:?${HOME}/.espressif/tools/esp32ulp-elf/2.35_20220830/esp32ulp-elf/bin/?##g" \
| sed -E "s#:?${HOME}/.espressif/tools/openocd-esp32/v0.11.0-esp32-20220706/openocd-esp32/bin/?##g" \
| sed -E "s#:?${HOME}/.espressif/python_env/idf4.4_py3.9_env/bin/?##g" | sed -E "s#:?${IDF_PATH}/tools/?##g" \
| sed -E "s#:?${IDF_PATH}/components/esp32/tools/sdk/include/?##g" \
| sed -E "s#components/esptool_py/esptoolcomponents/espcoredumpcomponents/partition_tablecomponents/app_updatetools:?##g") && \
unset IDF_PATH \
unset IDF_PYTHON_ENV_PATH \
unset IDF_TOOLS_INSTALL_CMD \
unset IDF_TOOLS_EXPORT_CMD'
alias lstty='/bin/ls /dev/ttyACM* /dev/ttyUSB*'
alias atio='tio /dev/ttyACM0'
NUM_CPUS=`getconf _NPROCESSORS_ONLN`
make-aliases () {
alias ${1}-make="(${BASH_ALIASES[$4]}; make -j$NUM_CPUS V=1 BOARD=${2})"
alias ${1}-clean="(${BASH_ALIASES[$4]}; make V=0 BOARD=${2} clean)"
alias ${1}-cleanmake="(${BASH_ALIASES[$4]}; make -j$NUM_CPUS V=1 BOARD=${2} clean; make -j$NUM_CPUS V=1 BOARD=${2})"
alias ${1}-dmake="(${BASH_ALIASES[$4]}; make -j$NUM_CPUS V=1 BOARD=${2} DEBUG=1)"
alias ${1}-dcleanmake="(${BASH_ALIASES[$4]}; make -j$NUM_CPUS V=1 BOARD=${2} clean; make -j$NUM_CPUS V=1 BOARD=${2} DEBUG=1)"
alias ${1}-load="(${BASH_ALIASES[$4]}; cp build-${2}/firmware.uf2 /media/${USER}/${3}/)"
alias ${1}-gdb="(${BASH_ALIASES[$4]}; gdb build-${2}/firmware.elf)"
}
make-aliases pico raspberry_pi_pico "RPI-RP2" cpmload
make-aliases picow raspberry_pi_pico_w "RPI-RP2" cpmload
make-aliases wannabe_lolins2mini wannabe_lolin_s2_mini S2MINIBOOT espload
make-aliases wspico waveshare_esp32s2_pico ESP32S2PICO espload
make-aliases combact Seeed_XIAO_nRF52840_Sense "XIAO-SENSE" cpmload
unset make-aliases
unset NUM_CPUS
the secondary env is specifically so that all circuitpython stuff are either using esp-idf env, or that.
Esp-idf removal from path was such a pain.
Hope these help.
onyx hinge
Use a subshell to confine the esp-idf environment changes? ```sh
$ X=1; (X=2; echo $X); echo $X
2
1
Oh.
Oh well..
Thanks.
Imma update it and edit the message above.
Literally spent 3hrs making espunload.
brazen hatch
It's bash, I will have forgotten by tomorrow
Updated it to use subshells.
It's cleaner this way. Leaving espunload in the aliases, in case someone wants to use it manually for flashing or something.
deft jay
https://github.com/adafruit/Adafruit_CircuitPython_ST7789/pull/35 thanks @jaunty juniper
onyx hinge
https://github.com/adafruit/Adafruit_CircuitPython_ST7789/pull/35 thanks <@19981...
I'm not sure why this order should matter! I'm also not sure why the baudrate is set separately, it's in the FourWire constructor.
stuck elbow
I'm not sure why this order should matter! I'm also not sure why the baudrate i...
it matters because the spi bus would be kept alive by the previous display and the pins would be taken
tame creek
i think it’s the known displayio persistence across soft restarts
jaunty juniper
I'm not sure why this order should matter! I'm also not sure why the baudrate i...
yeah I don't know about the baudrate in that example, it's weird
ornate breach
What’s weird is the learn guide for the st7789 displays released the display before setting up SPI
jaunty juniper
we could remove that
stuck elbow
it does nothing, as the fourwire will set its own baudrate anyways
jaunty juniper
it's not in any of the simpletest examples either, which the README code often is a copy of
onyx hinge
it's true, if the spi constructor was anything else but board.SPI() it wouldn't work. I was just wondering if there was a bug with board.SPI() being deinitialized by release_displays().
tulip sleet
@slender iron: @onyx hinge and I were going to over the 8.0.0 list this morning maybe 11:30am ET. But if you are available, we should include you. Could also be after the 2pm, but Limor mentioned meeting sometime after then as well.
i have a non-work mtg starting in 10 mins for an hour or less
slender iron
@tulip sleet later is better for me. we have a niece here to watch ari but I'm helping her as needed. later becca will be back and I can definitely get away
midnight ember
i'm not going to make the meeting today, read the notes in there so far.
