jd3

iTerm2 feels overwhelming to me. I think the default Terminal.app is good enough.

I agree with your point, though I can easily modify ~/Library/Messages/Attachments or ~/Library/Containers/com.apple.MobileSMS/Data on macOS, which is a platform similarly secure to an iPhone when you take sandboxing, sip, etc. into account. So Apple already supports macOS, which can be tinkered, no? I guess the argument is that by locking down iOS, a platform which was never open, they just never needed to handle that class of security risks which are introduced when customers have more freedom.

I did actually discover a bug in iOS (related to photos which have failed to sync to icloud for the past four years) by investigating why there were over 23k empty folders in ~/Library/Containers/com.apple.photos.ImageConversionService/Data/tmp/, fwiw.

tl;dr — free up disk space and delete ~/Library spotlight index and medianalaysisd caches

There's a really nasty bug in Golden Gate 27.0.1 which causes ANECompileService, mediaanalysisd, duetexpertd, and other apple intelligence-related features to thrash and permanently spike your CPU. I tested this yesterday on a machine with ~12GB of space left and the result was launchd writing 163 GB and ANECompilerService writing 70 GB in less than 24 hours. The noticeable effect is that the AI text-recognition for photos in QuickLook will not work and repeatedly download and compile ~/Library/Containers/com.apple.mediaanalysisd/Data/Library/Caches/com.apple.mediaanalysisd/com.apple.e5rt.e5bundlecache items. One in particular repeatedly fails and downloads a 69.9 MB file with model.mil and weights1.bin inside of the package folder. This will happen every 6 minutes, which further compounds the severity of the issue since the bug itself is due to low disk space. This folder has risen to 2–5 GB for me before, which I had to just delete manually. The filenames are <HASH>.tmp.<number which increases every time>.<random>.bundle.

I figured this out by watching `$ log stream` and seeing what was happening, which was lack of disk space while ANECompilerService was compiling ai features after opening a file in Preview or in Quick Look in Finder.

I fixed this by freeing up ~10GB (bringing me up to 20GB) and deleting the ~/Library/Metadata/CoreSpotlight and ~/Library/Containers/com.apple.mediaanalysisd/Data/Library/Caches/com.apple.mediaanalysisd/com.apple.e5rt.e5bundlecache/. After rebooting, I left it on overnight to reindex everything.

XQuartz performance has been broken for a couple of years now and I believe it is primarily due to an OpenGL issue.

I don't even bother writing mermaid by hand when I'm making decision trees. Instead of visualizing it using PrecisionTree in excel (I just use that for the actual data modeling), I've been first modeling the tree in plaintext, then pipe it through claude with a previous tree I've made as reference. I then ask it to use the content of the current tree and match it to the style of the previous one.

TIL! Fascinating they bothered recompiling the driver at all up until then.

That overengineered isight camera worked through multiple adapters up until the introduction of the M1, when they sadly finally removed the driver. I used to use it connected to my intel macbook pro for meetings in the the late 2010s for fun.

Google Workspace has one of the most perplexing interfaces I've ever had the displeasure of getting to know and use.

Namely, it seems impossible to archive users despite the fact that we have a business account and have already archived numerous former employees.

> You contribute to projects run by others with the understanding that others run the project, is this not the default assumption others have too when contributing to FOSS?

There are certain unspoken rules you don't break out of a of tradition of process and decorum — this decision broke that rule, overturning 5+ years of accrued experience and institutional memory, so I can understand why this might feel like a rugpull which would reasonably upset the average contributor.

These antics are acceptable in a BDFL project where the stakes are low, but try this in an actual major project with thousands of contributors and see how it goes.

I think the best case scenario I can think of is the microsoft typescript js->go conversion, whereas the bun zig->rust is probably the worst case.

I wrote a multi-page essay about this for our condo association last year. There's lots of complexity and externalities here — Pepco grid modernization, PJM interconnection cap auctions, AI buildouts in NOVA, etc; it's a really complicated issue.

The US app is still laggy (even on the iPhone 17 Pro) and constantly logs you out. My theory is that they set the login timeout to a low number to make it harder to accrue points.

We dealt with this headache for 7+ years at my former employer. Thanks so much for this.

The effects aren't too far off from reports of the maduro raid, though my understanding is that the US just used a standard LRAD

I stopped reading when I realized it wasn't a deep dive into the most interesting question I had, which is the technical hardware design process and finding a factory to actually take your design and manufacture it.