Hacker News new | ask | show | jobs
by bavent 248 days ago
Beyond that, it’s buggy and inconsistent. In dark mode, text boxes suddenly become light mode when they expand to add a second entry line. Buttons aren’t aligned within their containers properly. Some times buttons are in light mode when my phone is in dark mode, or if I open an app, it starts out in light mode and then suddenly switches to dark mode after a second or two. There is a noticeable lag when I back out of a message before it loads the other conversations in iMessage.

I think I had tracked 15+ things I would easily qualify as bugs the first two days after upgrading my phone - this would be absolutely unacceptable where I work, and we aren’t a trillion-dollar company with psychotic hiring standards.

Was this even QAed? I don’t like the look, but that’s a personal thing, these are actual issues that are not subjective.

4 comments

In my opinion Liquid Glass is still an alpha release. Nothing is really finished, it's conceptually unfinished, the changes are not well thought through, and it's really buggy.

I think they failed with Apple Intelligence (also a mess, without being useful) and needed something big. So they planned this big design change. When they realized they failed miserably, it was too late to undo it.

Liquid Glass was clearly rushed to cover for the total failure of Apple Intelligence.

Even the design being criticized distracts from fraudulently selling phones based on features never shipped.

The recently presented ChatGPT apps are what Apple Intelligence (and Siri) should've been. Some chat/voice interface, that can access data from installed apps and trigger actions.

It should've been a home run for Apple. ChatGPT starts with zero existing apps, Apple has one of the biggest app ecosystem, and with (Siri) Shortcuts they already have most of the necessary interfaces available for years.

Apple still has an ultimate advantage.

They have your context. OpenAI doesn’t know where you are. It doesn’t know what you bought or when you last called your wife, it can’t know your heart rate or your work schedule.

Apple can turn it around.

Great AI is a good model with lots of context. Your model can be the best, but if you need the user to provide the context it’ll never be a great experience.

After working with Claude code for a while now, I’ve become much more aware of how to convey context to a machine, and just how poor some humans are at doing it in conversation.

Your AI product is toast if you need people to make it work.

Yes, Apple still has the advantage over OpenAI. But OpenAI can also release some iOS and Android integration layer, that allows to connect with installed apps on the device.

If Apple doesn't get their act together with the next iOS release, it could be too late.

OpenAI can't, they're completely dependent on Apple and Google permitting such a thing. Unless you have a particular way in mind they could currently achieve this?
I think this will be an Apple Maps situation. Embarrassing initially, but a few years later, there’ll be a perfectly usable product.
I still find Apple Maps to be awful. It consistently gives poor directions. For example the exit you need to take from the closest major road to reach my neighborhood splits once you are on it. If you continue straight you reach a stop light where you can then take a left and then continue 1/8th of a mile to the turn for my neighborhood. Apple Maps will instead have you go to the right when the exit splits and then have you continue another 1/2 mile before taking a u-turn and heading back the 5/8ths of a mile to turn into my neighborhood. Google Maps does the right thing. I now warn visitors to use Google Maps or to ignore these directions from Apple Maps.

I also live near a large city and a couple of smaller cities with busy downtown areas. In each of them the main streets are virtually impossible to perform a u-turn on because of the large amount of traffic. Apple Maps will insist on giving directions which involve taking u-turns on these streets. Google Maps will instead route you the easier way around the block instead of insisting on an impossible u-turn which in the end is slower because of the difficulty in actually performing the maneuver.

Also I find the directions from Apple Maps when taking an exit which further splits into multiple exits to be highly confusing. The spoken directions from Google Maps is much better in these circumstances.

Every new release I try Apple Maps again just to see if it has gotten better in these circumstances and every release I am disappointed.

Well it's useable sure but also not very good/useful. From a directions/routing standpoint it's pretty decent and I'm OK with how the route planning works. But they miss many POIs and the data they have is often stale, and their version of street view, while smoother, is not up to snuff at all.

I think it's really annoying and a major reason I have stopped using it even though I was an advocate at first. Apple can't be bothered to invest as much as Google did to have a proper open map system, with a good web version where people/business can post/add data easily. At this point the privacy stick is tiring because we don't get anything from it and they will comply/sell the data the minute they can profit from it anyway (as they have shown).

So, you just end up paying more for a product that is clearly worse and won't become much better because of Apple's ideology and how stingy they are. They generate a lot of cash but are unable to invest it in proper competitive software.

There are many bad things to be said about Google, but at least they manage to serve pretty good software that is open to everyone...

the whole maps debacle was a plot against Forstall
A 'plot'? That must be some kind of weird online conspiracy theory. Apple Maps worked poorly on initial release for reasons that are entirely unmysterious (the lack of comprehensive and accurate geographical data). Forstall was of course fired as an eventual result of this, but the idea that the whole PR disaster was engineered as part of a scheme to oust him is just daft.
Also, I think they will work through bugs, but the losers with the older devices aren’t going to get it.
This is a fun one I think:

- Use dark mode

- Go to wikipedia (or any white page)

- Open the keyboard

- Watch the keyboard start in light mode and then resize very weirdly within its container as it switches to dark mode

Atleast it does on an iphone 12

One of the issues is almost certainly that the app developers didn't add the UIDesignRequiresCompatibility = YES[0] item to the Info.plist.