@slender iron I did submit an email last night and linked to a mastodon post.
should be at least 1 entry so far? 🤷♂️
slender iron
kk, will get it sorted out. thanks!
midnight ember
also i know someone else used the circuitpython2023 hashtag on mastodon to submit theirs but no idea if they submitted an email along with it
i apologize for my long notes this week. there's been a lot of movement on the tr-cowbell in the past 2 weeks and very excited.
also wanted to share a video of the modular enclosure i'm designing. still new to fusion 360 and quite proud of the progress. i can't code very well... but i can design.
random junco
<@&356864093652516868> In just over 2 hours at 2pm EST is our weekly meeting. I look forward to seeing your hug reports and status updates. See you then!
slender iron
@midnight ember I sent a test email from my personal email and it worked. did you send it to circuitpython2023@adafruit.com?
random junco
@slender iron Are you available to read the core section today?
slender iron
yup! I can switch off reading status updates and hug reports too if you like
random junco
great, thanks! By switch off, do you mean read for the folks who aren't able to attend?
random junco
if you don't want to read them all yourself
that would be great, I would appreciate the help, thank you
manic glacierBOT
onyx hinge
I should go look at the code but I'm puzzled about having e.g., pop() -- aren't CircuitPython's built in palette objects fixed-length? Oh, huh, it replaces the internal palette object with a totally fresh one.. https://github.com/CedarGroveStudios/CircuitPython_PaletteSlice/blob/main/cedargrove_paletteslice/paletteslice_acme.py#L153
If I'm not mistaken Dan made almost all "can't set that property" errors better.
errant grail
I should go look at the code but I'm puzzled about having e.g., pop() -- aren't ...
The length is fixed, somewhat. If you change the size of the palette, the bitmap image just uses what it needs. For example, reducing the palette size means that the image will substitute black for the missing elements. If you extend the palette size, the extra colors are ignored.
PaletteSlice repo: https://github.com/CedarGroveStudios/CircuitPython_PaletteSlice
PaletteSlice demo video. The image is expecting a 255-element palette but the slices are randomized and produce palettes of variable lengths. https://youtu.be/EwtSjBfnhWY
errant grail
I should go look at the code but I'm puzzled about having e.g., pop() -- aren't ...
Yes, a totally new palette object is created. Wasn't possible to inherit a displayio.Palette object, so the class composes it instead.
onyx hinge
I should have mentioned my #circuitpython2023 post: https://octodon.social/@stylus/109615118316334203
slender iron
thanks for hosting @random junco !
errant grail
Thanks!
slender iron
I'm back
onyx hinge
me too.
onyx hinge
That's where I went
random junco
If someone can merge the PR for the meeting notes, I'd appreciate it - thanks in advance: https://github.com/adafruit/adafruit-circuitpython-weekly-meeting/pull/98
@turbid radish Since the newsletter is already out, is there anything I need to do from the meeting? Just wanted to double check
mental nexus
I should have mentioned my #circuitpython2023 post: https://octodon.social/@styl...
Two thumbs up for the 2023 objective of finding performance enhancements for ‘displayio’.
random junco
Here is the notes document for the next CircuitPython Weekly meeting. It is on Monday January 9, 2023 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! https://docs.google.com/document/d/1CCaNlkBSC3KGohT2RsjaM-RgqQzSwrRbdAABbg3NdPg/edit?usp=sharing
tulip sleet
@onyx hinge @slender iron the baudrate setting for SPI flash is less complicated than we thought. THere is a max() done, but only once, and the speed is limited to be at or below SPI_FLASH_MAX_BAUDRATE. The speed is set on each read/write, but I'm guessing that is in case the SPI bus is shared with other devices.
slender iron
so its just the jedec read that doesn't respect it?
tulip sleet
right, so their fix makes sense
the limiting is done in spi_flash_init_device() which is called later
onyx hinge
@crimson ferry I think my question was unclear. I meant to ask whether you un-plugged the earlier device before plugging in the next device. Not how long you waited.
crimson ferry
ah, yes, ultimately trying to have more than one mounted simultaneously
onyx hinge
I hope you can try with #7410. It's a total guess, but it can't be good that every device's "unique identifier" is actually the same 🤯
crimson ferry
right, I'll try the artifacts later
i don't know about other linux flavors, but I can have multiples mounted simultaneously on Raspberry Pi OS
e.g., 4 Pico Ws will mount on Rasp Pi OS, but not macOS 13.1
jaunty juniper
it's a new issue in MacOS 13 too (which I don't have), except we have seen in previous versions the Finder open a file on the wrong board when mutiple ones were plugged in, even with different names
(but only the Finder)
crimson ferry
these 'bad' mounts are more fundamentally broken... they look wrong in shell
onyx hinge
I don't have a mac to do testing for myself
what I'm familiar with from linux/unix experience is that mounting a filesystem has two steps: creating an empty directory to serve as the mount point, then "actually mounting" the filesystem.
So I am speculating that the first step succeeds and the second step fails, because the second step is working by UUID and unintentially refers to the first (already mounted) block device.
But it's all speculation, I want to stress that like 5 times.
crimson ferry
@jaunty juniper you don't have macOS 13, or you don't have this issue?
jaunty juniper
I keep my macos version a few months/a year behind (or more if I don't like what they did like hiding the proxy icons)
I have a sacrificial laptop but it's too old for Ventura
tame creek
did Apple finally relent and let people install a newer macOS virtualized in an older host yet?
crimson ferry
gotcha, yeah I "fell victim one of the classic blunders" ...updating both Macs within a short time (once the UF2 issue was fixed)
manic glacierBOT
My code will actually execute and complete, but ends in safe mode. Shorter other bpm280 code piece will run correctly. Control +C has been unresponsive except in REPL.
