- "Unicode" button label way off center
- The 8/10/16 selector being off center in its own position
- The indicators for bits 31 and 63 are not aligned with each other
- x and + not being horizontally aligned (I believe this is an icon-font issue, seen on HN before so knew to look for it)
Also, the "32" label sits directly centered under the bit above it, but literally none of the others do, they're wherever-the-fuck.
The x, + etc don't look centered vertically either -- compared to the numbers to the left of them.
2's and 1's look like they are a different font size to everything else.
The padding on the buttons at the top is hideous -- the downstroke on the y almost touches the outside of the button.
I fear how awful this looks in localized versions, if they made any.
However, I suspect they both have the same ("Incorrect" seems too harsh a word... "Visually imprecise"?) layout constraints and they look different in practice because "63" is a wider number than "31".
Its actually the center of the button, I made a video for context: https://imgur.com/a/1Y9O8dS
> The 8/10/16 selector being off center in its own position
Might be due to the image compression, it looks fine on my MB.
Upvoted for putting in the effort, and because you make a correct point.
But the Unicode button is perceptually off center, because ASCII is a smaller word, and there's no visible boundary between the buttons. This comes up a lot in iconography, the classic example is a play triangle (like the media control) in a circle. Placing the triangle in the geometric center won't look centered, it needs to be a tiny bit to the right of that to account for the shape.
No separation between the buttons means you can't see the bounds which the words are centered in, so it looks off.
The 8 and 10 have the same problem, for the same reason. A visible background-gray line between the buttons would solve this problem, it should be 'squircled' to make it I-shaped and match the outer edges.
- ASCII is off center ~43/50 pixel margins
- Unicode is off center ~20/25 pixel margins
- Both have different margin sizes
- The button sizes of both are the same.
- The Hide button is offset from both 8/10/16 selector and ascii/unicode buttons
- Even if everything was correct, because there is no contrast between "Off" and background, it's going to look wrong anyway
Edit: zooming in closer it's maybe not outside the box at all, but there's some odd aliasing artifacts or something making the space above the highlight look bigger than the space below.
Honestly I don't think it makes it any better if the Unicode text is theoretically centered; the fact that there's zero separation between the options, and such poor spacing that it's difficult to tell and feels awkward either way is still terrible design.
The bug turned out to be in CFNumber, in Core Foundation. CFNumber does a lot of fiddly stuff at the bit level for performance, and one of their optimizations for exponentiation was incorrect. Somehow it was never found by tests or due to buggy behaviors it created in other apps, but by someone clicking buttons and thinking critically about the output.
Source: have made this mistake. Have been flagged.
And typing in "(-22)^21" gives "-71100888972574851072", but wolfram alpha insists it should be "-15519448971100888972574851072".
Looks like there are still bugs here.
At first I thought it was just an overflow error but no it's nothing like that. The math is indeed very clearly broken, as I play around with it on Sonoma on my M1.
I'm genuinely shocked. I though this kind of floating-point math was rock-solid, tested thoroughly over the decades.
"(-20)^21" = -2.097152e27 "(-22)^21" = -1.551944897e28
(-20)^21 = -2.097152x10^27 and (-22)^21 = -1.5519448971*10^28
That always amused me.
also mentioned by Raymond Chen, https://devblogs.microsoft.com/oldnewthing/20040525-00/?p=39... :
"Today, Calc’s internal computations are done with infinite precision for basic operations (addition, subtraction, multiplication, division) and 32 digits of precision for advanced operations (square root, transcendental operators)."
And they fixed the square-root-of-a-perfect-square bug a few years ago, https://www.reddit.com/r/Windows10/comments/89s53g/microsoft...
That post points out it’s probably just subpixel stuff causing the issue, but I think my thick, cheap glasses at the time were adding a layer of chromatic aberration to something that was already visually confusing.
I assume it’s kind of gone away at this point with all the high DPI screens these days. But I remember thinking at the time, if there was a public bug tracker, that issue would be a fun one.
You can't compensate for chromatic aberration with a coating. You need a compound lens made from multiple elements each with a different dispersion, e.g.:
https://en.wikipedia.org/wiki/Achromatic_lens
More expensive glasses lenses usually have worse chromatic aberration than cheap ones. The cheapest material for glasses lenses (PADC, often called by the brand name CR-39) has one of the best Abbe numbers (measure of dispersion).
