@solar whale @silver tapir not sure why I missed that, it's fixed now
#circuitpython-dev
ionic elk
onyx hinge
@hidden geyser thanks for looking into that cookie cutter path thing.
solar whale
@ionic elk when I put your PR on my board it wont boot - here is the log from the debug USB port
main builds and runs OK.
manic glacierBOT
solar whale
@onyx hinge are the recent build issues you have had with gcc10 or 9?
onyx hinge
@solar whale they are affecting main branch in github actions, which is still using gcc9 for the arm targets.
if you're talking about 3235
solar whale
yes -- just checking -- after my experience with gcc10 on the GCentral I'm backing off using it.
onyx hinge
so happy to have your testing
onyx hinge
@tulip sleet feel free to push to that PR branch, I'll try to be mindful and not force-push it
tulip sleet
it's looking good; i searched all over for cache clearing info, and didn't find it, but the cache tag seems important, and it will delete the cache stuff in a week or something like that. We could make the tag be generated on the commit or date or something like that.
onyx hinge
woo thanks @tulip sleet
solar whale
@tulip sleet what does the esp-idf cache tag impact?
tulip sleet
it identifies the exact cached installation stuff in /opt/hostedtoolcache or whatever it was called. There's some info here but I don't understand this in any depth at all : https://docs.github.com/en/actions/configuring-and-managing-workflows/caching-dependencies-to-speed-up-workflows
Also I should be saying "key", not "tag"
solar whale
ok -- so it only impacts the github actions, not my local builds...
tulip sleet
right; we were using a key that had 20200523 or something liike that to id it; I just edited it to 20200801. I think we should make that not be hardwired, but I'm not sure what it should be (commit? today's date? This week?` etc.)
solar whale
Ok -- I'm still trying to figure out why my builds of PR 3232 don't work but main does.
they build but wont boot... main works fine.
tulip sleet
do you have the same commit of esp-idf or whatever checked that main does?
tulip sleet
is it on linux or your rpi?
solar whale
Mac and RPi
tulip sleet
i'll try it myself
solar whale
when I pull and checkout the PR it shows 3 submoules are different ```jerryneedell:/Volumes/CircuitPythonBuild/circuitpython/ports/esp32s2$ git checkout pr_3232
M extmod/ulab
M lib/tinyusb
M ports/atmel-samd/peripherals
Switched to branch 'pr_3232'
yes, those have been updated since the pr was submitted, it appears
solar whale
actually I built it both ways (sync and not) and got the same results
tulip sleet
what is the commit for the PR? Are you trying it in adafruit/circuitpython or hierophect's fork?
solar whale
on adafruit circuitpython
tulip sleet
it's failing in the PR checks, too
tulip sleet
ah ok, I was mixed up; then i have no particular insight, sorry
solar whale
seem like USB is hosed... fresh install did not help -- I'll try cloning his fork
onyx hinge
laceback (most recent call last):
rxile "<stdin>", line 1pn irsrtGGZeroDivisionError: liision by zero
``` looks like I made a mistake changing how translations are compressed againThat looks compressed 😉
onyx hinge
good point, ship it
onyx hinge
OK, after scaring the clownsbugs away, I have determined that it's NOT a space improvement to store the trailing "\0" of each compressed string, rather than its length. https://gist.github.com/66521c5cb28729d7990795dfd539ca16 for posterity
lost nearly 100 bytes on trinket m0
solar whale
@tulip sleet I tried building from his fork -- same result -- I'll just wait until it is merged....
onyx hinge
the "nul byte" got a compression of 20 bits long(!) no wonder
@tulip sleet I think you had suggested this might be a savings but the numbers weren't with you this time
tulip sleet
ugh 🙂
onyx hinge
the values[] array (1 entry per distinct unicode appearing in the strings to translate] is 8 bits big for "latin1" and 16 bits bit for anything that goes outside of byte values 0..255. We could get savings for some languages by making the choices other than 8 or 16 bits (e.g., 9 or 11 bits), but that doesn't help de_DE where the unicode values are all below 255.
manic glacierBOT
manic glacierBOT
solar whale
I’m using one of these. https://www.adafruit.com/product/389 what board are you testing on? I’ll try the error recovery later today.
manic glacierBOT
manic glacierBOT
timber mango
Father Sense BLE: How long can I expect a charged 18650 battery to keep it on, provided it's just on, no sensors used?
Anyone know a ballpark figure?
tulip sleet
@timber mango I will measure the current drain, hold on...
about 10ma when idling in the REPL
@slender iron got a minute for supervisor_tick() q?
or would it be a jepler q?
timber mango
Thanks, @tulip sleet, much appreciated. Now, after 7 hours of sleep, it's still got enough power to turn on my flash light module.
I'm going to leave it in and see how much longer it'll last.
I test this because in my previous thing, I had a trinket powered by a lithium version of a small 3.6V camera battery. That thing lasted only a few hours, even in stand by.
manic glacierBOT
manic glacierBOT
silver tapir
Where can I get the link for today's meeting?
lone axle
The doc is pinned. https://docs.google.com/document/d/1i_l-w0lmHZg7vSoNgAvNWDoVwK1jU5dojDuJvzQgvMQ/edit?usp=sharing
beyond that I don't think there really is a link, it's just in the voice section of this channel
silver tapir
Thanks.
turbid radish
?serverinfo
digital shoreBOT
adafruit
onyx hinge
@gilded cradle Are there many "important" displays that are supported by framebuf but not displayio?
gilded cradle
@onyx hinge only eInks and Monochrome displays on the Raspberry Pi as far as I'm aware. RA8875 is the only one I can think of not on displayio, but it also doesn't use framebuf.
thorny jay
I know this is only H-2 for the meeting... but I will be lurking. I quickly filled the report, but I am totally unsatisfied with what I did. Either I have done nothing (highly possible) or I forgot what I did (also true). So I could only do it based in GitHub activity.
onyx hinge
@thorny jay happy to have your notes in any case. Not covering absolutely everything is fine
<@&356864093652516868> Our weekly meeting is in about an hour at 2PM ET. Please put your notes in the doc https://docs.google.com/document/d/1i_l-w0lmHZg7vSoNgAvNWDoVwK1jU5dojDuJvzQgvMQ/edit?usp=sharing and especially let us know if you're "lurking". It's super helpful!
Starting last week, you must be a Circuitpythonista to speak in the meeting. If you plan to speak but you're not a Circuitpythonista, please let us know and we'll add you.
See you here and in the voice channel in about 1h.
lone axle
@onyx hinge are you running the meeting today? I can paste the news links in here if so.
onyx hinge
@lone axle I think it's Scott and I assume I'll be taking notes. We haven't quite worked out the new rotation. If you want to paste links, though, that's awesome and you are welcome to do it!
manic glacierBOT
lapis hemlock
@onyx hinge Never dreamt of this: https://forum.micropython.org/viewtopic.php?f=3&t=7005&p=49844#p49842
onyx hinge
https://www.electrodragon.com/w/images/0/04/LY68L6400_0.3.pdf seems to be the chip on the unexpectemaker feather s2
@lapis hemlock wow!
serene warren
Lurking
lapis hemlock
I am lurking today, by the way.
manic glacierBOT
trim elm
lurking
serene warren
Sounds good
manic glacierBOT
modern wing
Good afternoon all you wonderful folks! Happily lurking, and the doc is updated to reflect as such 
turbid radish
Lurking
lapis hemlock
@onyx hinge but I would like to see the pictures.