Code top and REPEL as jpeg
text version of code[Hard crash code from ciruit.py
while loop installed; no rate code rate call in time wait loop
Cambria 8-31-2022 8 56 PM text backup, 1-2-2023 versi...
midnight ember
@slender iron sent it to adafuit, resending 😁
jaunty juniper
I am curious to see if 7410 solves the issues with the Finder opening a file on the wrong board. I rarely encounter it these days, though sometimes I am caught off guard by it. I would have to find a reproducible scenario, maybe check the volume IDs as curiosity, and test with this PR. (let's say tomorrow)
crimson ferry
You've gotten me in the good habit of opening the file from the bottom BBEdit selection. I think you can reproduce by resetting a board under edit and having another board with same filename.
jaunty juniper
note that tinyuf2 uses get_fattime() for the VSN too, would it cause issues if multiple BOOT drives are mounted ?
jaunty juniper
I am curious to see if 7410 solves the issues with the Finder opening a file on ...
and the answer is yes, it does !
slender iron
yup, I just asked on the issue if they are using nimble
@tulip sleet did you do the toml update for https://learn.adafruit.com/pico-w-http-server-with-circuitpython?view=all ?
tulip sleet
<@329766224093249548> did you do the toml update for <https://learn.adafruit.com...
I did a bunch, but there is a structural problem with that one where CIRCUITPY_* is used in one place and no-prefix in another.
Either this one or another one, I changed the presentation order.
"Creating Your settings.toml" file is mirrored.
orchid basinBOT
manic glacierBOT
onyx hinge
@slender iron now that web workflow is good, what if we add RNDIS to do web workflow over usb? Is it too crazy?
stuck elbow
and replace the circuitpy drive with a samba share
slender iron
@onyx hinge I don’t know what rndis is
onyx hinge
Networking over USB
crimson ferry
macOS 13.1 does seem to keep some stored state that can affect mounting. A couple of times as I've finished updating devices for the USB uuid, they wouldn't remount even after unplugging and delay. But if I took them to another Mac and renamed them (they weren't conflicting on the original Mac, but there were prior devices named CIRCUITPY), then they mounted on the original
the OS usually handles multiple same-name devices fine
manic glacierBOT
manic glacierBOT
manic glacierBOT
thorny jay
Apparently BLE support for Pico W is approaching, at least there are traces in the future 1.5 C SDK. https://www.cnx-software.com/2023/01/04/raspberry-pi-pico-w-bluetooth-le-support/
storm minnow
It has Bluetooth?!?!
jaunty juniper
yeah the module does bluetooth, but it was not implemented yet
stuck elbow
more blobs
storm minnow
I had no idea the hardware was there!
thorny jay
I am not under NDA or anything... but I have strange feeling about Pico and the Raspberry Pi Foundation.
They made the RP2040 where 2=#ofCore 0=Type(M0+) 4=floor(log2(ram/16k)) 0=floor(log2(nonvolatile/16k))... so we can hope some other RP#### one day.
But when the Pico was out, they already knew the PicoW was coming (it the PicoW sacrifice some of the PIO for doing the SPI to the wifi chip).
Then they announced the W and the WH (but for the WH with header we have to wait).
Now they might announce the BLE upgrade...
But at the end of the day, it is still the same chip, that the hype again and again, but it is still the same "product".
And the SBC is not evolving either (but it is not a good time to evolve since they have already trouble to get stock of the current).
They don't really have any new product (except very specific not for mortal CM4 into CM3 package) and they are just keeping us entertained with variation of what they have.
It is very likely that this has an impact on reseller, because who need a Pi HAT if there are no available Pi SBC.
And the "W" of the "Pico W" is not available separately, so company like Pimoroni are just making board with a Pico W soldered on it (and not sure if the antenna location is optimale... and there are no external antenna option!!!).
I don't know what is going on or if we have to expect anything, but Adafruit is making PiCow addon board... I just wait Lady Ada to do an Ask an Engineer dressed with her Raspberry Pi T-Shirt... hoping it mean there will be a new product announcement at 10AM UK time.
stuck elbow
I am not under NDA or anything... but I have strange feeling about Pico and the ...
it's a lot of work and resources to create a new chip
you can't churn them out every year
it also takes time for the ecosystem to catch up
esp32-s2 was released like 2 years ago and still has no full ble support, for example
end there is zero support for espnow
thorny jay
you can't churn them out every year
Yeah, I understand that, however... making the same with a variation is not new development, but sure a cost and why would they do that if the current is selling well. It is just that they just sell the same product again and again somehow, I am just a bit disappointed.
stuck elbow
each time you change anything in a chip design, you have to go through the whole costly process
personally, I'm happy for "unlocking" extra features (like the usb host), especially if I already have the hardware
or remember that time when CircuitPython got a 5x speedup for free?
ornate breach
I’m more disappointed in what feels like a cultural shift by RPi and what feels like a shift away from the maker community.
Like yeah, we have the RP2040 and it’s new variations.. and I understand business is still business, you have to prioritize business that keeps you afloat
stuck elbow
well, they incorporated, that's always a change for the worse
storm minnow
RP4080 when
stuck elbow
suddenly you don't just need to stay afloat, but you have to report inflated growth numbers
brazen hatch
RP4080 when
RP4080 Tie
storm minnow
Imagine if Adafruit made an RPi-type SBC ecosystem
ornate breach
But at the same time they’re alienating the people who made them who they are and it’s sad..
