UNDERDOG of PERFECTION

First, yes… I know you’re taking your chances when you install a beta OS on your primary device. As it happens, I was willing to take that chance with the macOS 27 Developer Beta on my primary Mac, because a) I had heard (and have seen myself) that it is very stable for a beta OS, and b) the Liquid Glass UI nonsense in macOS 26 was so bad that I was willing to risk running unstable software to get away from it.

And so, I have been running the macOS 27 Developer Beta for the past month, with very few problems. Oh, definitely not no problems. With an earlier build, Final Cut Pro — one of Apple’s own apps — was an absolute crash fest. In the process of editing a fairly simple video, I had it crash at least two hundred times in a single day. Fortunately, Final Cut is constantly saving your work, so I never lost a single bit of my project through all of those crashes… although it was definitely challenge to keep my ADHD brain focused to remember what it was I had just been working on 5 seconds ago.

The whole purpose of a Developer Beta OS is for developers to test their software against the new OS and see if there are any bugs they need to fix. I’m pleased to say, Apple’s own Final Cut team is certainly doing that, and with the latest beta, Final Cut seems to be back to its usual rock-solid stability.

Now granted, it may not technically be the purpose of a Developer Beta for users to test third-party developers’ software against the new version of the OS, but guess what… it’s gonna happen.

There are a couple of apps I rely on constantly throughout my workday, both of which have exhibited a couple of bugs with the new OS. They’re not showstoppers, just minor inconveniences.

I don’t want to name the companies, but let’s just say, one is a relative newcomer, building a client app for Gmail. The other is one of the most storied indie Mac development companies around, and I’ve been using their FTP client for literal decades. But frankly, they seem a lot more interested in their bright yellow handheld video game device than in supporting their crusty old Mac software at this point.

I submitted bug reports to the two companies a day or two ago, and I received responses from both of them within a half hour of each other today.

The mail client company thanked me for submitting my bug report, and let me know that there was already an update available that should fix the issue. I ran it, and it does.

The FTP client company sent me a cursory response linking to an FAQ page on their website that basically says, “hey dumb-dumb, don’t complain to us if things don’t work when you’re stupid enough to be running a beta OS.” OK, it maybe wasn’t quite that mean-spirited, but it felt that way.

Here’s the thing… maybe neither of these companies really wants or expects user feedback during the Developer Beta period, since they are — or at least should be — conducting their own extensive testing of their apps with the beta OS. But when they get that feedback, they should at least have the courtesy to feign appreciation, rather than assuming the worst about their own user base and dismissing their very legitimate concerns that bugs in an app that’s critical to their workflow may not be resolved before the final OS release.

Honestly, at this point I would have long since given up on that company’s FTP client, but sadly, the market for FTP clients is pretty slim. What am I supposed to use, Filezilla? Yuck. At least this is, in John Gruber’s words, a “Mac-assed app,” even if it’s buggy and feeling a bit neglected these days. Kind of like me.

Chapter 1

Chapter 2

Epilogue

Obviously, Google itself has a problem too. There is no official Google Authenticator app for Mac. And given the whole purpose of authenticator apps, I would only want one that was official.

“Let the market decide” and all of that nonsense. But I think Apple and Google both have a responsibility not to let their users be misled, especially about their own products. I’m of the opinion that, if you search “google authenticator” in the Mac App Store, nothing should come up, because users’ implicit expectation is that they’re going to get a Google-made app, but there isn’t one.

Likewise, I find it almost comical that Google’s AI Overview, of all things, is saving the day here. No, there is no official Google Authenticator app for Mac. But once you’re past the AI stuff, the first result is a sponsored link for a knockoff app.

The second link — the first “organic” link — is for the official Google Authenticator app in Apple’s App Store. But that’s the iOS version, the only real version of the app, and it won’t run on a Mac.

Which all gets me to my point. 2FA is better than passwords alone, but it’s a pain in the ass. Having an “authenticator” app that is always at-the-ready with your 6-digit code is more efficient (and more secure) than having to wait for an SMS message to come through, but it would sure be nice not to have to rely on a second device, especially when I don’t have my iPhone right by my side.

Of course, now that I think about it, the only times I’ve needed access to Google Authenticator when I’m using my Mac, and I haven’t had my iPhone right by my side, have been because my iPhone is plugged into a charger in another room.

That’s been happening a lot lately, because my poor, old, long-in-the-tooth iPhone 13 mini can barely make it to lunchtime without needing to be charged at this point.

All things considered, the answer to my conundrum of there not being an official Google Authenticator app for Mac is astoundingly obvious.

I need a new iPhone.

I’m working on a new WordPress site that has long dropdown navigation menus. I wanted to make the dropdowns two-column, but only if there are at least five items in the submenu.