Someone has a micropython for FPGA.
micropython port, I mean.
tidal kiln
lurking
manic glacierBOT
modern wing
No Cylons allowed? Hmm.
uncut nexus
lurking today
onyx hinge
(javascript is #1, tho)
onyx hinge
This is too cute not to share again, re: circuitpython day -- https://cdn.discordapp.com/attachments/327298996332658690/738609940498743336/quetzalblinka.png "I present to you, Quetzalblinka, the mascot for CircuitPython Day Centralamerica."
turbid radish
👋🏻
modern wing
...what's sleep? 💤
onyx hinge
[my typos fix themselves in realtime awesome]
thorny jay
I am afraid @slender iron falled for the "analogue pocket"... (pure guess from fpga + GameBoy + waking up early).
slender iron
yup, exactly
stuck elbow
is lurking
onyx hinge
modern wing
I just looked at the Analogue Pocket -- all sold out. 😕
slender iron
ya, it went very fast
inland tusk
I am lurking
onyx hinge
thanks @inland tusk
modern wing
⚒️
solar whale
no anomalies to report
ionic elk
@slender iron glad you finally got a chance to try migen!
turbid radish
Take a hike Scott
silver tapir
🙂
onyx hinge
to ble or not to ble? why are you asking me?
https://adafruit-circuit-python.s3.amazonaws.com/index.html?prefix=bin/mpy-cross/ if you want mpy-cross bins for desktop linux, raspbian, windws, or macos catalina
idle owl
@silver tapir No objections of course! It's simply a matter of doing it properly. 🙂
Technically speaking.
onyx hinge
@silver tapir I think it's jwcooper on github who is kinda the guru of the website and may have opinions
kattni can correct me if I have the wrong name
idle owl
I can get in contact with him and sort out how to move forward.
slender iron
@rigid birch
silver tapir
thanks.
slender iron
np
thorny jay
@silver tapir I am very much interested in your journey to translating Circuit Python in Spanish. Trying to measure the effort and priority for translation effort. Like what is the corpus of learn guide to attack. Except for learn guide and error message, what else are you translating and why.
onyx hinge
thorny jay
Yeah, that SD card learn guide I readed and was very interested... @onyx hinge
idle owl
@thorny jay What lanuage(s) do you speak/write? We're in the process of bringing up a French guide translator.
thorny jay
French.
onyx hinge
Kattni's up next
gilded cradle
I need to learn KiCad soon because I have a project coming up where I'll need to design a board.
silver tapir
@thorny jay Sure. (in a sec...)
idle owl
@thorny jay The more help, the better. If after talking to fede2, you decide you're interested, let me know.
thorny jay
And I forgot to hug reports @slender iron for demonstrating modifying a board with KiCad in his stream.
ionic elk
@onyx hinge looks like there's just one more translation error holding up your SDIO PR
gilded cradle
@thorny jay I'll have to watch that
ionic elk
@slender iron it's not bad
lone axle
I don't hear any extra noise
serene warren
I hear a hmmmmm in the BG (only when @slender iron is on)
onyx hinge
some of us just hmmmm all the time
slender iron
it's a very loud hum
lone axle
thorny jay
@solar whale Didn't you not share with me the previous toolchain with me? I was worry that would be broken once CP migrate to gcc10. So I am very much interested in your tarball for Raspberry Pi.
idle owl
@silver tapir I will contact you tomorrow with how we'll get started on translating circuitpython.org. I'm off the rest of today following the meeting. Sound good?
silver tapir
It's the next topic for on the weeds, but yes.
onyx hinge
@solar whale assuming it's in the couple dozen - couple hundred megs range I can host it on unpythonic.net
thorny jay
For language selection, please do not use flags. Flags are for country. Country and language do not match in a 1:1 relationship.
prime flower
@sterile bronze Are you developing your own curriculum for the class or using the metrox guide?
sterile bronze
yes I am
thorny jay
I work in a place where we have 24(?) working languages...
sterile bronze
It's based on an in class course I've taught before
solar whale
@thorny jay just sent link
lapis hemlock
logging off now, so long everybody!
onyx hinge
see you @lapis hemlock
solar whale
@onyx hinge DM your e-mail address and I'll give you acces
thorny jay
We will see when a non alpha version will be out.
Today, alpha-x is their only option?
onyx hinge
When 6.0.0 comes out I suppose
sterile bronze
@prime flower Didn't you write the metrox guide?
ionic elk
Having worked with the unicode support in Circuitpython last week, unicode characters take about 2MB by themselves
lone axle
Is there any interest in a Wio Terminal "helper library" that sets up easy access to all of the hardware that is on that device?
silver tapir
I'll write to Electronic Cats for their ESP32-S2 feather, to see if they'd like a cpy version.
ionic elk
So Chinese character support requires a flash chip that's about 2x larger than what appears on most feather boards and other breakouts
lone axle
you cut out for just a second but mostly good
prime flower
@onyx hinge maybe it could be structured to encapsulate the topics on the sidebar, could be round-robin.
unless its more unstructured like a project hour?
onyx hinge
If anyone wants the gcc10 that jerry made https://media.unpythonic.net/gcc-arm-none-eabi-10-2020-q3-update-armv7l-linux.tar.bz2
prime flower
@ionic elk +1
gilded cradle
A hardware chat might be interesting
slender iron
bleak tiger
The show and tells are tricky for my timezone but I'd like to join something similar on Discord at a time similar to this meeting
modern wing
I do see where you're coming from -- we're getting some cross-pollination from personal projects & moderation topics. Mainly because we're (mostly) here all at the same time.
I'm OK with those things coming into the CircuitPython meeting. And it's something to think about spinning off into a community-focused event? I don't know what it'd look like...or if there'd be enough momentum & individual availability.
idle owl
They did a couple.
silver tapir
Yeah, I teach a class at S&T time. I've been trying to go to JP's which is a couple of hours earlier.
onyx hinge
Thanks Scott!
Meeting calendar: https://open-web-calendar.herokuapp.com/calendar.html?url=https://raw.githubusercontent.com/adafruit/adafruit-circuitpython-weekly-meeting/master/meeting.ical or add to your calendar app https://raw.githubusercontent.com/adafruit/adafruit-circuitpython-weekly-meeting/master/meeting.ical
Next exception to calendar schedule: Week of September 8 for US "Labor Day"
modern wing
Thanks y'all!
turbid radish
Thanks all
mental nexus
Thanks everybody!
gilded cradle
thanks
idle owl
Alright, I'm heading out. Have a lovely day!
modern wing
Cheers Kattni!
silver tapir
So yes @thorny jay. I've completed 6 translations so far, with the big ones like welcome-cpy it takes about 1-2/h day for a week, with the smaller ones is takes less time.
It's a bit repetitive even if I like and enjoy the subject I'm translating.
The corpus has been defined by kattni in #learn-adafruit-authors, but the best one to start is the welcome-cpy.
I can't recommend it enough, one of the guides already has +1.2K views so it makes me happy thinking about all of the people how have gotten to that content that wouldn't otherwise. And if at least a 10% got a board or purchased something from adafruit, that to me is worth the work.
thorny jay
Thanks, I did not knew about that channel. I have never wrote a guide so I don't know exactly if it is translator friendly.
silver tapir
And as a disclaimer, I was only paid for the first guide, pre-covid times. The rest of the guides and now the site, I've been doing it as a way to help the brand to get back up, and also for the non-english people to get a chance to have the fun we do.
thorny jay
I cannot be paid due to my work. So I can only contribute freely.
And I am no professional translator, just native speaker. And Circuit Python fanboy.
silver tapir
Yes, same here.
thorny jay
I did participate in CoderDojo as "trainer" and I know it is difficult to do anything without English... so we do what we have French material available, mostly Scratch.
The situation is even more difficult for me in Belgium that is a multi-lingual country...
silver tapir
Something that's also on my mind, is translating the makecode for adafruit. I tried the one for BBC Microbit, and all of the blocks are translated so their isn't even "for" in english.
thorny jay
So you targeted the CPX and welcome guide.
Yes graphical interface can be a good start.
silver tapir
I think in order, it was welcome to circuitpython first. Then the "piano in the key of lime", the makecode for cpx, the cpy made easy for cp*, and "last" the product guide for cpx.
Oh and also, using the learning guide is super simple. The first step I do is to copy the pages. Then translate-rename them. The configure the urls for the individual pages, and then start with the bulk of the text.
You move page by page from draft to published, and when done, the whole guide for moderation. So it's really easy to know what is done and what is missing.
So yes, I'll recommend it fully. Specially in covid-times that I can't do some of the outside project I'd be doing right now if I could.
thorny jay
Yeah, I saw the piano. I have spanish speaker co-worker. I shared your guide as they have kids and might want to try electronics/programming with them.
Did you also heavily participated into the translation of the error message? Or did you double check those. The idea is to have some consistency and make sure the same English word is translated the same way to Spanish.
silver tapir
For now we are using the same images as the english guides. The one that would really benefit from it is the makecode one, but since it's not translated on the page, I can do it yet.