I would buy 10 AdaPuters
storm minnow
I would buy hundreds
brazen hatch
Imagine if Adafruit made an RPi-type SBC ecosystem
feathers kinda are that
just no ada cpu
stuck elbow
there are feathers running linux out there
storm minnow
Has Lady Ada done any CPU design?
ornate breach
Open source silicon moves definitely could open the possibility of Ada Silicon
ornate breach
Has Lady Ada done any CPU design?
She has a masters in electrical engineering, so she’s probably done some VLSI work
stuck elbow
it's a whole different world
ornate breach
it's a whole different world
But it’s getting a lot closer thanks to eFabless and the whole of the burgeoning open source semiconductor community
storm minnow
She has a masters in electrical engineering, so she’s probably done some VLSI wo...
Is VLSI a type of processor?
storm minnow
Ahh
stuck elbow
you can build a 4-bit cpu on a breadboard, but it's a long way from something like rp2040, not to mention the chips in sbcs
ornate breach
you can build a 4-bit cpu on a breadboard, but it's a long way from something li...
I mean people are getting 32Bit rISC-v processors fabbed on the skywater shuttles
stuck elbow
and I'm not even talking about silicon design here, just the design of the processor itself
storm minnow
Well, even if Lady Ada herself hasn't done CPU design, and if it's beyond what she's interested in doing herself, she could probably find folks to do it
I think even small CPUs are done by teams tho
ornate breach
I’d do it
ornate breach
and I'm not even talking about silicon design here, just the design of the proce...
Open source tools though have greatly narrowed the breadth of knowledge necessary to do it
stuck elbow
I think you greatly underrate the difference in size between raspberry pi foundation and adafruit
ornate breach
RPi is huge, an order of 10 larger than Adafruit
storm minnow
I know RPi is biiiiiiiig, but they didn't start off that big
stuck elbow
Open source tools though have greatly narrowed the breadth of knowledge necessar...
no, you still need the knowledge to do anything practical, the tools just make it easier to play
ornate breach
$60m vs like.. 600m in revenue
stuck elbow
I know RPi is biiiiiiiig, but they didn't start off that big
they started with government funding
storm minnow
oh...
stuck elbow
as a part of broadcom
storm minnow
Government, fund Adafruit
ornate breach
no, you still need the knowledge to do anything practical, the tools just make i...
I’m aware you still need knowledge, but the breadth of what you need to know is narrowed. You don’t need to know how to layout semi indigo layers for instance
Basic computer architecture and digital logic + HDL knowledge is all you really need these days. A little persistence to use the caravel user project template. But that’s it
stuck elbow
I’m aware you still need knowledge, but the breadth of what you need to know is ...
no, it's not like that. you don't need to know everything immediately to get everything working, because some decisions are already made for you in the tools, but to make anything practical, you still need to go back and make all those decisions, because those defaults are not going to be good enough
ornate breach
Heck, you can develop a chip in Wokwi.
I’m saying the barriers to get started are very low anymore
storm minnow
Y'all should just make a chip
ornate breach
I am
I’m doing my masters in electrical engineering focused on semiconductor development
stuck elbow
making a game in Scratch is not the same as developing an AAA title, and though it is helpful in teaching, the tool doesn't help you in the end
I agree the tools are improving greatly and making it more accessible, like Scratch does for simple games
ornate breach
Sure, but you have to start somewhere. And these tools lower that first step.
stuck elbow
but that still won't let you design something like the rp2040
stuck elbow
a practical one? I have my doubts
ornate breach
I mean, I did. It wasn’t Turing complete because I forgot a not instruction
stuck elbow
ah, like that, then sure
ornate breach
These days, many fabless companies develop their ICs in HDLs and then use HDL translation tools to layout. They still have dozens of engineers working on it of course..
But developing a circuitpython ready MCU? If someone worked at it for a year they could probably do it. Even with only a working knowledge of digital design
stuck elbow
but that's a very bright person not working on circuitpython instead
ornate breach
And this is because the tools exist that remove many of the tedious steps in chip design.
ornate breach
but that's a very bright person not working on circuitpython instead
But circuitpython already has a lot of supporting persons? Why take them away from it?
storm minnow
What makes a design "practical" that can't be done with the tools?
ornate breach
The most difficult part you’ll find is having a good enough understanding of computer architecture and how to translate that into tools that can program the device. If you’re doing open source, you’ll have to use COTS open source IP cores. You don’t get SWD, you have to use standard JTAG, you have to reduce memory complexity, etc..
storm minnow
The most difficult part you’ll find is having a good enough understanding of com...
I assume that's more an issue for CPUs for an SBC and less for MCU design
ornate breach
I assume that's more an issue for CPUs for an SBC and less for MCU design
No, it’s very relevant for MCU design. Microcontrollers incorporate CPUs in with hardware peripherals and memory.
storm minnow
Does memory complexity matter much for an MCU?
ornate breach
It can if you’re trying to squeeze performance out of small die areas.
storm minnow
Ah...
ornate breach
My argument is that you can design something that could run circuitpython in a year, I’m not saying it will run amazingly; but you could do it if you put in the sufficient amount of effort.
storm minnow
And that's the first step to eventually getting to amazing
ornate breach
It would probably not be very optimized or particularly powerful. But it could be done.
storm minnow
It's kinda like Intel with their Arc GPUs right now -- they're not really impressive compared to AMD or NVidia's offerings, but they're in the ring, and have a trajectory for improvement
ornate breach
It gives them a starting point
storm minnow
Exactly. Gotta start somewhere
ornate breach
For Adafruit, they wouldn’t necessarily have to even start with designing a chip. Develop IP that can run on an Adafruit fpga dev board, that IP can then be iterated on and improved, then translated into test silicon.
