The immediate future is easy to predict, because the same things that has been happening for years will continue to happen: Some of the exciting recent advances will be cancelled out by new compiler regressions due to fancy new compiler features. We users will create still deeper code stacks that racks up latency, even more so with Julia 1.9 now that latency has improved and we can “afford” to do so. — Julia’s latency: Past, present and future

Trying out J!

» jconsole
   comb=: [:;(,.&.><@:>:@;\.)^:(i.@>:@-~`[`(1:<@i.@{.~<:@-,2:))
   2 comb 3
0 1
0 2
1 2

Reading answer like this makes me think I should have spent more times with J when I first discovered it years ago.

♪ Avenged Sevenfold · So Far Away

The Many Faces of an Undying Programming Language. An entertaining discussion of Lisp, Scheme(s), and other dialects. #lisp #scheme

Deforming programs so they are ‘iterative’ in order that they do not run out of the stack we imagine to be so costly is ridiculous: if you have a program which is inherently recursive, let it be recursive. — The absurdity of stacks

The LAPACK (Linear Algebra PACKage) online document browser is really nice. TIL there’s even a routine to perform QR factorization without pivoting to a real or complex matrix (related R package: QR).

littler 0.3.18 on CRAN. It’s been a faithful and reliable alternative to Rscript for a long time now (17 years and counting). #rstats

Multiple Linear Regression in SQL with Only SUM() and AVG(). For those interested in multiple regression and residualized regressors as well! #statistics

If you type something that looks like a filename, you probably want to prioritize filename matches. When your search query looks like a filepath, you probably want to use an algorithm designed for filepath matching. This makes zf a “do what I mean” program. — A different approach to fuzzy finding

♪ Plain White T’s · Hey There Delilah