Clear benefits for (poorer) countries to join the EU.
www.euronews.com/business/202...
#eu
@decdiv.bsky.social
Clear benefits for (poorer) countries to join the EU.
www.euronews.com/business/202...
#eu
Lots of package updates after the SDK changed `--on-error` to `if-error` (and yes: there is a logic behind that change).
I hope I have updated all packages under my control.
#toit
Windows...
When doing a `stat` on a directory that is in the process of being created/removed one can hit a permission-denied error.
I just implemented a directory-based lockfile approach, and hit this issue with Windows. Now I need to retry when hitting that error...
#toit #toitlang
Toit v2.0.0-alpha.189 has been released.
Enjoy!
#toitlang #toit
Good progress on my lockfile package. Should be ready for a release as soon as the Toit SDK and host package have been updated.
This was the last major blocker for a pure Toit-based package manager.
#toit #lockfile
Just fixed an issue with BLE notifications. NimBLE was running out of resources, and instead of waiting and retrying we just gave up.
With that fixed my HID keyboard library seems to work nicely.
#toit #ble
Don't know what I would do without my oscilloscope.
Just looked at an issue where a modem would wake up when the ESP32 went into deep-sleep. Turns out, there was a 10us spike on the modem's reset line during shut-down. These kind of things are almost impossible to debug without an oscilloscope.
This has cost me soo much time:
> For data transferred from the HID Device to the Report Host, the Report ID is prepended to data received by the Report Host [...] before being passed to a USB HID Class driver.
Playing around with BLE HID.
On the one hand quite simple, but on the other hand finicky. Very frequently I have Android that works fine, but not Linux, and then Linux that works, but not Android...
#toit #ble
DSB (Danish railroad operator):
===
[...] buy your tickets on our website [...] - these tickets cannot be transferred to the app, as they are two different systems.
An app ticket is a digital ticket and a print-it-yourself ticket is a pdf file [...].
They can't be synchronized.
===
#dsb #fail
We just got the confirmation: Toit will have a booth at Maker Faire Hannover 2025 (August 23-24).
If you're attending, make sure to drop by!
And don't forget "\r\n" line endings.
29.05.2025 14:27 β π 0 π 0 π¬ 0 π 0TIL that tar archives can fail to extract on Windows if the target of a symlink appears later than the actual symlink (inside the same archive).
28.05.2025 14:48 β π 1 π 0 π¬ 0 π 0It's amazing how much effort goes into supporting Windows. `\` as delimiters for paths; a different way of spawning programs (without a way of passing arguments as array), ...
And today I spent the whole day trying to get Toit to work nicely on Windows systems with non-UTF-8 code pages.
Managed to create a debian package for Toit today. Hopefully the next release will build it automatically.
I also looked into a homebrew package, and that looks good too. (Still needs more testing).
Toit v2.0.0-alpha.180 was released.
See toitlang.org (latest news) for more details.
Just implemented RMT encoder support for Toit.
Users will be able to provide a start|between|stop-sequence (where "between" is between bytes), and then sequences for 1/2/4 bits.
Just implemented a UART using this approach. Pretty neat.
#Toit #RMT #ESP32
Still working on rewriting the RMT driver of Toit to the "new" API of the ESP-IDF.
I believe I made a bit of progress today. Pretty sure I found a race condition in the ESP-IDF driver.
#Toit #RMT
Shootout to GitHub. Our GitHub account had unpaid invoices from the previous owner. We filed a support ticket, and GitHub voided the missing payments without any issues. Thanks!
#GitHub
My external USB drive is acting up.
I used `UDF`, so it's platform independent. I'm regretting that decision now. The `fsck` for UDF consists of `int main() { return 0; }`. :(
github.com/pali/udftool...
Redbat monitor arm for laptops. Can be fixed on a desk. Swivels out of the way when not needed.
Bought two of these arms (designed for laptops) for my desk. One for my oscilloscope, and one for my soldering station.
They are now out of my way when I don't need them, but still easily accessible.
Found two issues in the i2c slave implementation of the esp-idf today. Explains why I couldn't get the Toit implementation to work.
Hopefully progress will be faster with those out of the way.
github.com/espressif/es...
github.com/espressif/es...
#Toit #I2C #buffer-overrun #ESP-IDF
Currently using wedrivers to write a test for some code that has a web-page.
For debugging I'm not running the browser in headless mode, and I have to say: it's pretty cool to see the browser change in front of me.
#webdriver
Wasn't really on the top of my TODO list, but couldn't stand it anymore. Just refactored a small portion of the host package. The 'fork' function should be much easier to use once my PR lands.
#Toit
Our sign-server (running locally at my home) isn't super stable. I think it's because it's running from an USB drive.
I just took a Toit-plug (smart-plug with an ESP32 inside) and wrote a small Toit program that checks the health periodically. If the server isn't reachable the plug resets it now.
'ping www.google.com' with response times of up to 4.8s.
On my way back from a great Hackaday Berlin 2025.
Internet on the train is a bit slow. I think I will just roam.
#hackaday-europe #ping
Just committed a flag to allow connections to TLS servers without validating the certificate.
Generally that's a really bad idea (due to mitm attacks), but validation is expensive, and sometimes it's just not worth it. While not recommended at least there is an option now.
#Toit #TLS
Tried to write a test for the updated i2c driver.
Somehow didn't want to work...
Turns out that the ESP32 silently used a 58x faster frequency than I specified.
github.com/espressif/es...
The lowest supported I2C speed is ~3kHz.
#Toit #I2C
oops.
Just accidentally turned off Toit's Windows signing server instead of turning on the light...
On a related note: I just discovered that I had forgotten to 'systemctl enable' the sign-service.
Interesting.
The RMT is faster on the ESP32 than on the ESP32S3. On the ESP32 I can easily emit pulses at the frequency of the APB (80MHz). On the ESP32S3 I have the following requirement:
3 Γ T_apb_clk + 5 Γ T_rmt_sclk < period Γ T_clk_div.
That's at least 3 APB ticks.
#ESP32 #ESP32S3 #RMT