I would recommend this to any programmer who uses high-contrast syntax highlighting. To me, it felt fatiguing every time I noticed differently colored words scrolling slight further than other words on a terminal screen on the same line.
This is directly contradicting the main purpose of glasses: to see clearly. So it's actually somewhat less safe to e.g. drive with glasses that have major chromatic aberration. No idea why optometrists brush it off as a minor glitch.
This is the same reason why window gaps are so popular in all tiling window managers. It just looks better.
It's a hard problem to solve optically and requires specially shaped lens. It's a common issue in telescopes, with higher end expensive scopes having these specially shaped lenses to reduce this effect.
> In conclusion, the off-center “x” is real and probably an artifact of the display or how it is rendered. It is unlikely that it is the result of chromatic aberration.
The icon is mis-aligned, or its the different color subpixels of the screen that are not produced at the same place. Tradidionally, red is to the left.
> That post points out it’s probably just subpixel stuff causing the issue, but I think my thick, cheap glasses at the time were adding a layer of chromatic aberration to something that was already visually confusing.
Now that you point it out, the X is way off center on my up-to-date M2, so I took a screenshot with default display settings and zoomed in to look at the pixel work.
The X is rendered asymmetrically. It appears to be about 0.1 pixels too far to the left and down, since the antialiasing has shaded pixels "outside the X" but only on those sides. The antialiased render of the red circle is symmetric. This matches what I see without zooming in and rules out my glasses.
I wonder if someone fixed the bug for low-dpi displays where subpixel rendering mattered a lot, but did so in a way that hard-coded whatever Apple shipped 10 years ago. Maintaining tall piles of hacks is hard.
Alternatively, maybe their font renderer is getting wobbly in its old age. The window manager is my #1 complaint about this laptop, but crappy font rendering vs. well-configured Linux is also on my list.
Which led to people like me making a fool out of themselves. Always been using Android, and listened to iPhone users singing the praise of the amazing UI and UX of iOS. Well, eventually iPhone 12 Mini released so I figured, "why not give it a try, can't be worse than my current Motorola Moto G gen4 right?"
Well, it is worse. I still have the phone because it still works, but that was my first and last iPhone. Everything is dog slow, not because poor performance but because of slow animations. Same on Android by default, but at least I can speed it up. And the UX makes you jump through hoops, things are impossible to discover unless you watch tutorials on YouTube, and the amount of UI bugs seems sky-high for something that sells itself as "Premium".
And then CarPlay is just an abomination! Even the most basic things like "I'd like to answer a call while still being able to see the map I use for navigation" seems to be completely ignored and it honestly doesn't make any sense at all.
Ugh, I almost look forward to accidentally dropping the phone so I can go back to having a non-distracting experience in the car again.
Edit: I just remembered the most egregious issue: How can I see the current year without having to open up a separate calendar application/put a huge widget on my home screen?
Today, it is more about maintaining your suite of apps, the Cloud with all your data, the little blue bubbles in your group chats, and a host of other issues that are more a priority for choosing one platform over another, for most people. If I were to switch to Android now, it would be a huge PIA considering the 10+ years of platform integration and thousands of dollars of app purchases, iCloud, etc, that has made up a significant part of my digital life. I'm sure it would be similar for people going in reverse. Apple knows this, hence why services have become an essential part of their business.
The past 10 years or so? Everything has gone out of the window. No one is left at Apple who cares.
Also quite frequently I'll swipe up to view notifications beyond the fold and they'll end up in weird places, like they'll jump further up than they should or jerk around.
There are so many cases where I touched a button and it’s so slow that I tap again, but when it finally responds, it does the thing twice or changes the UI under me and I tap a different button.
Or it changes color/flashes to acknowledge the touch, but does nothing until I’m super patient and try it again and it works.
Or it does nothing to acknowledge my touch and doesn’t execute the action, so I question my sanity.
The point is that it’s so inconsistent that I don’t have an evidence-based guess at the root cause. My gut says it’s the overuse of dispatch queues.
IO-blocking animations are everywhere on iOS, and sometimes they result in overlap (e.g. you can activate a widget and open an app if you press an app icon too fast after opening a folder). But having buttons on iOS animate in response to touch but not engage any further is mindblowing and infuriating.