Set that, and it doesn't use Liquid Glass in your app.

I set it for all my apps. One was designed by a professional designer, who absolutely defecated masonry, when I showed him what it did to our app.

I'm worried that Apple may end up ignoring that flag, and will force us to use LG. That would suck. It says that it's temporary, but I'll bet that Apple will be hating life, if they ignore it.

I'm not freaking out about Liquid Glass, but I don't like it. I completely agree that it is quite unusable.

[0] https://developer.apple.com/documentation/BundleResources/In...

I’m referring entirely to built-in Apple apps - Mail, Messages, etc. The in-house apps can’t even get it right, which to me means:

- devs are so siloed, nobody knows what’s going on - product is not communicating anything outside of individual fiefdoms - there is zero QA testing - no designers are actually signing off on the final results

…which all seem pretty typical for a large bureaucracy, I guess I just had higher expectations of Apple, since we pay a premium for their products. Some of these bugs are frankly pretty embarrassing.

Such reclusiveness is not an obligatory property of large corporations. Say, Google around 2011-2015 may have had fiefdoms, but at least things were quite transparent, you could know what other departments are doing, and see all the code. Facebook circa 2020 was surprisingly transparent and peer-to-peer, at least in the area I touched, messaging and storage infra. I've seen companies 1000x smaller that had incomparably more reclusiveness and opaqueness.

What I hear about Apple sounds more and more like what I used to hear about Microsoft, especially Microsoft of Ballmer times, when teams inside it clandestinely warred with each other, instead of cooperating.

Apple has this vision-driven culture, and the inclination towards internal secrecy, so that competitors won't steal their thunder. It worked relatively well under Steve Jobs, and whoever he assigned. It worked far less successfully when Jony Ive's ideas of usability made Macbooks into visually more sleek, but less loved devices. Whoever came up with Liquid Glass, has some interesting vision, but the gimmick value in its current implementation seems to dominate, and the usability shortcomings seem to be ignored. Technology-wise, it's half-baked. This means to me that Apple internally not in a good state, the leadership has trouble hearing the voice of reason.

Apple of course has an immense inertia. But giants like Nokia or General Motors also used to have an immense inertia, wads of cash, and dominant market positions.

Apple’s Mail app has been intensely buggy for years, and the bugs rarely get fixed.

(Search is comically bad.)

Search is so so bad. I just want to find an email with a word in it. All my Apple Devices fail at this.
Search is bad everywhere. If I open Settings, and then search for application $X, no results. If I search for $Y, $Y shows up. They are alphabetically next to each other, and I can see them both if I open the Applications submenu and scroll down to access individual app settings.

Why does one show up and one doesn’t? The one that doesn’t is a built-in Apple app, too. They both have identical settings for “show app in search”. This worked fine before iOS 26.

A lot of them are SwiftUI apps.

I feel that SwiftUI is not ripe. I use it for one of my apps (I have to, in order to use the charts), but it’s too limited to use for anything else.

It’s so bizarre. I wanted to use it for a menu extra and something as simple as animating the icon couldn’t be done. There are several of Apple’s own apps that use animated Menu extra icons and they’re probably doing the same hybrid AppKit/SwiftUI workarounds.
I won’t use UIViewRepresentable. I feel that it’s a kludge, and kind of negates the whole purpose of SwiftUI. I know that some of the “native” types are probably UIViewRepresentable, under the covers (like maps), but I feel as if it’s a “duct tape” solution. Also, some of the code gymnastics that I need to do, in order to implement “non-standard” functionality, are pretty crazy. SwiftUI makes it absurdly easy to do stuff that follows the intended workflow, but completely falls down, if you stray off the path. UIKit complains, but grudgingly goes along with you.

I actually want SwiftUI to work. I think they have a good idea, but it’s a massive undertaking, and really, breathtakingly ambitious, when you consider what it’s trying to do.

UIKit represents a mature tech that has been refined since 2008, and a lot of that is based on lessons learned, implementing AppKit, which has been around forever (especially if you consider that it came from NeXTSTEP, which probably started in the 1980s). With AutoLayout and UIKit, I can do pretty much anything I want.

> One of the issues is almost certainly that the app developers didn't add the UIDesignRequiresCompatibility = YES[0] item to the Info.plist.

Ah yes, let's require all developers scramble to try and fix their apps instead of spending time to actually fix and polish the design system we force down everyone's throats.

Sounds like it was QAd by AI.
And not just any AI, probably Apple Intelligence given the level of quality.