Mu doesn't have unicode on the repl console, so having cpy in spanish at least is a bit ugly in mu, and most of the screenshots are from mu.
thorny jay
Thank you very much for your insight. I believe I should start by (1) Create a small project (2) Ask access to the learn system to publish it (that would let me learn about the learn guide) (3) Contact that belgian company/shop that resell Adafruit board and translate (or make original) guide in French ... they would be the one (that I know) with the best interest in translating.
uncut nexus
Is anyone else having issues cloning adafruit/circuipython.git? After running the submodule sync and update commands I am ending up with an empty lib/protomatter directory
silver tapir
Oh, and another thing, I also tried to get the weblate translation to close to a 100% before starting with the first guide.
@uncut nexus Let me try it here.
It could be that it downloads it later in the process. Just a sec, my download speed is kinda meh...
onyx hinge
jepler@babs:/tmp$ git clone https://github.com/adafruit/circuitpython
Cloning into 'circuitpython'...
...
jepler@babs:/tmp$ cd circuitpython/
jepler@babs:/tmp/circuitpython$ git submodule update --init
...
Cloning into '/tmp/circuitpython/lib/protomatter'...
...
jepler@babs:/tmp/circuitpython$ ls lib/protomatter
Adafruit_Protomatter.cpp Adafruit_Protomatter.h arch.h core.c core.h examples library.properties README.md
``` it's working here for me fwiwWhen I try re-running the submodule update command I get an error saying "reference is not a tree"
slender iron
Here is the notes document for 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/1G2UoQCcNsiacWqUn8z_783TkmnNg6ytaANyaeFx0coE/edit?usp=sharing
thorny jay
Oh, and another thing, I also tried to get the weblate translation to close to a 100% before starting with the first guide.
@silver tapir Yeah, I saw that, and you discovered that they were keeping adding new string and that once at 100% you can return to 99% without knowing it. I am going to read #learn-adafruit-authors for the last few month, I will learn a lot there.
uncut nexus
@onyx hinge - I notice that you are not using --recursive on the update command. If I leave that off I get "Unable to checkout '761d6437e8cd6a131d51de96974337121a9c7164' in submodule path 'lib/protomatter'"
onyx hinge
I think that at one time we advised to use --recursive but we no longer do.
git submodule sync
git submodule update --init
``` -- BUILDING.mdOK - It's still on Dan H's guide at https://learn.adafruit.com/building-circuitpython/build-circuitpython
still getting the "Unable to checkout error" - used your exact commands
onyx hinge
jepler@babs:/tmp/circuitpython$
``` that does not cause me to see the error you are reportingHmm. So - are you running on Windows, Mac, or Linux and what version of git?
onyx hinge
debian linux with git version 2.20.1
uncut nexus
My git is older - I will try to update & see what happens.
Yep - that was it. No errors with a newer git. Thanks for the help!!
silver tapir
@thorny jay Is probably not necessary that you read the backlog, is more coordinating talk. Talk to kattni, and with anneb's help in no time you'll be concentrating on content.
thorny jay
I guarantee it is interesting, it show the progress, the problem encountered, and more. 🙂
solar whale
@onyx hinge @tulip sleet can correct this, but I think the --recursive is NOT recommended for the git submodule update --init only for the git submodule sync --quiet --recursive
tulip sleet
@uncut nexus Did you see the git submodule update --init --recursive anywhere but in the decorative screenshot? I removed it a while ago in the shell ocmmands here: https://learn.adafruit.com/building-circuitpython/build-circuitpython#fetch-the-code-to-build-2986719-1
I didn't think anyone was reading the actual screenshot, but I will edit it 🙂
uncut nexus
I copied the code from that page a while ago and put it into a shell script on my local system.
So - I was using older stuff - my fault.
Interesting, tho, that the older git version I was using caused my issues. Not something I would have thought to check.
tulip sleet
now the screenshot is not misleading either; we removed that --recursive was a while ago, mostly due to tinyusb, which doesn't really want its submodules brought in recursively
what os version are you running? I guess I would have expected it to keep up, but if it's Ubuntu 18.04, it might be too old. I used a PPA for git when I was running 18.04 to make sure it was current.
uncut nexus
Scitentific Linux 4.7 - A redhat rebuild
Might have something to do with Connie being the lead architect for Scientific Linux for many years 🙂
tulip sleet
Scientific Linux release 4.7 is based on the rebuilding of RPMs out of SRPMs from Enterprise 4, including Update 7. It also includes all errata and bug fixes up until September 03, 2008.
tulip sleet
ok, that's a little better 🙂
uncut nexus
still old
tulip sleet
i think you should upgrade to 7.6
i used Scientific Linux at work when Fedora was too unstable; that was around 2009
uncut nexus
Well, I still gett all the security patches. But, since SL has stopped putting out new versions (no SL8 😦 ) I'll probably be moving the system to something like the latest Fedora at some point
tulip sleet
i decided i liked .deb over .rpm at some point, after spending too much time chasing down dependencies
but it definitely is STABLE
uncut nexus
Yeah, Redhat's big customers don't like rapid changes.
solar whale
If you are using RedHat , be aware of this https://arstechnica.com/gadgets/2020/07/red-hat-and-centos-systems-arent-booting-due-to-boothole-patches/
tiny oriole
If you are using RedHat , be aware of this https://arstechnica.com/gadgets/2020/07/red-hat-and-centos-systems-arent-booting-due-to-boothole-patches/
@solar whale i mean, technically they have patched every security hole right? haha
manic glacierBOT
ionic elk
@tulip sleet @slender iron I've got an API conundrum. The ESP32-S2, handily, has its own dedicated Remote Control peripheral (RMT) which is capable of sending pulse trains with a carrier frequency, as well as receive filtered pulse trains in return. However, this peripheral has nothing to do with the PWM peripheral (the LED_C). So passing in a PWM object as a parameter doesn't make any sense.
It'd be really kind of a shame to use the LED_C for PulseOut instead of the RMT just because of the API, since the ESP32-S2 doesn't have that many pwm channels or independent timers (only 8 and 4 total respectively) and it seems silly to ignore the dedicated remote control peripheral in favor of hard-coding it. But I can't really think of another good way around this API thing other than "faking" the PWM object by de-initializing the LED_C channel it uses inside the PulseOut constructor.
What do you think?
tulip sleet
@ionic elk so, it sounds like it might have been a mistaken assumption to pass a PWMOut instead of a pin to PulseOut. Is there a reason why PulseOut could not create its own PWMOut on ports where there is no specialized RMT? I.e., I am suggesting changing the API incompatibly. We can do this for major versions. For backwards compatibility it could take both for now, and we could drop it in 7.0.0.
ionic elk
I haven't used the PulseOut extensively so there might be edge cases I don't know of? But I don't see any reason why the PulseOut constructor couldn't simply be fed additional args for the frequency of the carrier signal. And PulseOut shouldn't need to care about things like duty cycle anyway.
So I'd agree
slender iron
that is an artifact of how it's commonly done. I'm open to changing the API
tulip sleet
the RMT can also do PulseIn?
ionic elk
Yes, they're both managed by the same peripheral. 4 channels total
each channel can do either pulseout or pulsein. One is being used for the Neopixel so we have 3 left
Should I change the API as a part of my PR, or should that be done separately as a preliminary step?
We could combine it with the PWMIO change we discussed as well
tulip sleet
i think you could expand the API, and throw NotImplementedError on ESP32STM when given a PWMOut.
ionic elk
you mean in ESP32?
tulip sleet
ooops, yes
ionic elk
Ok, I'll throw something together and we can hash out details in the PR conversation
tulip sleet
a separate PR is ok since it will be across ports, and will be easier to review
are you using RMT for neopixel_write in general, not just the status NeoPixel?
I'm not even sure you can do it the bitbang way that samd does, because of interrupts for radio. Or the memory-expensive PWM way that nrf does it
ionic elk
RMT is for neopixel in general
tulip sleet
excellent; i wish we had that periph on the other chips
ionic elk
I know right? I love all the ESP32 peripherals. In spite of their silly names
everything is so easy. I'm thinking of moving all my personal projects over to it
and the docs are great too
slender iron
my brain is pretty fried so I think I'll do the gcc10 monotony
bonus that it'll make things smaller
slender iron
just pushed more to jepler's gcc10 branch
manic glacierBOT
onyx hinge
I'm so glad our community doesn't allow bug discussions to go like this one .... https://bugzilla.mozilla.org/show_bug.cgi?id=675533 [W: foul language]
manic glacierBOT
manic glacierBOT
manic glacierBOT
manic glacierBOT
ionic elk
@slender iron couple of ESP32 structural questions for whenever you get online:
- Does ESP32 need
FULL_BUILDturned off? That flag is typically for flash size reduction, but the ESP32 probably doesn't need that, right? @tulip sleet maybe you were right that we should rename it to something likeFULL_SIZE_FLASHto prevent confusion. - Should
CIRCUITPY_ULABbe turned on? - TouchIO was turned off, but that just requires Microcontroller, right?
