If you have an application running in a Kubernetes pod that uses relatively short-lived HTTPS certificates (like from Let’s Encrypt or Vault), and the application loads the certificates at startup, you may need to reroll (rollout restart) the deployment in order to trigger renewal of the certificates.

Some definitions:

• a is the lifetime of the certificate.
• b is the number of days before certificate expiry to start trying to renew the certificate, the renewal period.
• c is the number of days between each reroll.
• d is the number of times we expect a reroll to happen during a renewal period, a kind of safety factor. (In an ideal world, this would be 1, but sometimes CronJob rerolls aren’t triggered as expected.)

We know that the renewal period can at most be as long as the certificate lifetime, and we know that the reroll must happen at least once during the renewal period. We also know that if you reroll an infinite amount of times, the time period between rerolls becomes infinitely small. Thus, we end up with this formula:

In practice, you would set d to some value you feel comfortable with based on your experience, and then you can vary b according to your needs and come up with a c.

Or you could work the other way around and set c to a value that suits you (if you for example want a daily or weekly reroll for some purpose), in order to discover the ideal b.

This post also shows off a bit of MathML, which has been available in Firefox and Safari since 2011, but for some reason only arrived in Chromium-derived browsers in 2023.

Last year I wrote a op-ed in the digi magazine: AI Act kan gi et rush av halvferdige produkter. It was quite a buzz for me to see a big picture of myself “above the fold” on one of Norway’s biggest sites dedicated to technology news! Later that year it was translated into English for Sopra Steria’s corporate website: AI Act could lead to a rush of half-baked products.

I was inspired to write it after attending some meetings with EU officials who were quite concerned with upholding the dates mentioned. I felt that outside of the EU bubble the distinction between before and after the deadlines was under-communicated, and ripe for a lot of rush jobs once it was more widely understood. So I took it upon myself spread the news, as it were.

I think it did quite well, got me some new connections and allowed me to be invited into some interesting meetings I otherwise wouldn’t have, as a “subject matter expert” on the AI Act. The qualifications for being such an expert consisting of actually having read the whole act and writing notes, which is much the same experience as I had when the GDPR was new and scary. Of course, this is exactly the kind of thought work people are increasingly turning to AI agents to do for them. I definitely have some classical Luddite feelings on the matter.

Or, my rather unordered ramblings on what happened last week.

On the city and social stuff (PDX)

As a European I enjoyed it a lot! Portland has great public transport, is very bike-friendly, close to nature and is about the right size for me. It’s well known for its beer, but maybe less well known is that the Willamette Valley outside Portland is chock full of vineyards and wineries. I went on the FOSS4G wine tour and had a great time!

Another tour I was on is the Keep Portland Wired tour, which also was a blast! It’s a real shame that not more people showed up. (The closing keynote went over time, so I had to bail out to make the tour, I guess a lot of people prioritised the keynote.) We toured the offices of Jive software and Mozilla’s office/working space in Portland. At Jive we were showed around by the company’s founders, two really nice fellahs who were happy to to engage with a bunch of geogeeks. They gave good insights on their experiences starting and running a tech company in the USA, and a killer tour of the building. Their offices were in the old Federal Reserve building, so the facilities included a indoor shooting range and a bank vault with a massive steel door, both of which had in typical Portland fashion been converted into bicycle garages. Otherwise the offices were typical movie-style tech offices with cool gadgets and free snacks and food all over. Local Portland craft beer was available on tap, of course.

Mozilla was also awesome as expected (totally impartial judgement from the guy with a Firefox sticker on his laptop.) They had a slightly different vibe, as one might expect from a non-profit, idealistic type organisation, but were still equipped with a bike garage and Tardis stylings on their refrigerator. Nice conversations were also had in this space, and Firefox swag was secured! (They also had free beer and stuff.)
The FOSS4G parties were good fun too. The opening reception was interesting, a vey nice building filled with interesting folks. I missed the Null Island party because I needed a rest, but in hindsight after talking to other attendees, I kind of wished I went anyway. The Gala evening had a misleading name, insofar as anyone in a tuxedo would have been sticking out like a gangrenous thumb. The location however was nigh on perfect for the gathering, at the World Forestry Centre, with lots of cool interactive museum exhibits for all the geeks to play with and cool local food options.

I got to the Mapbox closing party after the walking tour. Super-hipster post-industrial space with heavy freight trains rolling past, and recent OSM edits being projected on to the wall. Beer, wine, snacks and a great atmosphere. What else could you expect from a Mapbox  party in Portland?

Got to meet some really cool people, like Lyzi Diamond from Code For America, who wasted no time in getting me to promise I’d have a go at gauging the interest for Maptime in Norway. Kristin Bott, who did a great job on the organising committee and whose vision for more diversity in our space I totally share.  Vladimir Agafonkin of Leaflet fame who actually remembered my last name from the internet and said nice things about the Leaflet plugin I made at work, practically making me swoon. Also, like a true Norwegian I spent lots of time talking to other Norwegians, which is very useful. Why not do your local networking on the other side of the globe?

