Announcement: here is something I worked on for a while - a language extending Datalog. It also does something that resembles type-checking. https://github.com/google/mangle
Transcriptions of lectures by Per Martin-Löf, available freely
There’s a soundness hole in Scala 2 involving wildcards that Stephen Compall reported back in 2015 — Dale and I found today that Java has the same hole! We’re trying to close the hole for Scala 3.3.0 without making too much trouble for library authors; details at https://github.com/lampepfl/dotty/issues/16789
If you have not yet, read this excellent page with commentary on Wiio's laws (Communication usually fails, except by accident...) https://jkorpela.fi/wiio.html
Published in 2016, this brilliant comic by artist Joe Dater somehow feels even more timely in 2023. #art
The Rust book but with interactive quizzes and some editing by PL experts and passionate educators Will Crichton and Shriram Krishnamurti https://rust-book.cs.brown.edu/
I just cut my #rust build time from 23 seconds to 2 seconds using mold. What sort of alien magic is this...
(you can do this too, see https://github.com/rui314/mold#how-to-use )
Inconsistency and redundancy in requirements is a guarantee for a failing project. Slowly but for sure it will kill any attempt to bring a project to success.
One of my professors during PhD used to say “You can drive a truck through the holes in any given paper. So you look for what you *can* learn instead.” And being the smartass grad students we used to think driving that truck was fun. After so many years, I now appreciate her wisdom more than ever. All scholarly work has limitations but it’s refreshing when people critically evaluate what’s the actual value of the research. It's about humility, honesty, rigorous intellectual work.
Found the time to add checking of struct (record) types to #mangle #datalog ... with structural subtyping and optional fields (a bit protobuf like, not option type). It certainly helps when there are almost no operations and lots of code to use for testing. I can't possibly write docs when the type system isn't done yet, can I?
Similarity as fuzzy equivalence relation
My superpower is visiting the library. I was reading up on ML and about every text treats similarity as the dual of distance.
Then I get Richter, Weber textbook on case-based reasoning (no ML, aimed at interdisciplinary audience) and it is the first time I see similarity can be seen as *fuzzy equivalence relation.* A pair x, y of elements is mapped into [0, 1], not triangle inequality, but a suitable notion of transitivity. Conjunction needs to be fuzzy (using a t-norm). And Gödel (G_k with k truth values, intuitionistic) and Lukasiewicz were there long before Zadeh talked about fuzzy sets in 1971.
It just helps to stick to logic as an organizing principle. But i need to study this more and work out details.
I had come across the notion of fuzzy sets while reading up on category theory and toposes, and I had heard about fuzzy logic (all the rage in the 80s?) but then I never realized to what extent fuzzy logic can be connected to ML. It seems once more that results of research from a past decade are hard to appreciate because of fashion/change of perspectives which prompts people to use different words and applications used for arguing validation or relevance. Fortunately the tenets of logic do not change. #research #ml
15y at Google, before that Scala (pattern matching FTW). Interested in many things PL from theory to implementation, also logic, category theory, also distributed systems, more recently infosec.
Follow the @datalog group!
#datalog #CategoryTheory #logic #types #systems #QueryLanguage #DistributedSystems