- Random probably shouldn't be turned on until TRNG is implemented, right?
- What does
CIRCUITPY_MODULE ?= nonedo?
slender iron
@ionic elk 1) ya we can turn it on if we habe all of the modules supported. 2) yup! 3) I think the esp has a touch controller so we could wait for that. 4) ya since that should be easy. 5) the ?= is assign if currently undefined
ionic elk
@slender iron ok cool. For the record I reworked the build flags earlier this year and we decided that FULL_BUILD should only be for reducing flash sizes. It doesn't pop up automatically on the website yet but I added a lot to the porting guide here: https://circuitpython.readthedocs.io/en/latest/docs/porting.html#porting
Since FULL_BUILD impacts lots of definitions like MICROPY_CPYTHON_COMPAT that really shouldn't be getting touched unless it's a SAMD21 trying to keep things under 256kb
I think TouchIO defaults to a bitbanged version if there's no common-hal version? So there's never any reason to not turn it on if you've created Microcontroller. But I should double check that.
I might take a look in the Litex port for stuff like this too, I think that also has a couple modules turned off that could be turned on
slender iron
@ionic elk that all sounds good. you clearly know it better than I do 🙂
solar whale
@ionic elk Doesn't the files system get wiped when the ESP32S2 is flashed at present -- so there is no code.py, correct?
ionic elk
I don't think it does? It keeps my code between loads
solar whale
OK -- I'll check -- I can only do that if I reload main first -- doing that now
ionic elk
Yeah try loading main and clearing it out
crimson ferry
@solar whale I was thinking about that last night, mine gets wiped too.
solar whale
yup -- no code.py present.
@slender iron I can build/run everything except samD51 with gcc10 ?? nrf52840,teensy41,stm32f405,samd21
manic glacierBOT
solar whale
@ionic elk built/flashed pr_3232 -- same crash...
ionic elk
welp, I'll keep looking into it
solar whale
Thanks -- just for fun, can you post a working .bin that I can try?
slender iron
@solar whale yup I saw. Will try and look today. Gonna tackle some WiFi stuff first
solar whale
@ionic elk aha -- I downloaded that and it fails the same way!! bizzare
My head hurts -- time to walk the dog -- before the Tropical storm hits .... back in awhile.
manic glacierBOT
manic glacierBOT
manic glacierBOT
onyx hinge
I hope I'm not wrong about this deeeee-licious binary size reducing fruit I just picked
onyx hinge
😅
solar whale
cool! will give it a whirl... I've broken everything else....
onyx hinge
tests are saying it's broken, maybe don't jump too fast (unix testsuite)
6 tests failed: builtin_float_round float2int_doubleprec_intbig float2int_fp30_intbig float2int_intbig math_domain math_fun_int
it may be something obvious, checking it out here too
solar whale
all that testing -- just going to find more errors 😉
onyx hinge
that's twice that the tests have caught my mistake just in this little PR, happy to have them. First was an unintional change in exception type, this one was matching a %q in a format string with a char* instead of a qstr...
tulip sleet
we wouldn't use the Arch build on GitHub actions, so I wonder if it makes more sense to test with the available preview download
it may not matter, but if there's a difference for some reason, it could be a timesink
solar whale
Doesn't PR 3231 use the 10.1.1 preview?
Has anyone else tried testing a SAMD51 gcc10 build on a Linux box?
tulip sleet
i haven't had time yet, but I could try it. I actually have had some weird SAMD51 issues while debugging bleio HCI with gcc9. If I turned off -flto, the USB did not enumerate. I got stuff like this in dmesg:
370979.084168] usb 1-1.3.4: reset full-speed USB device number 27 using ehci-pci
[370994.664111] usb 1-1.3.4: device descriptor read/64, error -110
[371010.276055] usb 1-1.3.4: device descriptor read/64, error -110
[371010.588061] usb 1-1.3.4: reset full-speed USB device number 27 using ehci-pci
[371026.128019] usb 1-1.3.4: device descriptor read/64, error -110
[371041.747976] usb 1-1.3.4: device descriptor read/64, error -110
[371042.059971] usb 1-1.3.4: reset full-speed USB device number 27 using ehci-pci
[371052.775945] usb 1-1.3.4: device not accepting address 27, error -110
[371052.975940] usb 1-1.3.4: reset full-speed USB device number 27 using ehci-pci
do you see that kind of error with the gcc10 build? If so, maybe it's not gcc10 per se, but some aspect of the compilation (-flto may have changed somewhat in gcc10, for instance)
solar whale
that was Ubuntu -- on Pi I see
ignore first post -- I see the same on Ubuntu and Pi
[779936.705348] scsi 7:0:0:0: Direct-Access Adafruit Grand Central M4 1.0 PQ: 0 ANSI: 2
[779936.705798] scsi 7:0:0:0: Attached scsi generic sg2 type 0
[779936.706214] sd 7:0:0:0: [sdc] 1048680 512-byte logical blocks: (537 MB/512 MiB)
[779936.706367] sd 7:0:0:0: [sdc] Write Protect is off
[779936.706368] sd 7:0:0:0: [sdc] Mode Sense: 03 00 00 00
[779936.706515] sd 7:0:0:0: [sdc] No Caching mode page found
[779936.706517] sd 7:0:0:0: [sdc] Assuming drive cache: write through
[779936.755314] sd 7:0:0:0: [sdc] Attached SCSI removable disk
[779937.821039] usb 3-1.4: reset full-speed USB device number 27 using xhci_hcd
[779938.026823] cdc_acm 3-1.4:1.0: ttyACM0: USB ACM device
[779938.028914] sd 7:0:0:0: [sdc] tag#0 FAILED Result: hostbyte=DID_ERROR driverbyte=DRIVER_OK
[779938.028922] sd 7:0:0:0: [sdc] tag#0 CDB: Read(10) 28 00 00 0f ff 80 00 00 08 00
[779938.028927] blk_update_request: I/O error, dev sdc, sector 1048448 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 0
what uses CountIO?
All of a sudden my compiler is complaining about it
Oh, I see, someone added it to FULL_BUILD
solar whale
If I leave it plugged into the Linux box, it keeps trying to access it producing more errors
@tulip sleet How is the wind in your area -- getting breezy here
tulip sleet
same here!
ionic elk
I have a wind turbine visible from my window and it's really going
solar whale
@onyx hinge So far -- I have just built the stock feather_m0_rfm9x with your size reduction PR -- now I need to see if I can remove a few things and freeze the rfm9x lib... Thanks -- I need all the space I can get.
But -- I may be done for the day -- it'll all be here tomorrow.... Have a good - rest of the day ...
ionic elk
@slender iron can we use GDB on the esp32 yet?
slender iron
I did manage to connect it on one of my livestreams
didn't really know how to use it though
ionic elk
Maybe I'll work on that a bit this week
would be nice to know
Especially for stuff like this null reference exception
slender iron
that's the correct timecode
the tools script has been enough for me when getting a load or store prohibited
turbid radish
?serverinfo
digital shoreBOT
adafruit
slender iron
cobalt grail
Thanks @slender iron !
onyx hinge
@solar whale ah right, if I can give you 1.5kB .. you'll take it
simple pulsar
Hello. I noticed nvm is listed for Gemma M0 in https://circuitpython.readthedocs.io/en/latest/shared-bindings/support_matrix.html but if I try it then I get ```>>> import nvm
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: no module named 'nvm'
Oh, I've just found it! >>> from microcontroller import nvm
onyx hinge
[nan, nan, nan, nan, nan, -0.909091, 1.0, -1.1, 1.21, nan, nan, nan]
``` I don't think this "optimization" of powf() is correct yet.
ornate breach
Yeah, I don’t think so
silver tapir
@solar whale ⬆️
Now I know why the neopixel PR fails only for you, and how to work around it. I hope you can get it working this way.
