Clojurians Log v2
announcements
solid-cljs v0.1.1 ClojureScript bindings to Solid: com.github.roman01la/solid-cljs {:mvn/version "0.1.1"}
by somewhat popular demand, I unarchived the repo, updated the library and pushed a release to Clojars
https://github.com/roman01la/solid-cljs
p.s. Claude Opus is really really good at writing macro heavy Clojure code
Clojars Download Stats, https://github.com/PEZ/clojars-download-stats An always up-to-date, complete SQL export of artifacts daily downloads since November 2012 (start of Clojars). Includes Babashka tasks for creating your own sqlite database from the exports and for keeping your local database up-to-date. The repo exists because I started to play a bit with Clojars daily downloads, and needed a database that would let me accumulate things over whatever time periods. Attaching a screenshot of a Joyride flare tracking shadow-cljs downloads since its release. (The dip Jan 2021 is due to a Clojars outage in download stats.) Ask me to plot something, and I’ll post more screenshots in the thread. 😃
looks interesting, but the claim that 2012 is the start of Clojars is strange. Clojars started in 2009.
Ah, I’ll remove that claim. 😃 Start of the stats at least.
thanks. out of curiosity do you know whether the user-agent data in the logs is public?
I mean, it's not included in this repository, but is it in the upstream data that this repo pulls from?
I was interested in Clojure and ClojureScript downloads, but those are fetched from Maven Central directly, Copilot tells me.
The user-agent isn't in the download data exposed by clojars.
@technomancy the
ring.core downloads over time. That massive spike in May 2020 is there for Reagent too, only more pronounced. But not for shadow-cljs.
My current theory about the May 2025 spike is that it was Covid work-from-home start. May 2025 is only the peak, it starts building in March, which according to Wikipedia is when the massive WFH shift started.
just a data point, Nubank eng grew from 160 eng (2018) to 4000+ (2025), convince me this is not a readout of Nubank's CICD pipeline
Their 10,000 microservices is probably more than the rest of the clojure usage combined
It’s not easy to tell if CICD is the main signal. If that early/mid May spike is a work-from-home effect, it indicates that individual developer machines is a big part too? Could of course still be a lot about Nubank anyway. Though, a while ago I had country stats in Calva downloads, and then the majority was from the US. I can still see very big impact from e.g. Thanks Giving.
tracking countries by proxy of their holiday schedules would probably be pretty effective. iirc that's how they found that Jia Tan wasn't actually Chinese.
Looks like so. 😀
Actually like so…
@dustingetz to your question. It does seem like the Clojars download data corroborates that Brazil is major Clojure hub. Using a basket of popular dependencies, and averaging using the weeks around national holidays, and using some 3-7 years for each holidays, I find that Brazil Carnaval Monday impacts downloads about as much as US thanksgiving: Country Holiday Average Effect Data Points 🇺🇸 USA Thanksgiving -19% 7/7 years 🇺🇸 USA Black Friday -21% 7/7 years 🇧🇷 Brazil Carnaval Monday -20% 3/7 years (clean) 🇩🇪 Germany Unity Day -3% 5/7 years (weekdays) I don’t see a trend that the effect is weakening in the US. For Brazil the number of reliable years is a bit too small to speculate about trends, but there also is none there to speculate about.
@pez How do you know which day of Carnival it is? I, as a Brazilian, know that is something like "its 40 days after the first full moon on some month etc etc" But I have no idea how to code a function that knows what day is it
@dustingetz this is also AppsFlyer CICD
@souenzzo i used hard coded dates from what I could find on the web.