It's also filled with obtuse interactions. (Did you know the iPhone's calculator app has extra buttons? You have to use the control center, unlock your screen rotation, and then rotate your phone to access it.) (Did you know you can erase digits by swiping left or right on them? You can't _access_ the hidden digits of precision this way.)
It needs to work on it for at least five more years, meanwhile you can buy one of the many inferior iPad calculator apps that are not hindered by Apple's vision of greatness.
Meaning you cannot have reversed (aka natural) scrolling on a touchpad, and standard scrolling on a mouse at the same time without 3rd-party software.
I’m not surprised. Apple’s first party apps have always seemed like afterthoughts that were lower priority than other things. (E.g. relative to what I consider great quality hardware.)
Maps was terrible for several years following the release, and is still not great.
Screen Time, especially the parental controls side of it, is almost unusable.
Find My Friends used to have all sorts of disconnects where it wouldn’t work, though admittedly it seems to have finally gotten better over the past couple years.
These are just some examples I can think of. But this bug in the OP doesn’t surprise me.
This describes iOS in a nutshell.
When I first came to HN it wasn't an issue. Now I have to use my own app for it so the font (and some other things) are workable.
According to my eye doctor the screen time is causing eyesight issues earlier. We're not designed to stare at a bright light 40cm away all day.
May want to look at some eye exercises - or at least something far away.
"In contrast, long-wavelength light is growth-inhibiting and short-wavelength light is growth-promoting in rhesus monkey (57) and tree shrew (58)."
Modern monitors have high amount of shortwave spectrum, blue is unusually shortwave.
FWIW in Firefox (and i guess Chrome and other browsers) you can have per-site zoom. Also addons like Stylus allow you to setup site-specific CSS rules (and HN uses a bunch of classes in elements that use the same visual style by default but can be altered with custom CSS). For example in HN one thing (among others) i have is to use a slightly darker background for every other comment to make it easier to distinguish between comments when scrolling.
If you like, replace ‘designed for’ with ‘suited to’.
Most people who say "designed" here aren't ignorant: they don't care about the distinction and say what's idiomatic.
I still see MacOS as the best choice for my desktop/laptop uses (browser and SSH), but I also have a documents folder that I’ve accumulated over decades. I still use various .txt files in the docs folder as my low tech note taking apps.
I use the Spotlight or Alfred keyboard shortcuts (that also use spotlight index?) for quickly opening the files when needed - and annoyingly my most important file - notes.txt - regularly disappears from the Spotlight index and suggestions. It’s been like that for at least 5 years, probably closer to 10. I’m not even trying anymore, will just open the file from command line with vi as the fallback step.
If it' just "this", Linux is perfectly capable and IMO even superior.
Nobody else offers the same combination of battery life to performance/weight, build quality, keyboard, trackpad, and screen. Of course it's not perfect for everbody and you might have different priorities but I think the MacBook gets most of them right for most people.
Some come close on a few of those points but if you want official linux support your choice is very limited. Perhaps that doesn't matter to you but I don't want to even think about if updating my daily driver is going to result in a broken webcam or flaky wifi or bad power management.
I'm holding out hope for the new snapdragon based laptops. They seem pretty close!
P.S. I've ran Linux/X (plus VMWare VMs with Windows) on my desktop machine with few complaints since the '90s and it was always the laptops that had issues, causing me to switch back to Windows after a couple of weeks of trial & frustration in the 2000s. I got Windows pretty performant & usable though, even attended Mark Russinovich'es Windows Internals class in London back in 2006 or so :-)
Less can be said about your typical Linux experience in in the 2020s where you will still inexplicably find yourself having to mess around trying to get Bluetooth/audio/webcam/sleep working reliabily.
With Spotlight you can never be sure. And to be fair, the Windows equivalent sucks just as much.
Almost anything will do for those?
I'm using Apple Notes and it fails in some random ways after keeping it open for 1-2 weeks: When I try to copy something I select, it copies some random stuff, dragging text won't work, I can check/uncheck todo boxes. Goes away when I restart it.
[0] says that was in macOS Ventura 13.1 and iOS 16.2.