manic glacierBOT
installed
adafruit-circuitpython-clue_nrf52840_express-en_US-20200803-c394af4.uf2
(boot message "Adafruit CircuitPython 6.0.0-alpha.2-143-gc394af412 on 2020-08-03; Adafruit CLUE nRF52840 Express with nRF52840" )
there are glitches on the display
and eventually the screen becomes two white bars
low sentinel
When you assign a method to an object where does that land in C?
class c:
pass
def f(self):
pass
x = c()
x.extra_method = f # << ?
hmm mp_obj_instance_store_attr looks promising
manic glacierBOT
manic glacierBOT
thorny jay
Gameboy + FPGA + Midi ... This is for @tannew and could be done with his Circuit Python GB card?
https://medium.com/@AdilSoubki/gameboy-synth-module-1e2d0132d86c
manic glacierBOT
manic glacierBOT
ionic elk
@tulip sleet is it possible to build Circuitpython on 128kb?
tulip sleet
@ionic elk probably not; you'd have to turn off pretty much everything. Try turning off math and then every io but digitalio and see how big it is.
ionic elk
I'm more asking because the only F1 boards I have are Bluepills, and they all have 128k max
I'm regretting I didn't press this point earlier when the F1 thing was first brought up. Seems like virtually nobody makes any kind of eval kit for the more powerful F1s, presumably because they're overshadowed by the low-tier F4s, which are nearly as cheap in the US
I feel bad I didn't spot it before that poor guy did all this work for the F1 port. Should have tried to push him toward the F401 instead
tulip sleet
i forgot to turn off math in the comment above, still 163kB. Edited comment to reflect that.
ionic elk
@tulip sleet thanks for checking it out
tulip sleet
are there F1 chips that are > 128kB?
ionic elk
Yes, but what I'm finding is that they're basically obsolete in favor of the access line F4s
onyx hinge
ionic elk
The F401 is cheaper than the F103 for basically all equivalent and even lesser-power F103s
Which is probably why there aren't any dev boards for the high-level F103s - you get better prices, more ram, and faster speeds with the F401
F103s are very popular but it seems like that's restricted to the 64k and 128k range of SoCs
tulip sleet
we always felt that the SAMD21 was a stopgap before better chips were available, if we could have skipped that generation we might have done so, so similar here
slender iron
@thorny jay they link to my repo 🙂
manic glacierBOT
ionic elk
@tulip sleet yeah I mostly just feel guilty that I walked Dahanzimin through a lot of obnoxious port work without realizing that the outcome was so impractical
I assumed from the dirt-cheapness of the F103C8 that the higher ones would have similar price advantages but they really don't.
solar whale
@onyx hinge @tulip sleet was it really necessary to remove math from the small builds? I think a lot of sensor libraries need it.
tulip sleet
@solar whale we did not remove math; I was just talking about an experiment for the smallest possible builds above (would it fit on a 128kB chip)?
it is unrelated to the squeezing work we are doing
tulip sleet
🙂 np
ionic elk
@tulip sleet it's neat to know the 163kB minimum number, in any case
solar whale
I wondered where all that space came from ...
slender iron
@ionic elk there is some boards listed here: https://stm32-base.org/boards/
with the larger stmf103 ics
ionic elk
Hmmm. I don't see purchase links. I'll try amazon for some of these part numbers.
slender iron
I clicked through and ended at taobao
ionic elk
I see some stuff for Amazon but it wouldn't arrive until October
@slender iron what did you click through on?
slender iron
the stm32 base sent me to vcc-gnd and then the board name
ionic elk
I can't get any of these links to send me to a purchase page. Are we looking at the same thing?
ionic elk
I haven't bought stuff off taobao in a while. I need to make an account?
slender iron
I have no idea
lapis hemlock
@onyx hinge Jeff, if I want to have a function that prints out some information (i.e., this information is not returned per se), how do I do that? E.g., in void ndarray_print(const mp_print_t *print, mp_obj_t self_in, mp_print_kind_t kind) { (void)kind; ... , where does the *print pointer come from? A real-life example would be numpy's info function: https://numpy.org/doc/stable/reference/generated/numpy.info.html .
manic glacierBOT
lapis hemlock
Another, very common, example is the log messages of function fitting, and the like.
ionic elk
@slender iron what's the intended behavior supposed to be, when a null pointer is dereferenced? It seems like most boards just ignore it? This was definitely happening on STM32 but it wasn't noticeably affecting performance
slender iron
that usually causes a hard fault that leads to safe mode
it should be fixed so it doesn't happen
ionic elk
I'm super confused then, because it definitely wasn't causing a hard fault
manic glacierBOT
ionic elk
But I even put a tracking message in Main to confirm it's null and there's nothing afterwards that could be could be changing that pointer...
slender iron
I don't have enough context to give you a pointer
manic glacierBOT
ionic elk
@slender iron it's not a big deal, it's pretty much fixed as it is and if you have no complaints after reviewing I'm fine calling it. I did want to draw your attention to the "Code done running. Waiting for reload." message, though. Is there a reason why we only show it when serial_connected has failed?
Seems like that should just always be there when your code completes
slender iron
it should be printed when a serial connection is made
you don't want to send it if the code completes before serial is connected
ionic elk
Right now it's
bool serial_connected_at_start = serial_connected();
//run code
// Wait for connection or character.
if (!serial_connected_at_start) {
serial_write_compressed(translate("\nCode done running. Waiting for reload.\n"));
}
forgive me if I'm missing something, but doesn't that prevent it from ever displaying?
the "wait for connection or character" implies that maybe this was originally a loop or some other delay mechanism?
slender iron
you are right, it won't be printed if the serial connection is made while code.py is running
if a reload is triggered it will show up the second time
because run_code_py is called multiple times
ionic elk
Right. I asked because I've never actually seen the "Code done running" message
So I'm trying to understand what circumstances it actually displays and why it doesn't simply display every time
onyx hinge
@lapis hemlock I think technically you want to use MP_PYTHON_PRINTER if you don't have another print object handy. In CircuitPython we usually refer to &mp_plat_print which is probably not correct
ionic elk
If you meant that "Code done running" should show up the second time there is a soft-reboot, I haven't seen that happen.
solar whale
I shows up on the PyPortal display
onyx hinge
@lapis hemlock I think technically you want to use MP_PYTHON_PRINTER if you don't have another print object handy. In CircuitPython we usually refer to &mp_plat_print which is probably not correct
@onyx hinge OK, thanks! But if I do that, can have a single implementation for multiple platforms?
lapis hemlock
So, this last example should be universally applicable, right?
OK, I will look into that.
onyx hinge
seems like MP_PYTHON_PRINTER must work in circuitpython
this is in header mpprint.h as you may have already discovered
lapis hemlock
This should be OK: https://github.com/micropython/micropython/blob/f6375ac3ebac28656a0a757952d32c265b1ba7aa/py/mpprint.h#L42
onyx hinge
# define MP_PYTHON_PRINTER &mp_sys_stdout_print
#else
# define MP_PYTHON_PRINTER &mp_plat_print
#endif
``` has something to do with whether an optional feature is enabledIf you say it is safe in circuitpython, then I think we are done. Thanks again!
slender iron
@ionic elk what do you use to connect to the serial output?
turbid radish
?serverinfo
digital shoreBOT
adafruit
manic glacierBOT
prime flower
@solar whale @crimson ferry I'm going to take out an issue on our fork, this seems like it may fix some issues where we fail out on large files https://github.com/arduino/nina-fw/pull/50/commits/73375c37fcd33f254334eaaf31c3bea4182ebefd
slender iron
@ionic elk try screen if you haven't. it may be that your terminal emulator doesn't send the connect signal
slender iron
prime flower
woah - this might overtake the v3 poster for my favorite
raven canopy
that is some very sweet posterage!
forest laurel
Hi all! I am working on a CP project using ItsyBitsy M4 and so far things are working great. I am extending CP with my own C modules so I'm already building custom UF2 images ... I wonder, since I've seen lots of discussion about overclocking the SAMd51 (and I know you can in the Arduino environment) ... is it possible to build a custom CP image with a different clock rate than the stock 120Mhz?
slender iron
@forest laurel ya, it should be possible to change the clock rate in a custom build
the clock stuff is in ports/atmel-samd/peripherals
forest laurel
@slender iron Thanks Scott. I'll go poking around in there and see what I can find!
