On a personal level: yeah, taking a 3-month sabbatical was a good call. I'm *way* more productive now than I was six months ago.
Currently speed-running ten years of ecosystem evolution for #Querki. (Which had been stuck on #Scala 2.11 / #Play 2.5 for a decade because of Dependency Dumb.) Often frustrating, but so far I keep managing to get past my roadblocks fairly quickly, and it's mostly fairly meditative stuff...
Anyway, step one complete: #Querki is finally running in #Docker locally.
One mistake was that I committed to an early deployment solution a dozen years ago which, sadly, was a dead-end. That was kind of invasive, and wound up pinning Querki to *ancient* versions of #Scala and #Akka (and libraries).
So now, we're switching everything to Docker, and probably just going to use native AWS for deployment for the time being. I may eventually switch to #k8s, but one problem at a time…
This is, BTW, not a call for everyone to pile into #Querki right now. As part of the upcoming work, I'm going to be downscaling it to a single node for a while, both to simplify things and make it a little cheaper to run.
Folks are welcome to join if it looks immediately useful to you, but I’ll give a yell when I think it's in a better place and I'm actively looking for more users, hopefully by March.
(thread continues…)
But #Querki is looking weirdly timely now, in a counter-culture way. It's anti-advertising, own-your-own-data, and while it's not *anti*-AI per se, it doesn't have any and might well never do so. (If we ever do add any AI, it’ll be strictly opt-in, not this "we use all your data for training without your consent” BS.)
It's intentionally anti-enshittification - I considered VC money early on, and ran screaming once I understood the implications.
thread/
It's never made a cent (I'm too perfectionist for my own good, so it's still beta), and is showing its age -- it's been running for 12 years now, and the UI just feels a little clunky and old-fashioned. It's still *useful*, but could be so much better.
I've largely neglected it in recent years, since I've had to focus on dayjob, but I'm planning to spend the next few months primarily working on it.
2/
Right - one nice thing about taking sabbatical is I can talk about what I'm doing, without NDAs in the way. So there will be occasional #Querki musings here, tagged as such.
Context, for those who don't know me well: https://querki.net/ is my own eensy garage startup, that I've been running for many years. It's essentially a hybrid wiki/database, aimed at individuals and small communities.
More info on the high concept at https://querki.net/help/#!What-is-Querki
1/
Article showing off a cool new library that illustrates #Actors and #FP working cleanly together using the #Typelevel type classes.
I may well wind up evolving #Querki towards this in the medium term: that's fundamentally Akka-centric now, and I've long wanted to move it towards Cats-effect 3.
https://mastodon.social/@cloudmark/112615743584339619
@mpilquist @alexelcu Ooooh, and that led me to Alex’ year-old article on integrating #Akka with CE3: https://alexn.org/blog/2023/04/17/integrating-akka-with-cats-effect-3/
That's directly relevant to my life, since I hope someday to bring #Querki into the modern age, using CE3 and fs2, and it's 100% based on Akka Cluster Sharding. (Still the number-one use case for Akka.) So that's going to get bookmarked for future reference; thanks!
Justin du Coeur