Hacker News Reader

Comment by Tepix | Hacker News Reader

It will be interesting to see how long Rosetta-2 will be supported. macOS 10.15 took away 32bit apps, which was a sizeable chunk so clearly Apple is willing to alienate some of its users in the name of progress and slimlining.

View on HN

A quick look at previous transitions shows...

Classic Mac OS support (both 68k and PPC): OS X 10.0 was released March 2001, dropped with 10.5 released October 2007. So about 6 years, 8 months.

PowerPC Mac OS X app support: First Intel Mac January 2006, support dropped with 10.7 in July 2011, so 5 years 5 months.

32 bit iOS app support: iPhone 5S with 64 bit CPU released October 2013, 32 bit app support removed in iOS 11, September 2017, so 3 years 11 months.

32 bit Intel app support (and Carbon framework): Mac OS 10.5 was first to allow 64 bit GUI apps[1], in October 2007, 32 bit support dropped in 10.15, released October 2019, so 12 years.

I'd probably count this nearer to the first three than the kinda drawn out 32 bit to 64 bit Intel transition, so perhaps about 4 years?

[1] There was limited 64 bit support before in 10.4.

Keep in mind that there is a much larger installed base of Intel machines than any of those other transitions. Apple is unlikely to want to alienate so many people. It will take several years for the percentage of Intel Mac users to fall under 50%.

I see Intel support continuing for closer to 10 years.

I thought the same at first, but the fact that Apple dropped x86 32bit support means a lot of old unmaintained Intel binaries are already out of the equation.

While I don't disagree, Apple users also seem to be the fastest to upgrade to new versions of Apple OSes and tech. They're already like 90%+ upgraded to the new macOS, etc.

> Keep in mind that there is a much larger installed base of Intel machines than any of those other transitions.

The OP asked about Rosetta2 though. How many Intel macs are out there doesn't mean much when fat binaries can support both. This is all about software moving.

The two are connected. The more niche (or custom) software is the more resistant it is to porting. There's going to be a lot of Wintel based stuff which is a quick and dirty port from Windows with few Mac native elements. The years where Mac was Intel compatible made the platform more attractive to people who worked in a mostly Wintel environment. There is almost no incentive for those who ported software to the Mac because it was easy to now port to ARM.

So there are two sides to that: those people are going to hang on to their Intel machines longer and they'll want Intel compatibility when they finally buy ARM machines. This is important to Apple because it can shrink the pool of potential customers.

I was on the last PPC during the ppc->x86. My experience on it was that after a year it was starting to be hard to find new releases containing a ppc build already, and those who did very frequently were poorly tested. Although OSX releases continued to support ppc, at the user level it was pretty clear I was on a dead end. Even on the OSS front I had to rebuild for PPC frequently myself, as more often than not the x86 build was the only one available.

Is that even a serious criticism? Users had about ten years of warning before 32-bit went away. The real impact of that on real users is minimal.

Yes, there were warnings, but what can I as a user do about it? I am impacted by 32 bit going away and I cannot do anything about it. That is why I am running Mojava instead of Catalina. On the bright side, whenever I am going to buy a new Mac, the ARM processor effectively doesn't cause any additional troubles on the compatibility side, and that was probably the reason for Apple to kill 32 bit with Catalina.

But there is something you can do about it. You're already doing it. You can run Mojave.

You want the platform to be weighed down with cruft forever because of a 0.01% use case? Do you want 8-bit support in there too, for the 68000?

68000 Mac apps were never 8 bit, the CPU had a 24-bit address space!

I'm sorry. Thank you for the correction.

There are some non-trivial apps used by real users that are 32-bit only, in particular the pre-subscription versions of Adobe Creative Suite.

And those aren’t the worst examples, as alternatives exist (subscription only and/or more expensive, possibly, but they do exist)

Especially in companies, there may be that ‘essential’ app somebody wrote in two decades ago that nobody has the source code to or, if they have, uses a language dialect no modern compiler supports, requires third party libraries of companies long out of business, etc.

