Dusty Pomerleau

Substitutes If guanciale is unavailable, pancetta can be used as a substitute, though the flavor will differ. Wikipedia montecatini.com.au

Hypothesis:

The ingredients in this sentence can be replaced with any 2 distinct nouns, and the statement will still be true.

I think at a certain level of expertise, differences of opinion become the rule, rather than the exception. Opinions converge as people learn more, and then among experts they start to diverge again. That's the case in my own field, at least.

It's a form of Gell-Mann, I think. For domain experts, all of the talks will be inaccurate in this way.

en.wiktionary.org/wiki/Gell-Ma...

formal verification with le creuset

Release v0.38.0 · jj-vcs/jj About jj is a Git-compatible version control system that is both simple and powerful. See the installation instructions to get started. Release highlights Per-repo and per-workspace config is now ...

v0.38.0 is out!

github.com/jj-vcs/jj/re...

I am helping organize the first ever community Rust India conference 2026! It will be soonish, if it interests you or want to be involved lmk!!!

If you do it with a figma, they'll soon ask you to do it with a figment of their imagination :)

$ jj describe -m "Hello, world!"

I just made the typo `podamn` while trying to start a container, and my brain has already decided that this is a permanent change.

What is it about installing something that improves your experience, compared to simply having it open in a browser tab?

I wanna say this out loud. A few months ago, I told Evan that @voidzero.dev should stop sponsoring me. I burned out, and I couldn't help Vite as before. But they decided to continue supporting me while I was down. And by doing so, they helped me get back on my feet sooner. Thank y'all ❤️

Honestly, I think the solution is the browser. All but one of the apps you mentioned already have a web app, so it's just a matter of putting effort into feature and performance parity. In more demanding scenarios, wasm is available, but it's a matter of convincing companies to leverage it.

Do you have an iPhone? Apple Maps has a "share my location" menu option that just sends a one-off set of coords rather than ongoing location sharing.

UB40 - The official global website for UB40 Welcome to the official UB40 website - discover upcoming tour dates, latest news, find out about new releases and visit our merchandise store.

ub40.global

Take me down to the Parallax city where the far moves slow and the near moves quickly

This post is right-handist.

Oh no.

Once you notice the mix of camel and kebab case in nix files, you can't ever unsee it.

Ah, my mistake, there is actually a dedicated module for each (sqlx-postgres::types::{chrono, time}), so it would be a bit of work.

Looking at it, there really isn't that much code required for any given DB:

github.com/launchbadge/...

I know it has been raised before, but maybe I will ask again if they are open to a PR, now that jiff is a bit more mature.

It's been a while since I looked at, so maybe I need to revisit, but I thought that the `query_as()` (function - as in your example) was Ok, but that `query_as!()` (macro) required custom support within sqlx.

I would love to switch, but I'm heavily dependent on `sqlx::query_as!()`, and I had the impression that isn't quite working with jiff?

I'm finally starting to understand how awesome UUID v7 is. I just deleted every `created_at` column in my Postgres schema. Poof :)

inside out :)

expected Result<Option<Record>, Error>, found Result<Option<Record>, Error> rust-analyzer (E0308) [100, 10]

Team... Can we please give our error types better names?

I was thinking about the situation backwards! The method for session creation gives mutable access to the generated ID, so that you can check for collisions. It just occurred to me that, even without a collision, I can use that mutable access to store the bytes of a fresh UUID v7 in the i128 :)

// We can't (easily) use UUID v7 for the session primary key, because tower-sessions mandates that // the `Record::id` is of type `Id`, which holds an i128. Instead, we use a UUID v4 that is // technically invalid, but represents the Rust i128 exactly, and therefore is completely // reversible. Postgres will tolerate this, but this UUID is not compatible with any external tool // expecting a compliant UUID. All of the IDs in the `session` table need to be generated in Rust // code for this reason. pub trait ToUuid { fn to_uuid(&self) -> Uuid; } pub trait ToId { fn to_id(&self) -> Id; } impl ToUuid for Id { fn to_uuid(&self) -> Uuid { Uuid::from_bytes(self.0.to_be_bytes()) } } impl ToId for Uuid { fn to_id(&self) -> Id { Id(i128::from_be_bytes(*self.as_bytes())) } }

Anyway, here's what I'm currently thinking of doing:

I need any conversion to be reversible, and I need to index on whatever type I cast the i128 to, because I'll primarily be looking up sessions by ID. I don't care about sorting.

I can store as text and index on that (slow).
I can make a fake, noncompliant UUID from the i128.
I can store as numeric.

I am up against this weird situation where the tower-sessions middleware uniquely identifies sessions with an `Id` that is hardcoded as a newtyped random i128. But the rest of my DB uses UUID v7 IDs, so I have this one outlier table, and exploring my options has led to quite a journey:

When I first started interacting with the Nix ecosystem, the approach felt pedantic in the extreme, but issues like this slowly radicalize you into celebrating that aspect of Nix.

I would kill for some static types, though. The function signatures are brutally dependent on convention.

I think within that subculture, the fact that the person who died was a white gun owner has an inverting "one of us" effect that makes it harder for them to have their usual response.