A couple of years ago I have started to experiment with the Raspi. Well, what can you expect from a computer that costs less than 50€? I thought. But as time went by, I started to use them more and more. Now I have had 4 of them (V3) in production for quite some time, mostly to control radiators and other smart home components through openHAB, an open source control center software for smart homes.
Today I was checking on one of them and while logging in, it’s welcome message filled me with joy:
Last login: Thu Oct 22 07:38:33 2020 from 18.104.22.168 Mi 9. Dez 13:02:39 CET 2020 @ xxxxxxx Uptime: 13:02:39 up 634 days, 23:17, 1 user, load average: 0,00, 0,00, 0,00
For almost 2 years the little guy was up and running, doing his work. Enduring internet outages and what not. I had not expected such a great performance. I hardly dare to upgrade him to the new OS version. Not for that I don’t trust the update process (I have upgraded most of them already to Buster), but because I want him to continue present me so pleasant stats!
PS: Kudos also to the folks from openHAB. While I find some functions quite cumbersome (not a JAVA friend here) I’m really happy with the RESTFul API that they provide. It has allowed me to develop some very helpful functions for automation on top of openHAB.
When we started to get serious about our holiday apartment business in 2008, I also got serious about providing IT support for the business.
At the time, it was very important to have your own website. I think about 30-40% of the bookings came from that source. With the little experience I had back then I chose Drupal to build the site. Note: This was before folks were using (i)Pads and smart phones to check for vacation homes. So we ended up with a page that until today sucks on small screens.
Next project was to create an internal system to keep track of our bookings. So I set off to write a small PHP system to create and track visitors, bookings, invoices and so on. What I had envisioned to be a very small app turned into a much bigger endeavour once I noticed that it won’t be just me using this app. So I wanted to add a nice UI and chose a very cool and hyped framework called Angular.js. The initial system booking system went life on my provider’s servers in 2014, the Angular.js version about a year later. Angular.js is now in long-term support. And starting with the next version (2.0) Angular.js was renamed in Angular and it underwent a major redesign and paradigm shift. Needless to say that it more or less requires you to learn a new language called Typescript.
After all that I started to enjoy hacking away again. Looking at the cost structure of our business one of the interesting places where we could lower cost and do some good for the globe appeared to be lowering our heating (aka oil) cost. Around 2016 I had started playing around with an ope-source project called openhab, connecting to closed solutions to control heating radiators. I implemented a system in openhab (or actually using openhab’s REST interface to connect the booking system and the heating control system. That was and is fun! And recently I started using Grafana to visualise the performance of the system. Just to find out that the closed solution (I’m using the MAX! system) is not up to par. So I’m learning a lot about the ‘hidden features’ of MAX! and try to adopt my control center software, running on Raspberry Pi’s at home and in the holiday apartments..
For the past years I was still fully employed. In June 2020 I changed from fully employed to working a maximum of 2 days a week on consulting contracts. This allowed me to sit back and look at the technical debt that I have accrued.
What I learned is:
Don’t create a website if you don’t have to
If you really need, think twice if you need to use a CRM of if a static page suffices
For each and every piece of software that you develop: Never forget the cost of ownership.
Write tests and documentation. You will not understand what you did 6 years ago without it!
Speed of front-end technology changes far faster than backend: Separate the two systems as much as possible (e.g., RESTful interface)
It sucks to use IoT systems (e.g., smart home hard/software) that is a) closed and b) developed with a focus on the hardware). In addition, the cost of the hardware is still very high and the ROI is not coming any time soon. Rather invest into more bold solutions (e.g., heat pump based solutions, solar energy)
It was and is a lot of fun to play!
So I’m going to invest some time into
Choosing a different technology for re-writing our public website (static page, goHugo?)
Learning Typescript (actually started a course @ Udemy)
Chose the right front-end technology for the internal app (e.g., React, Angular)
Make plans make major investments into saving a) the planet and b) heating cost
Ich kann wirklich nicht erklären, warum es so lange gedauert hat. Aber endlich habe ich mein Gersfeld Mesh-Up Netzwerk am Laufen. Was war das Problem: Wir haben eine Location, bei der vom Provider einfach nur furchtbares Internet geliefert werden kann (sagt der Provider). Die Idee war für eine ganze Weile, dass wir die Bandbreite der anderen Locations (andere Verträge, andere Anschlussmöglichkeiten) koppeln und dann auch bei der unterversorgten Location gute Bandbreite zur Verfügung stellen.
Die Hardware hatte ich schon seit Wochen (Monaten) herumliegen. Heute habe ich sie endlich zusammen gebastelt. Und siehe da: Es ward Licht.
Leute, die Zeit von Freifunk ist noch nicht vorbei. Mesh Networks sind eine Riesen Chance für kleine Orte wie unsere. Jetzt müssen wir nur noch schauen, wie wir alle Punkt mit Linien verbinden!
