Difftastic works really hard to find similarities between the before and after file. Sometimes this works wonderfully (see first screenshot) and sometimes it goes way too far (see second screenshot).
It's matched up a few variables and semicolons from adjacent functions!
I've released difftastic 0.35! In this release:
* Basic support for Makefile syntax
* Support for UTF-16 files (helpful on Windows)
* Several quality of life improvements, including the ability to see what languages are currently supported
Today I learnt that you can simulate a command being run in a TTY with `script`, e.g.
$ script --return --quiet --command "ls"
Unfortunately it generates output files called "typescript", which is a little confusing!
I had the pleasure of attending the Making Smalltalk event at the
@ComputerHistory today!
Adele Goldberg and Dan Ingalls were there, plus some videos from Alan Kay. "One of the tests of a programming language: can you write a debugger in it?"
@janriemer@mastodon.technology That's a valid use of dbg! as far as I can see, what's stopping you doing that?
One interesting feature of Rust: it has both print and dbg, so you don't need to use print for debugging very often. I don't even use print that often.
Given the number of game companies that have re-released old games by bundling an OSS emulator, perhaps it's only a matter of time before Microsoft includes Wine in something!
One nice thing about quadratic algorithms: if you can reduce your N, you get a really nice speedup.
I think I've used fewer third party libraries in bash scripts than in any other language. I just use the built-ins and invoke binaries.
I can't think of any other programming language where the stdlib felt like it was sufficient.
YAML is a funny language: it's not Turing complete, but it can have arbitrary code execution bugs.
On the limits and perils of being data driven:
https://twitchard.github.io/posts/2022-08-26-metrics-schmetrics.html
(Worthwhile improvements are often not amenable to A/B testing, and metrics can harm intrinsic motivation.)
I've just released difftastic 0.34:
https://github.com/Wilfred/difftastic/releases/tag/0.34.0
Minor release, but it fixes some build issues introduced in 0.33, so worth upgrading if you've had any issues with 0.33 :)
It's perhaps appropriate that a programming language database has its own syntax for entries!
https://github.com/breck7/pldb/commit/bde872b5b8e1cf9fb3114c1fbf3c40e405b5388d
It looks like a simplified YAML I think.
PLDB is an ambitious project to catalogue all the different programming languages out there:
https://pldb.com/lists/languages.html
It's showing a ton of potential (over 4,000 items recorded!) although it's still pretty young (I just fixed the LLVM IR comment syntax).
The bottom type has a different name in a bunch of languages. I've seen nothing, _|_, !, never and empty.
I don't know which is the most approachable to users though. `List[Empty]` reads very nicely for an empty list, whereas `never` reads nicely for error functions.
I've heard of quines, but today I learned about radiation-hardened quines!
https://en.wikipedia.org/wiki/Quine_(computing)#Radiation-hardened
These are quines where *any* single character can be removed and they still function. Fun concept, seems easier in a language with eval().
TIL about the box.leak() method in Rust:
https://doc.rust-lang.org/std/boxed/struct.Box.html#method.leak
This leaks the box, but it also gives you a reference that can live longer! E.g. you can leak a Box<u64> and treat it as 'static.
It makes sense, but being able to use a value after leaking is new idea to me.
@takloufer@fosstodon.org
@emacsomancer Fascinating, an authoritative slide! Thanks for sharing :)