It would take like, 4-5 years to fully get through
storm minnow
I wish I had the spoons to jump in on that
ornate breach
I wish I had the spoons to jump in on that
Start small, try the next Tiny Tapeout coming this year. Probably march timeframe
storm minnow
Maybe next year? I'm trying to clear my project plate before adding more things... I already added things this year >~>
ornate breach
Maybe next year? I'm trying to clear my project plate before adding more things....
There’s like a 2 week window for submitting
storm minnow
Yeah, there's no way I'm gonna make that DX I currently know nothing about that sort of design and have way, way too much on my plate to be able to learn any time soon
ornate breach
Lol, anyway drifting off topic and I have a meeting 😬
storm minnow
Oop.... ttyl
slender iron
First CircuitPython2023 overview blog post is up: https://blog.adafruit.com/2023/01/04/first-circuitpython2023-posts/
manic glacierBOT
GitHub: https://github.com/0xCB-dev/0xCB-Helios
Shop: https://keeb.supply/products/0xcb-helios
Pinout reference: 
crimson ferry
Is it intentional for the error to persist in the terminal title bar after a reload? (PyPortal)
orchid basinBOT
manic glacierBOT
hollow gazelle
the Scorpio page mentions ". It's also possible to flip the direction of the level shifter to make the 8 I/O pins inputs - say for making a logic analyzer - with a directional jumper selection also on the bottom of the PCB."
is any one looking at ways to collect logic analyzer data and present it to existing GUI applications?
blissful pollen
There was some work on hooking the RP2040 to sigrok / PulseView. I had a very rough implementation that basically dumped data out that you can load in. kmatch I think got more working
This project has a lot on it too: https://github.com/pico-coder/sigrok-pico
mental nexus
Here’s how far I got: https://github.com/kmatch98/tinylogicfriend
ornate breach
Here’s how far I got: https://github.com/kmatch98/tinylogicfriend
I am intrigued by this
tulip sleet
Is it intentional for the error to persist in the terminal title bar after a rel...
I don't think after a reload. Could you file an issue? Thanks.
manic glacierBOT
hollow gazelle
I was reading https://learn.adafruit.com/intro-to-rp2040-pio-with-circuitpython and the examples seem to feature using the PIO for output. I didn't see any overall architecture or examples for input from PIO. How would CP read from a PIO state machine?
blissful pollen
I was reading https://learn.adafruit.com/intro-to-rp2040-pio-with-circuitpython ...
An example like reads from rx uart: https://github.com/adafruit/Adafruit_CircuitPython_PIOASM/blob/main/examples/pioasm_rxuart.py
hollow gazelle
An example like reads from rx uart: https://github.com/adafruit/Adafruit_Circuit...
Thanks for the place to start - I guess I'll need to read about autopush and push_threshold
blissful pollen
I think auto_push is just if the PIO buffer gets pushed out or the PIO program has to explicitly call for that (been a while so I could be wrong). push_threadshold is where some of that takes place.
hollow gazelle
I *think* auto_push is just if the PIO buffer gets pushed out or the PIO program...
Other functions that I will study are readinto and in_waiting ( from https://docs.circuitpython.org/en/latest/shared-bindings/rp2pio/index.html )
blissful pollen
A lot are just passthrus from the actual PIO SDK so you can probably find information there as well if needed
hollow gazelle
A lot are just passthrus from the actual PIO SDK so you can probably find inform...
good point - I've been looking at PIO SDK a bit - but I didn't check out the source code for the CP wrappers yet - that's another good resource - ( I had been trying to think about how to approach the problem from the documentation side at first )
orchid basinBOT
sharp jacinth
What do folks think of the new P4 chip? From a cursory glance I think it's CP compatible
tulip sleet
What do folks think of the new P4 chip? From a cursory glance I think it's CP co...
it certainly looks interesting. Price and packaging seem to be unknowns at this point
sharp jacinth
Ah ok cool
Well I feel like I could justify a good bit higher price for something like that on a feather or smaller
brazen hatch
sign me up
almost a meg of sram
What kind of form factor though can have 50 pins..
jaunty juniper
a big one !
brazen hatch
I can only think of something like the PGA 2040
jaunty juniper
I don't know package names, but when I look at the Grand Central it's big
(that's a SAMD*51 and 70 pins)
brazen hatch
https://shop.pimoroni.com/products/pga2040
something like this but bigger
sharp jacinth
So inside the qt py samd21, is the silicon the same as the samd21 on the grand central?