manic glacierBOT
forest laurel
@slender iron Good morning Scott! I have been poking around in the peripherals/ subdirectory as you suggested. Also researching a lot on the cortex (finding mostly stuff on the Arduino side), and trying to piece things together. From what I can gather so far (I'm on github/main), I'm seeing code that is basically using the DFLL (48MHz), dividing it by 24 to "channel 5" (so 2MHz), then setting DPLL0 to 120MHz by using a multiplier of 59 (+1 = 60 * 2MHz = 120MHz). If I wanted to boos to say 150MHz, I could just bump the 59 to 74 in "init_clock_source_dpll0"? Am I on the right track? (I am happy to just try it out, assuming I don't have too much risk bricking by Itsy?) 🙂
slender iron
@forest laurel yup, I think that's the right track
forest laurel
of course I know I'll need to update my custom C module which uses CYCCNT for some timing stuff ... but thats OK - just wondering if this is even possible 🙂 Also, aside from the obvious like neopixel_write - anything else obvious you could thing of that might be disrupted by messing with DPLL0 drequency?
slender iron
¯_(ツ)_/¯
forest laurel
Well ... my first attempts were successful, at least with my code. I don't have my full board avaialble with the I2C sensors I'm using, but the code runs with previously logged data just fine. I adjusted the CYCCNT timings in my parallel neopixel bitbang code and it is working prefectly. My "frame rendering rate" to an array of 8x96 = 768 neopixels in total went from 204FPS to 259FPS by bumping to 200MHz. Will play around with it and see what breaks if anything, check CPU temp at stock vs. boosted freq, etc.
But initial results are promising
(the increase is due to reduced cost to calculate the buffers, of course the timing to push to pixel array is fixed)
But I'm also slightly overclocking the LEDs since they are SK6812 and seem quite comfortable at 1.1MHz vs 800Khz
slender iron
@forest laurel nice!
forest laurel
I'm really not sure what else should change for consistency (microcontroller.cpu.frequency still reports 120MHz, for example), or function (does SPI clocking depend on DPLL0?), etc. - but it is a start.
I would be happy to continue work on this with a goal of eventually contributing back a patch which would allow overclocking similar to the options added to Arduino the Adafruit products
slender iron
@forest laurel making it configurable would be amazing!
@prime flower this look familiar to you? https://forums.adafruit.com/viewtopic.php?f=60&t=168083
prime flower
@slender iron Possibly they're using an older version of ESP32SPI_Socket which used write instead of send?
I'll jump into that thread if that's OK
slender iron
that'd be awesome. thanks!
prime flower
Posted up.
simple pulsar
I've noticed CircuitPython makes a new copy of a bytes object if you do c = bytes(a). I've tested a few versions of big python and it doesn't do this. Is it suppoesd to do that?
manic glacierBOT
silver tapir
My local hardware provider just gifted me a sparkfun redboard with a cortex m4 module called "artemis".
But sadly, it has a serial programmer instead of native usb, so no cpy here 😦
I'll research later if there is a chance to program the module directly like in the esp32-s2, but I'm guessing that if they could, they would have saved the money on the serial chip.
mental nexus
Along the vein of adding a memory-saving way to handle text labels using bitmaps, I’m interested in extending Bitmap.c to copy a rectangular glyph bitmap into a bitmap. I have it running using Python pixel by pixel copy but it’s really slow relative to the existing label library. I’m totally new with updating CP, could I ask someone a few questions and suggestions on how best to tackle this?
mental nexus
Or if there are already other bitmap manipulation functions that I am unaware of, I’ll gladly use them.
manic glacierBOT
manic glacierBOT
Tested the feature on a nrf52840 m2 devkit, here is a current graph
The code is:
import time
@micropython.asm_thumb
def mem(r0):
ldr(r0, [r0, 0])
@micropython.asm_thumb
def mem_write(r0, r1):
str(r1, [r0, 0])
peripherals = {
'RADIO': 0x40001000,
'UART0': 0x40002000,
'UARTE0': 0x40002000,
'SPI0': 0x40003000,
'SPIM0': 0x40003000,
'SPIS0': 0x...
simple pulsar
I put in https://github.com/micropython/micropython/issues/6320 for the bytes() issue.
onyx hinge
OK that's a super cool idea bleeptrack. Thinking about what modern PCB manufacturing makes "feasible".
mental nexus
Is there a learn guide reference for the overview of what is “shared-modules” vs “shared-bindings”? I’m trying to add a new function and don’t fully understand how to get the hooks from C back into python.
I’m following this PR as an example, but don’t understand what is going on in shared-bindings/displayio/Bitmap.c https://github.com/adafruit/circuitpython/pull/2756/files
Ok found this. Will review and then post if it’s still unclear:
https://learn.adafruit.com/extending-circuitpython/inside-the-virtual-machine
stuck elbow
@mental nexus in short, shared-modules has the code that makes things happen that calls internal APIs, and shared-bindings exposes that to the python side
mental nexus
Thanks @stuck elbow. Now to understand how to modify these incantations properly. One more question. How do I specify these bindings so it is a function than can be used like mybitmap.dosomething(a,b,c) versus dosomethjng(mybitmap,a,b,c). Even better can you point to an example that has both of these?
stuck elbow
You mean you want a method on an object? Is the object created in C code or in Python?
displayio has many examples of objects with methods on them: https://github.com/adafruit/circuitpython/blob/main/shared-bindings/displayio/Display.c
here is a simple method that takes one argument: https://github.com/adafruit/circuitpython/blob/main/shared-bindings/displayio/Display.c#L262-L269
and then it's added to the type's locals dict here: https://github.com/adafruit/circuitpython/blob/main/shared-bindings/displayio/Display.c#L480-L494
mental nexus
Ok so in the locals dict, each class function gets an entry like this: { MP_ROM_QSTR(MP_QSTR_auto_refresh), MP_ROM_PTR(&displayio_display_auto_refresh_obj) },
And what is this info up at the top in “make_new”:
{ MP_QSTR_auto_refresh, MP_ARG_BOOL | MP_ARG_KW_ONLY, {.u_bool = true} },
Is this the default values for the function arguments?
Oops that is just the init for the class.
Thanks for the introduction to this. I will dig in a bit more and see if I can fight through it.
stuck elbow
that's argument parsing
for functions that take keyword argument or variable number of arguments
mental nexus
Thanks for pointing me in the right direction. Definitely a lot to absorb with all of this. I’ll see if I can get something simple working and then go from there!
onyx hinge
hmmm .. should I (A) re-spin this board to have an itsybitsy soldered into it or (B) stick with the trinket, but make a custom build that enables long ints?
given that it's not a product, (B) probably is the way to go
mental nexus
Trying to expand a CP library with a new function. It has 8 input variables. It looks like I need set the shared bindings with MP_DEFINE_CONST_FUN_OBJ_KW(displayio_bitmap_insert_obj, 1, displayio_bitmap_obj_insert);
But what does the second parameter (a number) specify? Where can I look to understand how this is defined?
onyx hinge
I didn't initially think about how this process of computing si5351 clock settings would require more precision than floats, but when dealing with ratios you end up with values like 915000000000/27923584
so in this case the 1 is the minimum number of arguments that is required, I guess?
mental nexus
I see a lot of functions do some kind of validation of the arguments. Where I can I find what these arguments mean and how to check my arguments? This is not what I'm working on, but an example from Display.c enum { ARG_target_frames_per_second, ARG_minimum_frames_per_second }; static const mp_arg_t allowed_args[] = { { MP_QSTR_target_frames_per_second, MP_ARG_KW_ONLY | MP_ARG_INT, {.u_int = 60} }, { MP_QSTR_minimum_frames_per_second, MP_ARG_KW_ONLY | MP_ARG_INT, {.u_int = 1} }, };
Seems like this is dealing with optional arguments also.
onyx hinge
again, not a lot of documentation. The things you can ask for are: bool, int, obj. additionally, it can be required or it can be kw_only. ```typedef enum {
MP_ARG_BOOL = 0x001,
MP_ARG_INT = 0x002,
MP_ARG_OBJ = 0x003,
MP_ARG_KIND_MASK = 0x0ff,
MP_ARG_REQUIRED = 0x100,
MP_ARG_KW_ONLY = 0x200,
} mp_arg_flag_t;
and the last thing can be a u_bool, u_int, or u_obj to match the MP_ARG_<type> and give a default value if not MP_ARG_REQUIRED.
mental nexus
How about the MP_QSTR_target_frames_per_second can that be any name or does that reference somewhere else?
onyx hinge
Yes, it can be any name, you can make up new ones. If you do, and you get an odd error during compile, try doing a "make clean"; the build system tries to keep the list of MP_QSTRs up to date but doesn't always succed in some subtle situations.
mental nexus
Thanks for the information, I'll dive in again and see if I can make it work. Also, for the benefit of my future self, I'm taking notes on this complicated scavenger hunt!
ruby atlas
@onyx hinge I'm looking into what it would take to make _pixelbuf be able to take ulab arrays. The first thing that comes to mind is to make pixelbuf accept the datatypes from ulab, the second is to make pixelbuf accept an array of len == 3 * n or len == 4 * n (when in RGBW mode or Dotstar mode).
