Tracing has comparable performance to partial evaluation, but with significant less implementation effort: http://t.co/BKkZhzFAkv
miniblog.
Tonic is a great REPL environment for node: https://tonicdev.com/ . I think web-based REPLs are still a massively underexplored area.
Neither mobile Firefox nor Chrome (as far as I can tell) have a View Source feature. I think this is a loss though I'd never noticed before.
Good software testing takes a 'guilty until proven innocent' approach.
I've written my first company backend! The API is a little unusual (you have a function where the arg varies) but it's nice to work with.
Joy (the programming language) can be very pretty. Qsort:
DEFINE qsort ==
[small]
[]
[uncons [>] split]
[enconcat]
binrec.
HolonU is an interesting, quirky IDE that stores source code in a database with a hypertext interface: http://t.co/31oaPd14WP
Updating all *ELPA packages gave me: 3495 files changed, 436904 insertions(+), 423268 deletions(-). That's a lot of elisp -- hard to test!
The lisp macro -> is more powerful than function composition. For example, this is legitimate elisp: (-> 100 (goto-char) (save-excursion))
Emacs' ielm is invaluable for playing with elisp snippets. Tip: use * to access the result of the previous expression.
Handy Emacs command of the day: 'M-x foo' -> jump to foo.el. It's replaced much of my use of 'C-h f'.
I'm also extremely impressed with the release tooling in Rust. Bors and Crater are both fantastic for keeping the language reliable.
I'm acquiring a taste for the Rust numeric types. usize usually represents an index, i64 a quantity, and Wrapping<u8> an emulation of hw.
I really like diff-hl-mode https://github.com/dgutov/diff-hl It shows lines added/modified/removed since last commit, which is useful contextual info.
Using LLVM with very dynamic languages is hard. A discussion of using LLVM for Smalltalk VMs:
Showing 5,131-5,145 of 7,583 posts