Just not as exposed?
brazen hatch
if it were to be pico form factor it would be loooong boi
jaunty juniper
since it's Espressif there's gonna be 20 different modules with confusing names is all I know 😉
brazen hatch
That's the best part.
jaunty juniper
So inside the qt py samd21, is the silicon the same as the samd21 on the grand c...
oh I mistyped it's a samd51, but chips do exist in different packages and variants, compare the feather M4 (ATSAMD51J19) and the grand central (ATSAMD51P20A)
https://www.digikey.com/en/products/filter/microcontrollers/685?s=N4IgTCBcDaIIIBUDKcCyARArARgFLYE4QBdAXyA
https://www.digikey.com/en/products/filter/embedded/microcontrollers/685?s=N4IgTCBcDaIIIBUDKcCyARArARgApgAY4QBdAXyA
manic glacierBOT
manic glacierBOT
tulip sleet
<@&356864093652516868> Reminder of today’s CircuitPython Weekly meeting 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 to the notes document: https://docs.google.com/document/d/1CCaNlkBSC3KGohT2RsjaM-RgqQzSwrRbdAABbg3NdPg/edit?usp=sharing
lone axle
<@&356864093652516868> See above about meeting today ^ (this didn't ping, I think due to missing t in the role tag)
tulip sleet
<@&356864093652516868> See above about meeting today ^ (this didn't ping, I thi...
Thanks! I did it from my phone. edited, but it doesn't ping when you edit, so your ping was still needed
lone axle
Does anyone happen to know if / how we instruct RTD on what version of the theme to use for building the docs for the pages that get hosted in their infrastructure?
Ahh, maybe it's just based on the date the project was created:
Ahhhh, yeah that does seem to be the crux of the issue. Any of our libraries created before that date use an older version of the theme that doesn't handle copying example code correctly (it includes line numbers which then have to be manually removed) :/
tulip sleet
Ahhhh, yeah that does seem to be the crux of the issue. Any of our libraries cre...
maybe we can change that default
lone axle
If I am understanding the recomendations on this page: https://docs.readthedocs.io/en/stable/guides/reproducible-builds.html#don-t-rely-on-implicit-dependencies If we include a pinned version (or even maybe unpinned?) inside of docs/requirements.txt, it gets referenced from there in .readthedocs.yml and I'm assuming their instance will install from there and then end up with the same version we use locally.
I'll make a note to discuss it in the weeds.
tulip sleet
@lone axle I'm not sure if this helps any but here is the "Advanced Settings" for a typical library. Note the "ignored" stuff at the top, so it may be irrelevant.
lone axle
interesting. Yeah I'm thinking our config file is taking precedent. We specify python 3 in there and I believe it uses that when it makes the builds. Not certain though.
brazen hatch
I presume the sus stdin on C3 is known right?
If you send characters during boot, it breaks down entirely and it occasionally desyncs and requires double presses.
tulip sleet
I presume the sus stdin on C3 is known right?
not sure, if you think it is something to document or is something we can fix, let us know. Add feedback to the QT Py C3 guide (see left sidebar) if you think we should adda caution
brazen hatch
I'm talking about beetle c3 but fine
When I find time I will go write a detailed report / issue.
Its just seems so frequent ppl should be complaining about it like all the time
this is the reason I havent spoken about it
I have been fighting it since beta 0
though beetle c3 is a board I only use to specifically test c3
I hate it
I just had to test stackless
wraith crow
The c3 had issues with the serial repl for a long time and only recently became usable at all. There's probably not a lot of people using it yet.
brazen hatch
Oh, so its documented then?
wraith crow
I don't think the issue you're seeing, there was a serial timing issue
it would drop characters randomly
brazen hatch
Ljinux has a full 'curses' library
It can break pretty badly on c3
special character input is a 70% success rate
like arrow keys and things are 3 bytes
once one fails to go through, it progressively breaks down further
wraith crow
I'm thinking you should definitely submit an issue when you get a chance. The last time I was using the c3, I was so happy it worked at all that I would have expected some minor problems
brazen hatch
alright will do, thanks for the info!
onyx hinge
if there's not an open issue we typically don't know it exists; and even if we knew and forgot to file the issue, that doesn't help the community. So, filing bugs is almost always a fine idea (searching for closed ones first doesn't hurt though)
idle owl
@plucky tulip In the notes document for the meeting - please put your updates in alphabetical order, ignoring the host. 🙂 It basically doesn't matter anymore because we simply read the notes, but still, it can help.
If that makes sense anyway.
tulip sleet
@plucky tulip if you would like to speak, we can add you to the right discord role for that
idle owl
Their update says text only. But good call, Dan.
We can add you to the role regardless, if you'd like weekly meeting updates. 🙂
tulip sleet
i was just seeing who was in the chat already
idle owl
i was just seeing who was in the chat already
Ah ok!
plucky tulip
Oh, sorry, didn't know about alphabetical thing. If you could add a role, I would apprecieate the updates 🙂
idle owl
Oh, sorry, didn't know about alphabetical thing. If you could add a role, I woul...
Done!
onyx hinge
You may have to leave and re-join the voice chat before you can send audio in there
not sure if that discord limitation still exists or not
tulip sleet
lone axle
gilded cradle
Is the CircuitPython forum the same as the Adafruit forum?
lone axle
lone axle
Is the CircuitPython forum the same as the Adafruit forum?
I think so, or at least I'm not aware of any others that are used as much as the Adafruit Forums
midnight ember
yey nice one 2231puppy! gotta be the haptic fidget spinner.
gilded cradle
👍
midnight ember
Congrats on the board 2231puppy!
digital ibex
Thanks!
midnight ember
Welcome back Kattni
idle owl
Thanks!
midnight ember
Where are you selling them and what program are you using to design the enclosure?
digital ibex
Where are you selling them and what program are you using to design the enclosur...