It is already easy (though I'm not sure how expensive/fast) to create a ulab array from pixelbuf.
onyx hinge
that would be neat
solar whale
weird -- github just told me I could not comment "at this time" then on retry accepted the comment...
manic glacierBOT
mental nexus
@onyx hinge, thanks for the guidance this morning. Thanks to your inputs, I succesfully made a rudimentary bitmap slice copy. It has a huge performance impact on what I was trying to do (copy font glyphs into a larger bitmap) even with a simple strategy. I added some notes to the related issue #3004.
onyx hinge
Cool, I'm glad my little hints were worth something to you!
mental nexus
Your not so little hints pointed me in the right direction. This code is pretty difficult to piece together for me so a little bit of a roadmap helped a lot. I can’t say I understand everything I did so far but at least this is a first step in that direction. And even minimally-working (albeit far from elegant) is a big step and good positive reinforcement on a Friday!
Thanks a lot @onyx hinge and @stuck elbow
ruby atlas
from _pixelbuf import wheel as colorwheel
with timer():
buf = ulab.linspace(0, 256, 256)
vec = ulab.vector.vectorize(colorwheel)
for n in range(256):
pixels[:] = vec(buf)
pixels.show()
ulab.numerical.roll(buf, 1)
with timer():
for n in range(256):
pixels[:] = [colorwheel((i+n) % 256) for i in range(256)]
pixels.show()
Took 3.28809
Took 4.15796
ulab version is faster!
ornate breach
<@&327289013561982976>
tulip sleet
thanks @ornate breach
ornate breach
🙏
manic glacierBOT
orchid basinBOT
orchid basinBOT
fringe trench
Sorry I have been out of touch, lost power in the storm Tuesday, still out.
idle owl
@fringe trench Sorry to hear that. I hope you're staying safe.
fringe trench
I am. I offered to help with Circuit Python and promptly disappeared, so since my phone is currently charged and mobile data is ok, dropping a note to explain why.
idle owl
@fringe trench No worries. The important thing is that you're ok. We'll be here when you return.
orchid basinBOT
solar whale
@tidal kiln Should we wait for additional review/comment or would you leke me to merge/release the rfm9x change today.
tidal kiln
meh, i'd say go ahead and merge/release. you tested 3 cases, i tested 1, seems adequate.
ivory yew
oof, anyone ever bricked a SAMD21 board? I can't connect to it with openocd anymore.
ivory yew
I hope not
ivory yew
yeah I guess I did. 😦
orchid basinBOT
manic glacierBOT
slender iron
@ivory yew I did have to use atmel studio once to get one back
were you doing pin muxing stuff? it can be particularly hard to get debug access if you switch the mux away from swd
ivory yew
lol
yeah I imagine so
I managed to bring it back from the dead by basically doing a timing attack on it and resetting it until OpenOCD connected.
My problem was that I set a peripheral's clock config to a GLCK that wasn't configured and then tried to write to a synchronized register.
that apparently is very bad.
slender iron
ah ya. maybe hanging the memory bus then 🙂
glad you got it back!
@worn lark this is a better channel to ask about the RGB status pixel because it'll require some internal changes to work
lone axle
It seems there may be an issue with the system that aggregates the links for https://circuitpython.org/contributing
raven canopy
yeah, i saw it earlier and DM'd kattni. i'm not sure what the issue is, as all of the recent CI/Jekyll builds passed. 🤷
mental nexus
@idle owl I’m working with @lone axle on some updated examples with the display_text library and we’d like to add some font files for the examples. What kind of license issues do we need to consider? Is there someone that vets these kinds of things for inclusion into the libraries?
lone sandalBOT
manic glacierBOT
hybrid scarab
👀
idle owl
@mental nexus That is the point of PRs, or at least the intention of PRs, for someone else to catch things like this. That said, things get missed. I'm not sure what kind of licensing comes with font files. I'll have to look into it, and get back to you.
hybrid scarab
Aren't fonts ostensibly not-copyrightable? Though putting that to the test would be... fun
mental nexus
Ok, thanks kattni. We’ll include some fonts in a PR and see how the review proceeds.
hybrid scarab
Does the "fonts" Python module work here, or are these extremely display-specific?
Actually I guess CircuitPython is a different world WRT to resource distribution
idle owl
@hybrid scarab In theory, yes, but licensing isn't necessarily copyrighting.
hybrid scarab
True!
lone axle
I have run across some fonts in my work that were definitely locked down in some way and came with hefty fees to use depending on how they were used. I'm not sure of the specific legal mechanism in use though. The ones we are looking at here are much more common and seem to be open for use as far as we can tell.
hybrid scarab
Not sure if it helps or is relevant but I've collated these mostly under OFL - https://github.com/pimoroni/fonts-python
No complaints... yet
lone axle
Ooh that is good to know. Thank you 😄
There are a few here: https://www.theleagueofmoveabletype.com/ that are all open sourced.
hybrid scarab
Oooh! Thank you
I think a handful of people other than me actually use my shove-fonts-on-pypi-because-life's-too-short packages, but it would be nice to add these
crimson ferry
Doesn't address the permissions issue, but perhaps useful to know that there are a bunch of monospace fonts included (linked) within the circuitpython repo https://github.com/adafruit/circuitpython/tree/main/tools
hybrid scarab
AH! That's how you do a reuse license for a single file- I had an svg earlier that I just pasted some HTML comments into
lone axle
whoa
are all of those usable within the system somehow? or you'd need to make a new build to utilize them?
crimson ferry
I'm really not sure. I just needed some fonts in the early days of CP when there were only a couple in learn guides, and came across those.
hybrid scarab
That Tecate/bitmap-fonts repo is a treasure-trove
crimson ferry
I tend to use terminus-font-4.39 bc it has a good range of sizes from 12-32 pt
hybrid scarab
In a lot of use-cases I need just one good vertical size, without any weirdness, and with PIL-compatibility. Might have to look over these.
crimson ferry
Sometimes I'll even automatically switch to a smaller font if some dynamic text is bigger than my Label 😉
hybrid scarab
e-Ink being the exception
ionic elk
Would anyone have a moment for me to rubberduck at them about PulseIn on ESP32-S2 today? I'm really bogged down with the FreeRTOS ringbuffer and could use a second opinion.
thorny jay
Wait, is it @hybrid scarab on Adafruit Discord server? Are you going to join for the meeting? I need to update my Hug Report...
hybrid scarab
Ahaha! Hello there @thorny jay
I've been lurking here for a long time, but it's an intimidating server 😆
thorny jay
About font and copyright... from usenet reading 25+ years ago, low res bitmap font could not be copyrighted, but TTF or vector font could. At the time I was collecting and distributing a collection of link to cyrilic font, both for DOS(?) and Windows.
But I don't know if this was tested in court since then.
hybrid scarab
Still not something I'd want to test in court against any significant font publisher though- they seem to enforce their own interpretation
thorny jay
There are enough copyright free font that are meaningfull for Circuit Python typical screen resolution.
hybrid scarab
Would be nice to collate those bitmap fonts which have permissive licenses, I find myself duplicating effort in that regard a lot
When I start porting our other boards over to CircuitPython there will be some that need specific fonts, too
thorny jay
If you search for font to use in CircuitPython... there are a lot in various learn guide (go for PyPortal guide) in here: https://github.com/adafruit/Adafruit_Learning_System_Guides
There are a lot of duplicate in that repo, because each project try to use one that best fit their need.
So putting together those already verified font would be a big progress...
Should that be "In the Weed".
