@DavidSherret

Software developer. Prev @deno_land. Working on ts-morph, dprint, and more.

 Toronto, Canada
github.com/dsherret
Joined January 2009
  • Tweets1.1K
  • Following446
  • Followers1.3K
164Media
Crunched the latest CVE/Pope numbers.
Bar chart titled "CVEs per Day by Pope" showing average published CVEs per day of each pope's reign, with a photo of each pope under his bar. Pius XII, John XXIII and Paul VI: 0. John Paul II: 4.5, counted from the CVE program's 1999 launch. Benedict XVI: 15.8. Francis: 50.0. Leo XIV: 219.2. Data from the official CVE list as of October 23, 2026.

ALT Bar chart titled "CVEs per Day by Pope" showing average published CVEs per day of each pope's reign, with a photo of each pope under his bar. Pius XII, John XXIII and Paul VI: 0. John Paul II: 4.5, counted from the CVE program's 1999 launch. Benedict XVI: 15.8. Francis: 50.0. Leo XIV: 219.2. Data from the official CVE list as of October 23, 2026.

Bar chart titled "CVEs by Pope" showing published security vulnerabilities (CVEs) during each pope's reign, with a photo of each pope under his bar. Pius XII, John XXIII and Paul VI: 0. John Paul II (1978–2005): 9,099. Benedict XVI (2005–2013): 45,340. Francis (2013–2025): 220,948. Leo XIV (2025–present): 116,850 so far. Data from the official CVE list as of October 23, 2026.

ALT Bar chart titled "CVEs by Pope" showing published security vulnerabilities (CVEs) during each pope's reign, with a photo of each pope under his bar. Pius XII, John XXIII and Paul VI: 0. John Paul II (1978–2005): 9,099. Benedict XVI (2005–2013): 45,340. Francis (2013–2025): 220,948. Leo XIV (2025–present): 116,850 so far. Data from the official CVE list as of October 23, 2026.

calle
@callebtcOct 22
I'm tired boss
7
267
More progress compiling Prettier to Wasm then running it in dprint. 32 threads then 1 thread.
32 threads: 11.4s compiled vs 182s in Node.js

ALT 32 threads: 11.4s compiled vs 182s in Node.js

One thread: 135.6s compiled vs 364.8s in Node.js

ALT One thread: 135.6s compiled vs 364.8s in Node.js

David Sherret@DavidSherretOct 17
Current progress compiling Prettier to Wasm.
Plugin in dprint, 32 threads: 22s Prettier 3.9.10 experimental CLI, default workers: 182 s Plugin in dprint, 1 thread: 260s Prettier 3.9.10 classic CLI, 1 thread: 365s

ALT Plugin in dprint, 32 threads: 22s Prettier 3.9.10 experimental CLI, default workers: 182 s Plugin in dprint, 1 thread: 260s Prettier 3.9.10 classic CLI, 1 thread: 365s

3
1
15
977
It's actually not faster when strictly looking at how long it takes to format text, but dprint traverses files really fast and the Wasm scales better on multiple threads (this chart is JS warmed up, which the Wasm doesn't need to do).
Shows progress of the formatting getting faster at text, and currently being slightly slower than Node.js

ALT Shows progress of the formatting getting faster at text, and currently being slightly slower than Node.js

90
It's very hard to innovate on module resolution in JS because the TS user experience would be bad. TS needs something like content mappers, but for module resolution—you specify an npm package in a tsconfig and then TS sends it the raw specifier and referrer info to resolve.
1
9
430
Current progress compiling Prettier to Wasm.
Plugin in dprint, 32 threads: 22s Prettier 3.9.10 experimental CLI, default workers: 182 s Plugin in dprint, 1 thread: 260s Prettier 3.9.10 classic CLI, 1 thread: 365s

ALT Plugin in dprint, 32 threads: 22s Prettier 3.9.10 experimental CLI, default workers: 182 s Plugin in dprint, 1 thread: 260s Prettier 3.9.10 classic CLI, 1 thread: 365s

David Sherret@DavidSherretOct 11
Replying to @DavidSherret
I've been looking into getting prettier compiled as Wasm the past week with equivalent performance to v8... initially had some poor investigation into it (github.com/dprint/dprint-plu…), but I resurrected an idea I had in the past and it's looking more promising.

Investigation: running prettier as a Wasm plugin (too slow for now) · Issue #107 · dprint/dprint-...