Probably Etsy and Blender or Tinkercad respectively
midnight ember
looking forward to seeing the new pedal restoration, the ibanaz stuff was neat to see.
thorny jay
Also I pushed this to Twitter but a minute too late for the notes (it could be interesting for LEGO fans): https://twitter.com/DavidGlaude/status/1612531378507096088?s=20
onyx hinge
you may need to explain the johnny 5 reference .. I think I get it but that movie was a looong time ago 🙂
midnight ember
midnight ember
onyx hinge
classic
storm minnow
It’s only… checks notes 36 years old?!?!
idle owl
"for real this week" 😂
storm minnow
How did we get old
thorny jay
lone axle
<@382939733107408897> https://www.lego.com/en-us/product/table-football-21337 ?
Yep, that is it.
there are manual little pegs for keeping score, but I've heard they are tough to use. Plus I can't skip out on a chance to embed circuitpython device into something else!
If I get everything working I'll publish the steps to "retrofit" the stock set with the bits needed to add sensors into it.
idle owl
"buglet" 🤣
midnight ember
keyboard guru
digital shoreBOT
idle owl
I love it.
lone axle
Open source hacking vacation sounds awesome. Fascinating project with the display emulation as well, interested to see where that leads
midnight ember
🆕 spudgers
onyx hinge
spudging videos are the new unboxing videos
midnight ember
the right to repair stuff is really taking off, ifixit is getting a lot of stuff in like that
installer wizard? what is this. interesting.
idle owl
the right to repair stuff is really taking off, ifixit is getting a lot of stuff...
They even reference the Apple tech support product. It wasn't something Apple sold, it was what we used for in-house repairs.
onyx hinge
installer wizard? what is this. interesting.
for espressif family boards, as I understand it
idle owl
I had a few left from being there, but no idea where they are. It was a long time ago. 😄
digital ibex
I think I'm gonna go text only, it's getting a little busy where I am
midnight ember
never underestimate the usefulness of a quality spudger
the line numbers are an issue for beginners who intend to copy code
digital ibex
Actually I've gotta go, bye all!
thorny jay
installer wizard? what is this. interesting.
I think it relate to ESP32 install/upgrade from circuitpython.org . @gilded cradle is that it?
midnight ember
there are some examples in learn guides that contain repl >>> and it's kinda the same issue too. we do see beginners occasionally in help with that have unknowingly copied repl characters into their code.py
onyx hinge
well that makes my list of libraries by creation date less useful 🙂
testrepo is old enough but kattni's other points stand
maybe irremote added 2018-01-26
thorny jay
Thanks.
errant grail
Thank you!
onyx hinge
@lone axle happy for the chance to work on my web API skills, they are still rudimentary!
midnight ember
Thank you for hosting Dan
ember iris
Thanks all, have a good one!
gilded cradle
Thanks
lone axle
I wonder if it would be possible or worthwhile to try to include a clarification message on the page for boards that don't have a stable release yet. It's ideally a short window of time and devices, but I have seen it come up a few times. A message explaining that it is normal and expected for those device to have no stable release at this time might help re-assure people they have the right thing.
midnight ember
List for batch 2 of TR-Cowbells is now full. Cannot accept anymore inquiries.
lavish saffron
@lone axle Thanks for all your hard work with the reviews of the Wiznet driver. I've updated the other pull requests to include the changes that you just merged. Hope that makes life easier for you.
idle owl
List for batch 2 of TR-Cowbells is now full. Cannot accept anymore inquiries.
I seriously wish I had a use-case for it. It looks really neat. But I genuinely don't. Haven't jumped into the black hole of synth yet.
lone axle
<@382939733107408897> Thanks for all your hard work with the reviews of the Wizn...
For sure, Thank you for all of the improvements! That will definitely help me test them out, I appreciate it, thank you.
midnight ember
@idle owl 16 switch linear macropad. eventually i intend on using it with OBS to switch cameras while recording youtube videos. basically like any macropad.
idle owl
<@330227457296957440> 16 switch linear macropad. eventually i intend on using i...
Oh neat. Fair enough, I was thinking it was more a sequencer or whatever the terminology is. I also don't have much going on that I've considered a MacroPad for either. Point is, I wouldn't want you to put in the effort for me to not find something to do with it. But it looks really excellent!
tulip sleet
Here is the notes document for next Tuesday’s CircuitPython Weekly meeting. The meeting is Tuesday due to a US holiday on Monday. 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/1aNx4nRN0TXR7JaEv5dAu139IgZDz8aMCPL-wS6WICZI/edit?usp=sharing
midnight ember
yes the hardware format is designed primarily for that but with mode switching you can make it a macropad. sequencers were originally video broadcasting devices for switching cameras, then someone started using it exclusively for music buttons. history of sequencers are rooted in broadcast video.
so they are naturally great for someone like JP or streamers who have multiple cameras.
idle owl
spudging videos are the new unboxing videos
I only vaguely registered you posting this, and just now thought to go back again to read it properly. Hah! "I'm Kattni, and this is Spudging Episode 4: Removing Batteries"
midnight ember
is that a real episode? i'd totally watch it.
idle owl
is that a real episode? i'd totally watch it.
🤣 Not yet, anyway.
(I'm actually lol'ing over here)
onyx hinge
there are certainly youtube channels of taking electronics to bits, big clive comes to mind. it can get a bit repetitive, mind you.
idle owl
Yeah, that's why I picked something super small for my apparently-not-so-silly example.