At one time, I started to group them together to have a collection to choose from when in need.
ivory yew
@hybrid scarab definitely intimidating if you don't mute a ton of the channels
thorny jay
I am pretty sure those bitmap font are generated from a vectorial source and made to the size required.
ionic elk
@slender iron you prepping for the meeting or can I ask you a couple quick questions about found_main and the other run_code_py setup code?
slender iron
I'm in a separate meeting atm
ionic elk
ok, I'll ping you after the main meeting is over
hybrid scarab
Oh boy converting physical_feather_pins into a CookieCutter-based library has been... oof
thorny jay
So you plan to keep using "physical_feather_pins"?
Do you have other library relying on that?
hybrid scarab
At least until something mirroring its functionality is merged upstream- although tbf I don't know how much relies on it
thorny jay
I believe it is only the Enviro+ library.
Now that lrt559 is solved, I think all the other you use the adafruit supported library...
hybrid scarab
You might be right 😬
hybrid scarab
No secrets on the CircuitPython front, still reeling from this 😆
It made a good, self-contained project for Nat to do when he was interning for us- allowing me to forge ahead with my workload without having to think about it. Unfortunately that's now coming back to nip at my heels
I'd like for us to do Breakout Garden FeatherWing, but there's a lot of software support that has to happen before that's feasible
thorny jay
Is this a secret project: https://twitter.com/DavidGlaude/status/1291839360887853057
hybrid scarab
Not that secret if it's on my Twitter 😆
But yes we have Stemma/Qwiic to Breakout Garden adapters in the works- actually really handy!
thorny jay
How many Breakout Garden do you fit on a FeatherWing ?
thorny jay
Wondering where the "reverse insertion" protection take place... is it on each breakout?
hybrid scarab
Yes, the breakouts are protected themselves
thorny jay
I think you could fit 3 but the middle one would need to be "not on the same row".
So maybe 2 I2C and 1 SPI in the middle.
lone axle
Interesting. I hadn't seen these Breakout Garden devices before.
hybrid scarab
That's because they're so awesome their majesty has blinded you 😆
thorny jay
This is a pre-Stemma solution to make sensor that are modular but can also be used as breakout board and soldered.
hybrid scarab
Those encoders and pots have a Nuvoton 8051 1T driving them, for fun
thorny jay
I believe the Rotary POT is not an announced product yet...
lone axle
Oh wow. I'm definitely going to have to get an order in. 😻 @ the RGBW Trackball breakout.
hybrid scarab
Yeah the edge-connector you stuff into a socket works really well- although unlike Stemma soldering it... kinda breaks that
Also WIP. We don't tend to have SUPER SEKRETS, but we don't shout too much about stuff like this in case it never gets released
The APDS 9500 is a particularly troublesome sensor
thorny jay
Adafruit use the APDS9960, standalone and on the CLUE.
Wait... I know rather well the Pimoroni product line... and the Adafruit recent (2-3 years) product. Does that make me an expert?
vivid temple
is the esp32-s2 feather already available to buy? can’t find anything
if not, what are the alternatives?
hybrid scarab
@thorny jay insofar as I'm aware the 9960 and 9500 are frustratingly unalike, and the 9500 datasheet is full of "here be dragons", where it mentions initialising undocumented registers to undocumented values.
lone axle
Those are he ones currently available that I am aware of. There may be others that I don't know about though.
thorny jay
Here is a comparaison of both... https://pdfs.semanticscholar.org/41af/aea923aafb3d37bfeabb318dd3ace870d36c.pdf
tulip sleet
we are using the Saola WROOM and WROVER boards. Get the one with PSRAM (WROVER, I think).
thorny jay
ESP32-S2 you have one by Unexpected Marker but very few were produced, and one from that Cat company in Mexico that might be supported in the next 6.0.0alpha.
And make sure you have an antenna if you take the one with the antenna connector.
vivid temple
okay, thanks guys - wifi already working?
thorny jay
@slender iron is working on it... you should watch his video, or follow the meeting in 24 minute for the lattest news.
vivid temple
24 minute?
slender iron
it is soon
thorny jay
Last thing I have seen in ESP32S2 the scanning was working...
But I did not watch the full video and that was 4 days ago.
vivid temple
that’s a youtube live broadcast in 22 minutes?
lone axle
it's in the voice channel here in Discord
vivid temple
ah okay
vivid temple
@slender iron are you going to implement esp-now as well?
vivid temple
sure focus to get the main things working first 🙂
but if there is time later on ... 😛
slender iron
my goal is to get more folks working on the port than me
vivid temple
i have a bin for micropython with esp-now working. They just haven’t accepted the code in their main branch because of the style
lean marsh
is there an off topic or general chat here?
slender iron
@vivid temple I'm happy to help you get it into circuitpython
@lean marsh #general-chat or #general-tech
idle owl
<@&356864093652516868> Hello! CircuitPython Weekly meeting in 15 minutes. Please update the notes document with your Hug reports and Status updates (regardless of whether you're attending the meeting), or let us know you're lurking. It's super helpful. Looking forward to seeing everyone soon! https://docs.google.com/document/d/1G2UoQCcNsiacWqUn8z_783TkmnNg6ytaANyaeFx0coE/edit#
thorny jay
@idle owl I would love if you could have a second look at my UART enhancement for the PM25 sensor: https://github.com/adafruit/Adafruit_CircuitPython_PM25/pull/6 . This would be handy for usage in Feather Wing Enviro+. Or maybe @hybrid scarab can have a look at it and see if more things can be ported from Pimoroni Library to Adafruit Library?
hybrid scarab
It's on my TODO list- right now I'm calling it a day, not been the picture of health lately
idle owl
@hybrid scarab Hope you feel better and get some needed rest.
@thorny jay I'll try to get to it this week. Added it to my list.
thorny jay
I don't know for UK... but it is hot in Europe, at least where I am.
hybrid scarab
Thanks @idle owl
thorny jay
For health reason, I don't have to (physically) go to work tomorow... not because of Covid... but because of the heat wave and because the air conditioning is not working well in my building!
serene warren
Lurking - noted in doc
prime flower
Lurking as well
onyx hinge
.. grabbing the list of lurkers ..
modern wing
Good afternoon all you wonderful folks -- happily lurking, as noted in the doc
turbid radish
Lurking today - if you have newsletter content, ping me or Tweet to @anne_engineer
thorny jay
Since my microphone works... maybe not lurking after all.
uncut nexus
Lurking today, thanks!
bleak tiger
@thorny jay yeah super hot, had highs above 35 for the last few days!
onyx hinge
@thorny jay mute for now though?
It feels like a short meeting notes doc but .. it's still 14 pages 😅
thorny jay
I don't care if it's 35C° in the day... but it should go below 20°C during the night.
bleak tiger
Right!
onyx hinge
@lone axle I can grab links today unless you want the honors again
slender iron
@bleak tiger if you want to speak in the meeting we'll need to add you to the circuitpythonistas role
thorny jay
That will be hard...
bleak tiger
Cool, nothing to add today but I'd like to be added for future meetings?
tidal kiln
lurking
onyx hinge
@bleak tiger OK, let us know -- we can also add you if you want to get Discord reminders for the meetings.
bleak tiger
Oh that'd be useful, please add me
onyx hinge
@bleak tiger done! and welcome!
sterile bronze
lurking
lone axle
@onyx hinge I can get them 😄
onyx hinge
Yup hope you like it
bleak tiger
Cheers
turbid radish
Purple
serene warren
lurking
onyx hinge
@lone axle it's your new semiofficial role then
inland tusk
lurking
onyx hinge
[I'm just going to assume that illegible user name is lurking, let us know if you're a legitimate person]
slender iron
🙄
onyx hinge
(we have merged 39 pull requests and a total of 178 commits since circuitpython alpha 2)
ionic elk
@gilded cradle that's cool!
gilded cradle
Thanks
ionic elk
What kind of dev boards do they have for the MP?
gilded cradle
I have the STM32MP157C-DK2 which has a display and it's running Linux.
lone axle
errant grail
Late and just listening today.
ionic elk
@gilded cradle how would you rate it alongside other boards in terms of value? Seems like a good tradeoff of features vs price to me, but I don't have as much experience with SBCs.
hybrid scarab
I'm lurking and just listening
gilded cradle
Pretty good @ionic elk, but it's designed for more advanced users for sure because you either need to compile packages or you need to compiler a firmware image to get it working with Blinka. I'm currently working on trying to find/create an easier path for new users.
ionic elk
That's great to hear. I have a friend who's working with linux on STM32 stuff but I think he's trying to do it off an H7... does the MP even run on an ARM core?
slender iron
MP1 is a cortex-a core iirc