I wouldn’t be surprised if they had to rewrite the text editing engine in Notes, or at least parts of it, to accommodate for this change. And if there’s anything more terrifying than modern Apple rewriting parts of macOS, it’s them doing it for any cloud-based functionality. shudders
Seeing that I’m not the only one, I need to remember to restart notes at the first sign of an issue, rather than trying the action over and over trying to figure out what’s going on.
Ha, I am too used to it being more accessible to file a bug report, having spent most of my career with GNU/Linux (contributing and using since 90s).
A special place in hell exists for such code. No surprise the coder responsible is not keen to visit it.
I took some screenshots and I do not see the misaligned numbers at retina or non-retina resolutions, but I do see the odd bevelled edges on the 8/10/16 "tabs": https://imgur.com/a/PqqkWai
Apple have pretty much given up on making things look correct on non-retina displays, so many things are positioned at what turn out to be half pixel steps. Depending on whether we're talking fonts or shapes things can jump by a pixel or become blurry. I wrote about this here: https://blog.gingerbeardman.com/2024/01/25/running-modern-ma...
There is: https://feedbackassistant.apple.com/
That said, it isn’t very user-friendly, and I find that they don’t seem to pay much attention to it. When they do respond, it tends to be some form of “#wontfix. Please close this.”
That looks like a fairly ugly little bug. I suspect they know about it now, thanks to the HN Bug Reporter. It tends to highlight these types of things.
By not having a public record of bugs Apple conveniently hides the sheer number of them and how many people they affect.
This creates an incentive for the big players to improve their process and proactively catch bugs.
I've seen bugs reported to Apple's bug reporter get fixed in subsequent OS versions, but almost never in updates to the current or previous ones. This is a fundamental flaw with their process that provides a historical track record of them deprioritizing certain bugs. Which is why we should probably pivot away from internal bug reporting services and move towards third party bug trackers.
The AAPL market cap is $3.48175 trillion as I write this.
I've had enough cases where a simple screenshot or log snippet should have been enough to accept the bug report but instead they were closed because I cannot in good conscience attach all the data they want from a Mac that I use for my day to day work.
:shrug:
If you are submitting a build for Mac (in my case, as a Mac Catalyst companion to an iOS app), you can’t reset the build number to 0, after you change the main version.
For example, if you go from 1.0.0 (106) to 1.0.1, you would expect it to be 1.0.1 (0), but it won’t let you submit a Mac build with that build. It must be 1.0.1 (107), even though the iOS build is fine, with 1.0.1 (0).
This forces me to keep updating the build number on both builds (because I sync them). I used to use the build number as an indicator of release status, but this pooches that. Not the end of the world, but annoying.
I first got a “cannot reproduce” response, where they wanted me to submit a sample app (In the original report, I actually sent them a link to my full app source code, as it is an open-source app -most of my work is open-source, and I have a number of repos that contain full source for shipping apps).
I responded, saying I would not, because it would require creating a whole fake app, with fake releases and whatnot, and it wasn’t worth it, as I had already sent them a link to a shipping app, that exhibits the problem, and also, they were quite capable of doing that, a hell of a lot more easily than I could.
I then got a second response, saying something like “Oh, I see. It is a string issue, not a numerical issue. Works as designed. #wontfix. Here’s how to close a bug.”
I gave up, and closed it.
I also submit feature requests and usability issues.
My experience is pure anecdata.
I usually end up closing the reports, after a number of months of them being ignored.
Also another reader mentioned to enter Feedback Assistant in Spotlight. It's the first time I see that app.
However, switch to a different calculator mode (like scientific), and Windows inexplicably removes the pinning feature.
This baffling decision feels so actively user hostile it is deserving of an award for poor design choices.
Linux and Windows use actual pixels for their rendering, and even with anti-aliasing, it looks sharp. If you’re stuck with macOS, aim for 4k at least.
And as far as bad design goes, why are the bit position indicators on the right (0, 32) center-justified underneath their digits, whereas the ones on the left (31, 63) are left-justified?
I remember using fonts of 8pt in an IDE to "squeeze " the potential of the monitor.
This is infuriating and the same for all the big companies (at least Google, Microsoft, Apple); you have a serious issue and simply no way to talk to a representative. The best you can do is post something on Hacker News and hope it somehow gets picked up.