Also, hardware support may be an issue. Expensive hardware can have long lifetimes, and hardware companies tend to be bad at upgrading drivers order hardware they don’t sell anymore.

Those companies should pay the developers of those “essential” apps for the update. I’m tired of companies cheeping out and blaming everyone else when they don’t want to pay devs.

And real users can keep running that old software on real similarly-old computers. I don't think the number of still-functional old Macs is so small that they're going to be impossible to find any time soon.

I can understand being unhappy that Adobe doesn't offer an upgrade path from an obsolete version of CS to similar licensing terms for a modern version that runs on modern computers. But I think that complaint should be directed at Adobe, not Apple.

So end-of-life releases that have been replaced with newer builds? Any application that doesn't survive the progression of technology doesn't deserve to-- software requires maintenance. We expect hardware to get bigger/faster/stronger but are unwilling to hold software to the same standard?

I am wracking my brain but cannot think of a single application that I ran on previous epochs of hardware (whether it be 32bit, PowerPC, Itanium, or anything else) that I am now missing out on.FOSS has really mitigated this a ton too because if there was a good idea from a defunct codebase, inevitably a project would spring up to fill that void.

Professionally, legacy software is the bane of my existence because I am expected to move heaven and earth to make crappy, long outdated and inherently insecure proprietary enterprise apps work on modern networks with modern operating systems and modern security requirements. Let it all burn.

Software that runs perfectly fine, performs just what I need, and that I already paid for, that has no security implications... and you want me to now start paying monthly for the privilege of using the same software. No thanks.

I know software requires maintenance. When it’s cheaper to buy another computer that runs the old software just fine than paying for “updates” and “improvements” and “bug fixes” that do nothing for me, I think people are going to stop paying for such “updates”.

And a massive chunk of Steam library which will never be updated.

> Users had about ten years of warning before 32-bit went away.

Citation needed. I remember 32 bit being deprecated just 12 months before its support removal and there were no signs earlier. I just had to recompile my apps to x64, but some 3rd party tools I use never got an update and my 2013 Macbook will stay on Mojave forever.

64-bit apps debuted something like 12 years ago on the Mac. The writing has thus been on the wall for circa a decade, as I said. You figured 32-bit would be supported for what, 20 years? 30?

> The writing has thus been on the wall for circa a decade, as I said.

So do you say that there is writing on the wall for removing 32-bit support in Windows and Linux too? I certainly don't. If Apple deprecated 32-bit while introducing x64 then the writing would be on the wall but they kept supporting and updating 32-bit.

> You figured 32-bit would be supported for what, 20 years? 30?

At least one hardware generation lifetime from moment of deprecation (5 years would be bad, 10 would be great). There were 32-bit AU plugins still being sold in 2018 and some of them never got an update. KORG never updated some of their synth control apps for synthesisers that are still on sale - they just updated docs and wrote that macOS is not supported anymore.

I saw someone post on a non work slack about their steam library, where they had a bunch of games that previously ran on mac, being decimated, and they meant decimated literally by catalina. 90% of their games were 32 bit.

The good news for steam users is that you don’t have to purchase the game separately for each platform. The decimated users can build their own Windows gaming PC and easily regain access to all of their old games. I doubt they had any Mac-exclusive games that also happened to be 32-bit only. Do such games even exist? Maybe.

Funnily enough that is the inverse of the historical meaning of decimated (to destroy 10% of something).

Good to know, thanks.

Sounds like the writing is on the wall for 32bit linux but you still have some time.

https://ubuntu.com/blog/statement-on-32-bit-i386-packages-fo...

At WWDC they said they're starting a transition period that'll last 2 years.

So I expect them to drop support in 1.5 years

No way. Rosetta-2 was introduced as part of the transition, why would Apple sunset it so quickly?

This is an in accurate reading of the word “transition“.

Transition of their hardware, so in two years all their devices will be Mac silicon. Probably. I think the Mac Pro may not migrate, given that it uses server grade hardware and Mac's silicon is not there yet (I think / assume / I'm probably wrong)

Reply on news.ycombinator.com