Trends & Impressions (PDX)

• Macs. Everywhere! Come on folks, hooking up a Linux machine to projectors has gotten much better, only like 37% painful. You have to suffer for your art!
• Javascript is surging ahead. FOSS4G isn’t a web tech conference yet, but you could probably pick your talks to make it look like that if you wanted to.
• Lots of buzz about vector tiles and personalisation of maps. Nice to see the stuff that was future-tech back when I wrote my thesis is rapidly becoming reality. (Still bleeding edge though, but three more years from now it’ll be pretty pervasive, I think.) The talk on adaptive maps by John J Czaplewski was a very interesting conversation starter on what we might do with these technologies.
• WebGL is  also maturing nicely, a lot of libraries and apps coming together now. We shall see which ones emerge victorious from the Thunderdome of adoption rates in a few years.
• Speaking of the Thunderdome, in my opinion both the Leaflet + Mapbox GL combo and OpenLayers 3 are in sum promising a lot of the same things when they’re done, but neither are done yet and have different things finished. (OL3 even replicated MapBox Gl’s big party trick by showing off drone/sat video overlaid on a map!) So there is still room for both!
• Like Atlefren mentioned, projections are hard and people would rather go shopping, so use WGS84 for vectors and Web Mercator for maps unless you actually have a compelling scientific reason. (Like if your data *actually* has the kind of accuracy and precision that warrants a local projection, or you’re mapping something north/south of 80 degrees. His observation on more and more non-classically trained folks entering the field also holds true. (AKA. you don’t need 5 years of univerity geomatics to put a pin on a map, and on some levels it probably doesn’t help, either!) But I still feel that scientific rigour has its place, especially for those who are making the tools everyone else wants to use. (But then again, that might just be the five years of university talking.)

Random Observations (PDX)

• Somebody needs to make a LCARS interface for GRASS.
• Compostable utensils and cups that still feel like plactic confuse the hell out of me, I never know which bin to use.
• Dear Portland: “MAX Light Rail” and “Portland Streetcar” are really just two different names for “Trams in Portland”. Why not just call them “the large tram” and “the small tram”.
• Did not see any werewolves or supernatural policemen while visiting, did see at least two locations I recognised from episodes of Grimm.

FOSS4G NOR

Two days after I got back from Portland I attended the local Norwegian FOSS4G event, where 86 happy hackers and useful users of open source geospatial software in Scandinavia’s most coast-liney country (award-winning fjords!) got together for a day of presentations and chatting. Together with Alexanno I did a ten minute lightening talk on impressions from FOSS4G PDX, in which our rate of speech steadily increased throughout as we tried to cram in more and more information in the allotted time.

The other talks were also very interesting, like Bjørn Sandvik’s presentation on the impressive open source-based stack the journalists at NRK use to publish maps with their news stories, and the talk on how the Norwegian Defence Research Institute makes maritime monitoring systems  for the coast guard using open standards and open source. We were told that the Norwegian Armed Forces mandate OGC compliance in their internal secured systems, which is pretty cool. OGC standards, open source software, and things that are expensive and make other things explode. Can it get any better?

I believe all the slides from the talks should soon make it out to the internet, sadly not full recordings like the big FOSS4G does. Next year we’re hoping to attract a bigger crowd of students and others to FOSS4G-NOR, making it less dependent on the big companies. All in all I’ve had two great FOSS4G’s in two weeks, and I hope to attend many more in the future!

I’ve been neglecting to blog the last three days because I’ve been going full steam ahead since the last blog post! After I finished the GTK#3 template for MonoDevelop, I moved on to use it to create a new Mono.Addins subproject named GuiGtk3. This subproject was needed so that apps depending on Mono.Addins.Gui (Pinta, Tomboy, F-Spot)  could move forward, while still allowing GTK2 projects to use the original Gui. The API is identical, so all you need to do is drop it in place of ye olde one and change the using statement. Right now the pull request is pending.

With that successfully concluded, I took over where Stefan Hammer left off on Tomboy, and using all the knowledge I learned from Pinta, Mono.Addins, Mirco Bauer and Bertrand Lorentz I powered through Tomboy’s porting work. (Hopefully there will be a technical post/memo for future use forthcoming…) The net result is Tomboy, in GTK3, using GSettings, and with every last scrap of horrible icky C code evicted!

Right now the code is living in this branch, but we can expect a release in time for the next version of Ubuntu, when all the needed library dependencies have made it in. (If you want to run it, you need to pull down the Mono.Addins.GuiGtk3 branch mentioned previously and make install it to have the GTK3 Gui available.)  There is one thorny issue remaining, which is the state of GTK3 on Windows and Mac. Someone will have to make installers for this on these platforms before we can make GTK#3 releases there, but I think that the current attitude is that Linux is the lead platform and the others will just have to wait.

The port is complete, but we don’t yet take advantage of all the features of Gnome Shell and/or Unity, so among the things we have noted for future improvement are improved DBus connections to allow direct interaction with the shell and making the panel applet an add-in which can be deactivated where appropriate.

The hackfest, summarised

I didn’t get to finish porting Pinta, but by tackling it head-on I learned some important things that were very useful when I did the projects I’ve completed this week. (Talk about diving in on the deep end!) I managed to make important contributions to tooling and dependencies for future GTK#3 work, and I managed a complete port of my first open source love, Tomboy Notes. We have had a great time together and been super productive, and there is no doubt in my mind that the benefits of inspiration and  direct knowledge transfer that come from being in the same room cannot be matched in any other way.

Finally, let me once again thank our gracious sponsors:

Norkart AS, Norway’s premier supplier of Geographic Information Systems and related consulting and my most lovely employer (website, logo)

Collabora Ltd, Open Source Consulting (website, logo)

Hotel Schottenpoint, Our hotel partner (website, logo)

Novacoast IT, Professional Services and Product Development (website, logo)

The GNOME Foundation, providers of the GNOME desktop (website, logo)

Venue sponsor:

• The University of Vienna (website, logo)
• Institute for Theoretical Chemistry (website)