I worked at a company that paid Microsoft a lot to have a 1-day SLA for support. When I contacted them, I got a reply back weeks later saying "hey sorry I missed your email". About two weeks later (which was the time it took to email back and forth), it was clear that I had to insert another ticket and mark the subject as something else (that was not directly related, but apparently the team responsible for that subject was also working on the functionality I found a bug in. There was no way for me to know this since it was something internal to Microsoft). So, I had to go through the whole procedure again.
Once I did that, the reply was "oh yeah, we dropped that functionality but the documentation doesn't mention it. we recommend you use <technology X> for this". Where, of course, technology X did not support the feature I was trying to use.
But it inherits baggage from the limitations of the handheld calculators of the 1970s. Why can't I use the - button to write a negative sign? What does "AC" mean? The scientific calculator is an even worse design. There's a ton of invisible state, like the value stored in memory, or whether you're inside parentheses. The user has to hold the whole sequence of operations in their head, without clicking a single wrong button. Want to repeat a calculation with a different operand? Tough.
Graphing calculators like the TI-84 that let you see and edit your input are so much more usable. Even better are notebook-style interfaces like Mathematica. It's a shame Apple won't pay homage to those designs.
It's using floats on 32bit (...which means only the watchOS currently, I guess?) and doubles elsewhere.
Are there any modern UI frameworks that _don't_ use floats/doubles?
I was gonna guess CSS, but even that has supported sub-pixel precisions on HiDPI displays for a while now.
[1]: https://stackoverflow.com/questions/5709698/html-sub-pixel-b...
I was just talking about this yesterday: somehow TextEdit on Mac has been wrecked. TextEdit, which is essentially a wrapper around NSTextView, was more or less "perfect" 15-20 years ago. Now I experience a bug where the window is blank when I open a document until I click inside a window, and scrolling performance in a long document is atrocious. For example, if I try to scroll backward, from the end of the document, it stutters and can lose my place. This doesn't depend on the document; it happens all the time.
I guess that Apple rewrote everything a few years ago with TextKit 2, and it shows, but not in a good way.
The impression I get from Apple is that Craig Federighi has given engineers license to keep churning out new features and not worry much about bugs, or design, or the user interface. And if something becomes a massive problem, they just pause on features for a couple of weeks, which is like rearranging the deck chairs on the Titanic.
"after I took screenshots and restarted the calculator, the misplacements were gone. I am looking out for it to happen again."
Same on my old MBP as well.
I guess it is a feature.
Just guessing -- it doesn't look accidental to me.
Assuming you last left the calculator in "Programmer" mode the calculator displays the value "4".
Ideally, pressing enter in Spotlight would simply replace the text in the input bar with the result. The equivalent Alt+Space tool in KDE (Plasma Search) performs math this way and it's amazing. I haven't used Quicksilver or Alfred in a decade but I'm sure they do the right thing, too.
Otherwise both Spotlight and Plasma Search are both pretty great. Type something like "14oz to lb" and they both display the result (though Plasma Search displays the exact "0.875 pounds" while Spotlight displays the rounded "0.88 pounds").
Overall I'm mostly disappointed with first-party Apple software. Being one of the richest companies in the world I have higher expectations.
That would have never happened under Jobs's watch.
Antialiased text always looks blurry to me after looking at pixel fonts all the time.
Not necessarily cosmic rays but things like marginal timing can cause errors like this, especially on GPU buses/VRAM that tend to have less protection.
GPGPU and now AI has made accuracy of results more important, but before that, GPUs were regularly ran at the limits and it was assumed that occasional barely-visible artifacts or otherwise computation errors whose results aren't noticeable were acceptable. (Imagine you're playing a 3D game and a few pixels in a frame occasionally have incorrect values, or some shapes are a pixel off --- unless the errors are massive, you're unlikely to notice.)
Is 'Hide Binary" enabled or disabled? If it's enabled, why is it a different colour to the slider that has presumably selected Base-16?
Are the binary digits editable?
Are "ASCII" and "Unicode" mutually exclusive as you'd expect, in which case why are they both the same colour?
The Apple calculator is a frustrating mess to use.
PCalc does scientific, engineering, and A to B calculations for most things.
I wouldn’t go back to Apple’s calculator app even if Apple gave me a credit for the PCalc app.
Nope.
There seems to be more to that custom view than a standard line of text; something about that view is not optimized for low-dpi displays.