This records an investigation into replacing the process plugin (V8 via deno_core) with a Wasm plugin, so nobody has to repeat it. The conclusion is that it works functionally but is too slow, so t...

github.com
1
17
2,143
A huge benefit is this allows distribution of the TypeScript compiler as JavaScript so it can keep working well in JS until JS can catch up.
Chris Tate
@ctatedevOct 17
New from Vercel Labs: tsc-ts TypeScript was written in TypeScript Then rewritten in Go We ported it back to TypeScript Compiled it with scriptc And it's 2-4x faster than tsgo
5
23
2,574
I had a similar idea when I became distraught after the TS team told me about their choice of Go... for several weekends I worked on a subset TypeScript to C# compiler that exposed things like Parallel.For Unfortunately, at the time, it was a ton of work so I gave up.
1
1
226
Nowadays this kind of stuff is trivial... super exciting, but also I feel bad for the TypeScript team because it's becoming easy for other people to come up with better solutions. You no longer need to spend millions of dollars to do this kind of work.
1
6
280
You can use bun's formatter in dprint now with `dprint add bun` (dprint.dev - code formatter for all your code formatters) It's also available to try in the playground here: dprint.dev/playground/#code/… Of if you want to use it in JavaScript: npmjs.com/package/@dprint/bu…
Jiahan Chen
@jiahan_cOct 12
👀 `bun lint` and `bun format` just landed - 600k+ lines added in one PR - 1,000+ built-in lint rules - Fast (of course) github.com/oven-sh/bun/pull/…

Add `bun lint` and `bun format` by Jarred-Sumner · Pull Request #44805 · oven-sh/bun

What does this PR do? Adds two commands: bun lint: a linter that reads ESLint and oxlint configs. 1,021 rules built in, type-aware rules included. bun format: a formatter that reads Prettier and o...

github.com
1
12
1,007
Try it out in the playground here: dprint.dev/playground/#code/…
1
1
214
I've been looking into getting prettier compiled as Wasm the past week with equivalent performance to v8... initially had some poor investigation into it (github.com/dprint/dprint-plu…), but I resurrected an idea I had in the past and it's looking more promising.

Investigation: running prettier as a Wasm plugin (too slow for now) · Issue #107 · dprint/dprint-...

This records an investigation into replacing the process plugin (V8 via deno_core) with a Wasm plugin, so nobody has to repeat it. The conclusion is that it works functionally but is too slow, so t...

github.com
1
1
1
1,386
After many years, finally got the Roslyn (C#/VB) formatting plugin for dprint (dprint.dev) compiled to Wasm so it runs sandboxed instead of launching an executable. Amazing how far C# has come.
Replacing the process plugin for `@dprint/roslyn` with a Wasm plugin.

ALT Replacing the process plugin for `@dprint/roslyn` with a Wasm plugin.

1
12
740
David Sherret retweeted
dprint code formatter@dprintfmtOct 4
Range formatting is now supported in the json, markdown, toml, dockerfile, and jupyter plugins. Additionally, most of the bugs in all these plugins have been fixed and many new config options have landed. Run `dprint config update` on your project to get the latest versions.
1
1
84
David Sherret retweeted
Matt Pocock
@mattpocockukOct 3
My hot take from chatting to @poteto is that we should use MORE abstractions in the AI age You can use them (combined with harsh lint rules) to reduce the design space available to the agent and constrain them only to good decisions. Combined with the fact that high-leverage abstractions let you do more with less code - so, more token efficient. Plus, unwinding the damage from a bad abstraction is much cheaper with agents. This runs counter to a lot of folks thinking that agents just want to read the raw code. They can, but they're not maximally efficient that way. Be braver! Design abstractions.
263
214
108
3,915
449,515
I understand this, but I find PRs can serve as a good starting point/context for landing a patch. It’s like an issue with token/time savings and some projects still need guiding AI. I really hope more projects don’t take this route—I want to continue contributing places!
Sindre Sorhus
@sindresorhusOct 1
Due to AI, I have disabled external pull requests on all my repos. Open source, as we have known it, was fun while it lasted. (It’s been 15 years for me) I will still maintain projects and handle issues.
3
12
1,384
That said, a lot of PRs I see seem like the person is using a bad model or they don’t know how to provide direction to AI. Other times they’re not targeted. In those cases, I don’t feel bad closing or just rewriting their PR and landing with attribution when it makes sense.
101
Nowadays you can do anything and it won’t ever tell you no.
zack
@zack_overflowSep 26
Looks like the gains were not from dropping down to asm And rather just better design: DOD (SoA), only re-painting what was changed, writev to reduce syscalls, etc All of which you could in Rust without asm!
1
6
562
One way you can bring value is explaining why something is a bad idea.
83
I know sometimes I just have to get over UI changes, but isn’t moving the back button far away to the top left corner in chrome mobile an objectively bad decision? Now I need to adjust my grip and stretch for it. Feels like c-suite design. I don’t think I’ll use “Ask Gemini”…
1
6
670
Here's a screenshot I found online of before. The back button is easy to reach in the bottom left corner.
1
2
137
Load more