Just came across an article in a German internet news portal. The headline claims that Uber won’t ignore official rulings anymore. That is an interesting statement. It makes it clear that Uber thought in the past it is above all and everybody. And it leads to another thought: At least here in Germany they had to stop the bullying and behave. I really appreciate that the German cities have started to take a stand and show the Überflieger that the same rules apply to everybody. The Germany manager of Uber, Christian Freese sounds now like the wolf who ate chalk. He even claims that Uber is trying to cooperate with the traditional taxi firms.
UberBlack has been prohibited in Berlin, and UberPop has been prohibited Germany-wide.
And while we talk about it: Berlin has taken a stance and is reclaiming rental space from AirBnB and similar companies. But more on the friends-rent-to-friends companies in a different post.
In the last couple of month I’ve started to experiment with data. Won’t call it big data, because my datasets are rather small. Like visitors in our holiday apartments or feedback from EclipseCon Europe attendees. But like a real guy I didn’t want to do with MySQL, I wanted to play with the the new toys. So I installed MongoDB.
Anyway, it turned out to be more difficult than I thought. Cleaning up the data and getting them in (the same) shape turned out to be harder than I though. But help was just around the corner – coming from Udacity: A course on Data Wrangling, just what I needed. Turns out it was just the right course at the right time. And forgive me, if I don’t bore you with all the details that I had to clean up.
I’m certainly not a big data expert yet. But I start to understand how to approach my problems and what I can do with my information. And visualizing that information is even more fun. Now you can guess what that heatmap shows 🙂
I’ve been one of these folks who bought a Ninja Block pretty early after they appeared. This was in the time before I knew much about Arduinos and Raspberry Pi’s. My block turned out to be fun to play with: I could read humidity and temperature and stream my webcam. Through its nice little REST interface I had a chance to link all this into my small web universe.
Anyway, I liked it so much that I wanted to buy another one for another location, again with all the sensors you need. Turned out that everything is sold out, and the Ninja folks say that they are planning the next big thing – the Ninja Sphere. And even small things like a temp sensors continue to be sold out 🙁
I don’t know when the announcement for the Sphere started, but by now it feels like a year or two. Nothing has happened since then, at least nothing that I can see. No release dates, just a link to a weird release plan in one of their blog posts.
After my last post I was actually contacted by 2 people asking for more current information on the website that I had built. In particular, they were interested in the conditions of the winter sports facilities that we have in the region (ski-lifts, cross-country trails).
I looked around, and the only information available was on the web sites of the facility operators. No central place where all the data was collected and made available. Since I had never done screen scraping before, I wasn’t really sure what to do.
Reading up on Stackoverflow and other resources I learned that I had to read an HTML site, turn it into a DOM object and find the right places with the right information for the facilities (closed, open, good conditions, red.gif, green.gif). Looking around I found a nice helper library that served me very well with my first version: For every webpage to get data from I wrote a little PHP script to capture the data.
This worked well for the first facility, where the website was quite responsive. The second one was making more trouble with regard to response times. Now I had a 6 sec wait before my page was displaying. That wasn’t really acceptable, because I have still 2 more places to scrape.
So I took the Saturday afternoon to make it work asynchronously. It turned out to be quite easy: I continued to use my PHP scripts, but converted them into functions that could be called with AJAX calls, returning JSON data. From there it took only a couple more minutes, and I was finished. Displaying the site itself is really fast again, and since the scraped information doesn’t show up in the visible part of the browser things can take a little longer. But even scrolling down right away is fun: I enjoy watching the data show up!
They design, program and manufacture what they call intelligent autonomous vehicles, the next generation copters. The idea is that the hexa-copters can be used for unmanned surveillance missions that needed way more expensive methods in the past: One example that Joerg Lamprecht, the CEO of the company describes involves these vehicles inspecting building status or taking high-precision arial photos.
As with all the other manned or unmanned vehicles I’ve been dealing with lately, these machines rely heavily on the software that comes with them. While there’s a whole bunch of other issues like robustness, weight and construction, the big challenge is to provide software that doesn’t need an experienced pilot to fly around a building and take head-photos.
Logically, the company has a couple of cool software developers to provide a platform system for the copters that integrates all access to navigation, flight control, camera adjustment and such. On the other hand they are very interested in developing apps that can be used on top of the platform. These apps could be maneuvers such as loopings or turns, or it could be even more complicated tasks like ‘fly around this building in a spiral and measure it’.
Time was flying by, and just before we had to leave Joerg took us to their manufacturing hall to see a bunch of young kids building the copter platforms. After the visit I was wondering what the banks and insurance companies have to do in the future to keep their developers from running away and doing cool stuff like Aibotix is doing.
Both Christian and I look forward to the arrival of the IoT!