The two-column part is easy:

.sub-menu {
  column-count: 2;
}

I’ve (finally) been learning more about some of the “new” conditional selectors in modern CSS over the past year or so. It occurred to me that there might be a very easy way to do this, checking if the submenu has at least 5 items. But I wasn’t sure it would work, so… I tried it:

.submenu:has(.menu-item:nth-of-type(5n)) {
  column-count: 2;
}

Much to my delight, it works! Submenus with four or fewer items are displaying as a single column, and the ones with at least five are displaying as two columns.

Now, this of course is using CSS classes WordPress creates for its navigation menus, so if we were talking about barebones vanilla HTML5, you could use this:

ul:has(li:nth-of-type(5n)) {
  column-count: 2;
}

This also doesn’t really account for tertiary nested menus, but I only have this site set up to display two levels of depth on its nav menu.

Thanks to my On This Day plugin (not the plugin I was asking ChatGPT for help with in my last post), it’s really easy to “surface” (I hate that as a transitive verb) some really old posts here on my blog, which has now existed for about 24 years… frighteningly close to half of my life.

I clicked on the link to a post I wrote on this day in 2008. It was about links to articles I didn’t want to lose track of.

And it’s interesting for a couple of reasons. First, there was — was — an image at the top of the page. What it was, I have no idea. But that, like all of the other images on the blog before about a year ago, are now gone. I deleted them, because I was tired of getting harassed into paying hundreds of dollars out of the blue for random images that were lurking, unseen by anyone, on ancient blog posts no one reads.

I’ve usually been pretty good about only posting images I created, but when you’ve been at this for two decades, there are going to be a few times when you get lax. And there are a couple of news agencies who have contracted with a company that is absolutely ruthless about shaking down unsuspecting bloggers. What a racket.

But I digress… I wanted to talk about the links on that page.

They’re both dead, too, but they still go somewhere. The domains they linked to are still active, but now they redirect to 404 pages on completely different domains. Whoever owned those domains either sold them or let them lapse, and now they’re something completely different.

A lot of people want to fight “link rot.” But not me. I relish it. The Internet is an Orwellian memory hole, and fixing broken links in blog posts from 2008 just accelerates that. The pages I’m linking to have long since disappeared. But their URLs remain in a zombie state here on my blog as a testament to the fact that these things once existed.

Like the website Veer.com. I don’t even remember exactly what it was, and it had been years since I had last contemplated its existence, but once I saw that link, my immediate reaction was, “Oh yeah, I remember that site. Well… no, actually I don’t, but I remember that was a site, and I used to like checking it out semi-regularly.”

There are so many lost worlds like this. No longer present anywhere on the Internet*, but lingering in the dusty backrooms of Gen Xers’ minds.

*I lied, of course. The Wayback Machine exists. I managed to find a snapshot of that URL from June 10, 2008, about a week before I posted about it.

It even still has the image of the coffee mug I was interested in! But I’m not going to post it here. I know better. Finally.

The following originally appeared as a post on my YouTube channel.

Trigger warning: not entirely negative stuff about A.I.

I have the script written for the first half of my “5-String Bass Showdown” video (the second half will be unscripted), but I may not get around to making that video this weekend, because one half of my “day job” (indie WordPress plugin developer) is bleeding over into my video time.

I spent about 10 hours sitting in front of my computer today. The first half of that was beating my head against the wall over an extremely obscure problem I was having with Block Editor development. (I am an old school PHP developer and I have never bothered to learn React, so I keep the dev side of the Block Editor at arm’s length.)

Anyway… after literally 5 hours of almost no progress, beyond homing in on exactly where, but not what, the problem was, and getting nowhere reading 6-year-old GitHub discussion threads and even older StackOverflow threads, I finally resorted to…

Asking ChatGPT. (With DuckDuckGo’s Duck.ai as an intermediary, of course.)

Luckily I had gotten to the point where I could succinctly state in one sentence exactly where I was stuck, and provide ChatGPT with the relevant JavaScript function from my code.

In less than 5 seconds, it returned a cogent explanation of why my code was having an issue, along with a minimally rewritten version of my function that solved the problem.

It even mimicked my somewhat idiosyncratic coding style.

As Dr. Gnome would say, “Uff da.”

After finally taking a break around 3 PM to eat lunch and ponder my existential crisis, I decided to throw a few of my other “back burner” issues at it. The first one, which I thought would be the simplest, ended up being unresolvable. Two others were simple code refactoring (JavaScript is not my strongest suit), and it knocked those out of the park.

So… I’ve come to realize, A.I. like ChatGPT does have a role in my work. It can be a good partner to troubleshoot and optimize the code I write in languages I don’t know like the back of my hand.

But to that end, I think of it in a very similar way to how I’m using the stem splitter in Logic. It’s a tool to streamline the things I already do, and fill in the gaps. But ultimately I still have to be the one doing the thinking.