- A mix of parody and seriousness, on Haskell and Ocaml: take 1 & take 2
- What’s new in Scala 3
- A look at Raku (and some myths)
- On Paradise Lost
- On math vs simulation
- I might have posted this already, but: a shot of Emacs in Tron Legacy …
- On web pages that are … built to last
- Crux, an open-source bitemporal graph database backed by Kafka (and featuring a testimonial for Confluent Cloud)
- The title says it all: Running a bakery on Emacs and Postgresql (!!)
- If anyone remembers Terry Davis, a BBC episode of The Digital Human about him, his life, his work (it’s one of those truly bizarre achievements)
- Cheering Drew DeVault for an insipiring side-project taken to Alpha and beyond
- On really teaching computational thinking
- Carl Hewitt (of Actors) on Reusable Scalable Intelligent Systems
- Extremely comprehensive set of musings on text editors
- Looking back at twenty years of Erlang
- Bit of absolutely hilarious satire (or at least that’s what I thought it was)
Note: Many parts of this post are true or at least plausible, but I did not actually install Emacs in my car. Neither should you.
- Ways in different programming languages, of achieving fearless concurrency, another comparison of “actor frameworks” across languages
- A video about live-coding … from a decade ago
- An interview of Arthur Whitney (of K and Kdb) by Bryan Cantrill
- Speaking of which … he’s part of an interesting new company now.
- An interesting comparison of Swift and Rust
- On switching to Plan 9
- A “real-life example” of using Datomic
- An adventure with the JVM, from one of the folks behind JRuby
- Another deep-dive into an example of JVM bytecode
- A “FUD FAQ” for Scala
- On the weird orbital movements of Neptune’s moons
- File this under “weird things parasites make their hosts do”
- An interesting interview on “the aesthetics of programming tools”
- Couple of examples of applying modern perf tools in Linux
- Command-line tools in Clojure!! Which work fast thanks to GraalVM!!
- Updates on Project Valhalla for the JVM (Part 1, Part 2)
- Interesting take on database indices vs hash tables
- Putting too cool things together: Smalltalk and Graal_ (one very old, and the other very new)_
- A look at new Garbage Collectors for the JVM (and a brief look at the equivalent in Go)
- Applying the lessons of the Golang scheduler to the Tokio scheduler in Rust
- Speeding up Nixpkgs by avoiding subshells
- Pessimism about engineering software
- The Racket programming language is heading in a new direction
- Ben Lynn’s “most functional compiler”
- Someone tries TLA+ and comes away impressed
- Something recommended by a co-worker: CMU 15-721, a course on Advanced Database Systems
- Matt Godbolt (of Compiler Explorer) on “why C++ isn’t dead”)
- Urbit is here
- Comparing APL and C (!)
- Remembering the era of Flash
- Clojure at “the immutable bank”
- Zen and the art of software maintenance
- Shareware and floppy disks …
- Reminiscing about different versions of Microsoft
- I recently discovered all the different “server-less options in GCP”
- Looking back at the first version of Redis, written in … wait for it … in Tcl !!
- A periodic reminder about microkernels
- Why text editing is so hard to implement
- (Re-)Assembling the Apollo Guidance Computer (!)
- Wondering whether Perk 6 should be named as a totally different language
- A look at “typed Lisp”
- Bob Martin on “why Clojure”
- Rakuten And Egison
- In defense of stable platforms
- And now, a quote, something about languages both natural and artificial, spoken and digital:
It makes perfect sense to say that the target demographic of English is English speakers—or more precisely, the target demographic of Modern English is Middle English speakers who wanted a few simpler rules, some continental vocabulary, following other contemporaneous European languages in not having þ and distinguishing i and j, etc. It was a relatively small change and very intentionally served a community of people who already spoke Middle English well. It was the Python 2-to-3 of English.
The target demographic of Esperanto, meanwhile, was the whole world: people who already had a language, people who already had a workable lingua franca in international contexts (French, later English), and in particular people who weren’t familiar with the European language patterns that Esperanto was largely based on. So it had limited success. A Japanese or Indonesian or Persian or Swahili diplomat not familiar with any European language would be better served learning French or English than Esperanto, because those languages are roughly equally foreign, there are many more resources for learning French or English, there is a larger community and more people to speak to, there is a larger corpus of works, etc. And an Anglophone or Francophone diplomat has very little incentive to learn Esperanto, either.
Modern Hebrew, on the other hand, had a well-identified target demographic: Jews from around the world migrating to the reestablished state of Israel who lacked a shared everyday language. Some liturgical Hebrew was already familiar to most of this population, and there was a strong cultural willingness to see a reestablished Hebrew language. So while it is in many senses a conlang, it was far more successful than Esperanto and now has a large community of native speakers.
So the question of whether a programming language—especially one that is so much like a conlang, not an incremental evolution based on use in practice (like the C standards committee accepting compiler-specific dialectal changes)—has a target demographic is a fair one.
(yep, not a lot this month …)
The shrew-like networked GUI equipped microcomputers of Apple were released as products only two years after this central planning dinosaur was postulated. Eventually, decades later, someone built a mechanical golem made of microcomputers which achieves a lot of the goals of fifth generation computing, with independent GUI front ends. I’m sure the Japanese researchers of the time would have been shocked to know it came from ordinary commodity microcomputers running C and using sorts and hash tables rather than non-Von-Neumann Prolog supercomputers. That’s how most progress in engineering happens though: incrementally. Leave the moon shots to actual scientists (as opposed to “computer scientists”) who know what they’re talking about.
As I’m absorbing more of the general flavor of event-streaming and everything around it (… in the beginning was the log …), I keep gettin a bit of deja vu and realized that there’s an uncanny resemblance to something else, a different place I’ve encountered similar relationships between things in a different concrete form … in Clojure.
I feel Clojure keeps getting compared to other “languages for talking to machines”, and discussions tend to end up rehashing either the static-vs-dynamic-typing cliche, the lisp cliche(s), the scripting-Java cliche, and so on.
What is missed (I feel) is the symbolic nature of the language, which allows it to really represent concepts at an arbitrary level of abstraction … something that will be incomprehensible if all you ever compare it to is examples of computing factorials in other languages.
In particular, it is a sort of “language of the system”. The same patterns I see Apache Kafka getting involved in, for distributed data-processing systems, are the patterns I imagine with symbolic objects in Clojure code.
So instead of a log with brokers and producers and consumers and streams, I see …
LazySeq, with transducers; instead of Schema Registry,
spec … and so on.
This is not a fully-formed thought yet, but … dunno, I’m enjoying this parallel so far.