Arduino in Go and Go in Arduino
We started the day 2 in Golang room where Ron showed how to code Arduino in golang. They provide a framework that can be used for not only simple, but also a bit more complicated apps. He showed few practical demos from blinking led, through button-controlled lights, to Linux (busybox) based dron simulating solar system repair. I must admit it was really cool to see a linux controlled dron flying above our heads, doing a flip and that all communicating with a base via wi-fi.More info: https://fosdem.org/2017/schedule/event/go_gobot/
Hacking cars to make them drive alone
A similar talk, this time not in golang room any more, was Open Source Car control introduction by Josh Hartung -- they buy a cheap Kia car and put an Arduino, inside. They rather spoof wheel assistant rather than creating new complicated controlling.. so they basically do a slight man-in-the middle attack. The reason is that when anybody needs to do something with a car, he needs to hack it, and the only standard is the OBDII, which does not offer what is needed.Safety can be done by different ways, the safest so far seems to have strict policy, that touching a driving wheel or break hard resets the system, which puts a car to the stock settings.
For assuring good enough product quality, they use coding standards, SW diagnostics, code review process and unit integration testing.
Autonomous car racing looks like seniors driving (approx speed just few km/h), but it looked interesting, given those are not cars produced with milions given into research. And waht is the best, anybody can join and hack own car, the cheaper the better.
More info: https://fosdem.org/2017/schedule/event/open_source_car_control/
Rest of the day in Python devroom
Brian Bouterse explained how CPython (the runtime) can be debugged using gdb to find some issues in the runtime itself. However, GDB can also work with the python code and corefile in a bit smarter way, so users only see python stack functions for example. And why should we use debugger for python? For example if you have some problem that is hard to debug or you cannot share reproducer to the upstream. We can make the Python better by fixing such corner cases. It's a powerful technique, that every Python developer should know of.More info: https://fosdem.org/2017/schedule/event/python_gdb/
Threads are goto of our generation, mentioned Ewoud Van Craeynest in his talk Asynchronous programming with Coroutines in Python. It does not mean we should not use them, we should just use it less. Or use some framework that will force us to do it more correct, like asyncio.
asyncio allows to call some functions asynchronously by using decorators, can create event loop, can call functions with delay or at some point in time.
No callbacks, easy to use. Usable for SSH, file handling, IRC, HTTPD, RPC, testing, and other use cases.
In the end Ewoud mentioned, that some reimplementations of parts of asyncio also exist, because not everybody liked the way upstream implements the async framework.
More info: https://fosdem.org/2017/schedule/event/python_coroutines/
Ben Nuttall was talking about IoT in Python. GPIO Zero helps implementing Raspberry apps using pre-prepared objects where user only specifies pin number and calls methods to work with the sensor.
Quite complicated and large device hierarchy demonstrated wide range of objects available for users.
Remote GPIO (pigpio) allows to run the same code and in the end control remote device.
Picamera is a able to capture 1080p30, 720p60 VGA90 video, so you can create your mini camera using few lines of code.
Energeie allows to remote control real 230V plugs, an example of remote controlled tortoise lamp showed one of the possible usages. I guess there are more useful use cases though.
Sense HAT is then a complex module that returns many values like temperature, humadity, etc.
More info: https://fosdem.org/2017/schedule/event/python_raspberry_pi/
The last talk on this year's Fosdem was prompt_toolkit by Jonathan Slenders. This tool brings nice and very advanced prompt for cmd-line python apps.
Features included today: autocompletion (using table), syntax highlighting, colors, multi-line edits and many more.
Tools that use prompt_toolkit library already proof its readiness: wharfee (Docker client), pgcli, mycli, ipython, screen splitting, etc.
Since the library includes a lot of around text-editing, pyvim was then quite easy to write, at least to show up that it is not that painful to implement a real editor in pure Python. Jonathan plans to continue in re-factoring the library, so the API can change a bit in the future.
More info: https://fosdem.org/2017/schedule/event/python_prompt_toolkit/