Because with something like that, if you combine too much into one, you'll run out.
midnight ember
spudgers seem silly until you need to replace the battery in your car's key fob, those suckers are precision tight and almost always with a waterproof gasket.
ornate breach
spudgers seem silly until you need to replace the battery in your car's key fob,...
You misspelled needle nose pliers
lone axle
So far I've hit a bit of a snag with the shinx-rtd-theme issue. I added sphinx-rtd-theme to docs/requirements.txt inside the irremote library to test. But the problem is inside of RTD it executes pip install on it's own default hardcoded version first, then afterwards it executes pip install on our requirements file.
When ours runs it gets "requirement already satisfied" for the library since it was installed in the previous step at that hardcoded default version.
If there were some flag or way inside requirements.txt to specify that you want to force it to get the newest one, that could be a fix. But as far as I can tell that doesn't seem to be a thing that exists.
We can specify a version in the file i.e. sphinx-rtd-theme==1.1.1 and that does make it get the stated version and then does indeed fix the example copying issue. But that means our version is now pinned and we won't use newer ones as they become available. We currently have sphinx>=4.0.0 in the requirements, so maybe the crux of this issue was discovered before? Or maybe we were just more keen on a pinned version for sphinx.
So far though it's looking not possible for us to force RTD to always use the latest version. We can force a specific version (or greater or lessthan with pip syntax), but not force the latest version. Hopefully I am missing some part of the puzzle and there is actually a way. I've created an issue here to ask the RTD folks: https://github.com/readthedocs/readthedocs.org/issues/9876
stuck elbow
So far I've hit a bit of a snag with the shinx-rtd-theme issue. I added `sphinx-...
I'm not sure you actually want to always use the latest version. Pinning the version to a known working one seems like a reasonable thing to do. Then you can upgrade it as you have time to test everything with a newer version, or as you hit problems with the old one.
idle owl
So far I've hit a bit of a snag with the shinx-rtd-theme issue. I added `sphinx-...
I'm pretty sure you can do the same thing as we did with sphinx to force it if we want latest.
Oh hmm.
Didn't finish reading before responding apparently.
Wait, isn't "greater than x.x.x" going to get latest? Or is it a freeforall at that point.
stuck elbow
Wait, isn't "greater than x.x.x" going to get latest? Or is it a freeforall at t...
only if the one that rtd installs doesn't already fulfill the requirement
lone axle
Wait, isn't "greater than x.x.x" going to get latest? Or is it a freeforall at t...
I'm unclear on the specific behavior. But I think it will get the lowest one that is greater than the given value, so it will get the "next newest one" after the specified version, but won't keep getting new ones as time goes on and more new ones are released.
If I am wrong and that does actually get the newest ones, that would solve it for us though I think.
stuck elbow
If I am wrong and that does actually get the newest ones, that would solve it fo...
I think it gets the newest one that fits the specification, as long as the one already installed doesn't fit it
idle owl
@lone axle Please tag me on that RTD issue. We have an odd situation with our account because we wanted to move it to a business account, but that would have broken everything we already have, so they enabled the features on our account, but we're still "community" or whatever it's called. Anyway, point being, if there's something that they can do on their end, it's possible I can make that happen.
They didn't have a migration tool or anything yet for going from a massive community account to a business account.
Thanks!
idle owl
I'm unclear on the specific behavior. But I think it will get the lowest one tha...
That seems silly to me.
lone axle
I can see upsides of both pinning, and always taking the latest. I don't have a strong opinion either way. In this case RTD has chosen to use latest version for that library by default on projects created after a certain date, so I'm hoping it's at least possible to make the same behavior for ones created before.
idle owl
But fair enough.
lone axle
I will try with the Greather than and see what version it lands on. right now for that library 1.1.1 is latest "real version" but they do have some 1.2 rc's
idle owl
I can see upsides of both pinning, and always taking the latest. I don't have a ...
I typically prefer taking latest. I would rather run into needing to fix things, but keeping up with things, than to realise in 6 months we have to update the thing and do even more work then. But that's me.
I understand both sides of that camp though.
lone axle
yep, different libraries are better or worse at breaking change releases as well so it all depends on so many things
idle owl
Indeed.
lone axle
tagged you in a comment on there. Heading outside to spray off some things while it's not quite so cold. back in a bit.
idle owl
tagged you in a comment on there. Heading outside to spray off some things while...
Thank you!
lone axle
Ah, I do think Deshipu is right about the greater than taking newest, rather than "lowest that is newer than"
Looking at sphinx for reference it looks like it gets a much newer version when it installs that one.
So maybe this is actually easy fix of just adding the greater to whatever version is "new enough" to have the fix, or really any new-ish version.
Collecting sphinx>=4.0.0
Downloading Sphinx-6.1.2.tar.gz (6.7 MB)
Oof ok.
idle owl
```
Collecting sphinx>=4.0.0
Downloading Sphinx-6.1.2.tar.gz (6.7 MB)
```
That seems arbitrary... if it's not latest?
lone axle
yeah definitely. I hope it is the latest because it would be so weird of one to land on if not.
tulip sleet
I will try with the Greather than and see what version it lands on. right now fo...
I think you can set a range, greater than (or equals) than this, and less than (or equals) to that
lone axle
Yep I think that is true, I saw some with a range and a few other combinations. Apparently you can use commas to make arbitrary multiple logic rules for it.