Laurence Gonsalves

Many #git users think --force-with-lease makes a force push safe. It *is* strictly safer than --force, but: - It only guards against one narrow scenario - It's easy to disable that protection without realizing it

Many tools (like certain IDE plugins) will also fetch automatically in the background, which results in the implicit lease being updated behind your back.

Many #git users think --force-with-lease makes a force push safe. It *is* strictly safer than --force, but: - It only guards against one narrow scenario - It's easy to disable that protection without realizing it

Also note that if you git fetch right before git push --force-with-lease (without an explicit lease ref), it's exactly as if you had used plain old --force.

Many #git users think --force-with-lease makes a force push safe. It *is* strictly safer than --force, but: - It only guards against one narrow scenario - It's easy to disable that protection without realizing it

With plain --force, you'd silently nuke those new commits. With --force-with-lease, the push fails, because the remote sees that the branch isn't pointing where your local repo expected it to be.

Many #git users think --force-with-lease makes a force push safe. It *is* strictly safer than --force, but: - It only guards against one narrow scenario - It's easy to disable that protection without realizing it

The first time I was saddened by the news of Optimus Prime's death was when I watched Transformers: The Movie (1986) in theaters. The second time was today. RIP Peter Cullen

As a kid, Transformers was one of my favorite after-school shows, and Optimus Prime was my favorite character. I even dressed up as him for Halloween. (Visibility was *not* good. I ended up inadvertently "transforming" when I tripped over the curb of someone's driveway because I couldn't see it.)

We've been getting some renovation done on our home for the past couple of weeks, and this morning there was a terrible high pitched tone coming from this work. It was loudest in the hallway to our side door, and second loudest in my office, which is almost right above that hallway. 🤕

Outside, I noticed that a hose had been connected to an outdoor spigot near the door, but hadn't been completely tightened, so some water was dribbling out the connection at the wall. I tightened the hose connection a quarter turn.

We've been getting some renovation done on our home for the past couple of weeks, and this morning there was a terrible high pitched tone coming from this work. It was loudest in the hallway to our side door, and second loudest in my office, which is almost right above that hallway. 🤕

This sound had been going on all morning, and I assumed it was unavoidable. However, when the workers went on break, this tone continued, so I went to see if I could find its source. I assumed it was outside our side door, since inside the house it was loudest in the hallway leading to that door.

Ever since Google search and Google assistant on my phone started using different voices I've noticed that search's voice has begun using some pretty weird pronunciation in some cases. Like today I asked it to add a series of numbers, and discovered that it has a lot of trouble with small numbers.

I've been using Python for over 25 years now, but only got around to really learning how its type annotations work recently. It got me thinking again about something that seems like a bit of a "truism", but I think has some interesting implications...

For example, Python's type annotations. I don't think anyone would come up with "-> type:" as the syntax for return types in a new language. Python had the constraint that they didn't want the addition of type annotations to the language to have any effect on the meaning of un-annotated code.

I've been using Python for over 25 years now, but only got around to really learning how its type annotations work recently. It got me thinking again about something that seems like a bit of a "truism", but I think has some interesting implications...

Another example from Java is that several features were explicitly designed to work the "old way" if prior versions would consider the code valid, even if the old way was arguably inconsistent with the "new way", like the way ambiguity is resolved when passing arrays into varargs.

I've been using Python for over 25 years now, but only got around to really learning how its type annotations work recently. It got me thinking again about something that seems like a bit of a "truism", but I think has some interesting implications...

For instance: the use of "@" in both Java annotations and Python decorators. Prior to the introduction of those features, an "@" (outside of a string or character literal) was an error in both languages.

I've been using Python for over 25 years now, but only got around to really learning how its type annotations work recently. It got me thinking again about something that seems like a bit of a "truism", but I think has some interesting implications...

The first option means you're breaking backwards compatibility, as programs that used to do one thing now do something else. Many languages try to avoid this, but Python 3 infamously chose this option. A str in Python 2 is very different from one in Python 3, for example.

I've been using Python for over 25 years now, but only got around to really learning how its type annotations work recently. It got me thinking again about something that seems like a bit of a "truism", but I think has some interesting implications...

Even their outage map doesn't work.

App passwords can be used to sign into bsky.app, not just third party apps. You can't create an app password while signed in with one. The UI doesn't do a great job of explaining *why* you can't, though. It highlights the name field as if that's the problem, and doesn't show any error messages.

Most apps that let you authorize other apps with OAuth also have a page somewhere in their UI to revoke these authorizations. I would have expected something like this to be in Bluesky under Setting → Privacy and Security (which is where App Passwords are managed), but it isn't there.