Anne O. Nymous (not.delete@this.real.address) on June 24, 2020 4:41 am wrote:
> hobold (hobold.delete@this.vectorizer.org) on June 24, 2020 3:33 am wrote:
> > Maynard Handley (name99.delete@this.name99.org) on June 23, 2020 9:14 am wrote:
> > > hobold (hobold.delete@this.vectorizer.org) on June 23, 2020 1:07 am wrote:
> > > > Maynard Handley (name99.delete@this.name99.org) on June 22, 2020 11:57 pm wrote:
> > > >
> > > > > I've posted the wikipedia PPC->Intel transition page here like 10 times. I've explained
> > > > > many more times than that how this transition will follow that exact same script. And
> > > > > yet people refuse to listen up till the minute that the announcements occur.
> > > > >
> > > > Apple's history of transitions, both m68k -> PowerPC, and PowerPC -> x86, did follow the script
> > > >
> > > > 1. have a complicated and unreliable translator that works good
> > > > enough for stage presentations and very simple programs
> > > >
> > > > 2. quickly abandon the translation layer before it ever learns to deal with the hard cases
> > > >
> > > > Been there, observed that. Twice.
> > > >
> > > > You know how the saying goes: "Any sufficiently advanced
> > > > technology is indistinguishable from a rigged demo."
> > >
> > > So what is your complaint?
> >
> > My complaint is that Apple is a marketing driven company, not a technology driven company.
> > Apple is promoting the image of being a technology leader. But that is merely propaganda
> > (or call it advertizing if the former term is too laden with war connotations).
> >
> > When Apple transitions ISAs, they punish those developers who had put extra effort into
> > tuning their software. Rosetta 1 didn't support AltiVec, Rosetta 2 doesn't support AVX.
> > Customers who invested in such software are getting punished as collateral damage.
> >
> > When Apple transitions ISAs, that happens because they
> > had bet on the wrong technology, or because they were
> > unable to make it succeed, despite all the marketing. When Apple transitions ISAs, that is the correction
> > of an earlier mistake. In the "best" case, an Apple ISA transition is an act of planned obsolescence.
> >
> > Rosetta 2 is stage magic. My complaint is that we, knowledgeable computer
> > experts, are falling for the same trick again. And again. And again.
> >
> >
> > The worst part is that Apple is not even transitioning. In the video keynote
> > https://www.youtube.com/watch?v=GEZhD3J89ZE
> > at 1h46m00s, Tim Cook first says the transition will take two years. Immediately afterwards he
> > says new Intel based hardware is in the pipeline and will be supported for years to come.
> >
> > If a transition doesn't end, is it really a transition? It is all marketing. ARM Macs get the promise of a
> > bright future, so that enough customers actually buy them.
> > Intel Macs get the promise of comfortable retirement,
> > so that enough customers actually continue buying them. It is all marketing. We all fall for it.
> >
> > Third party software vendors having to do double the testing? Apple doesn't care. Third party Mac software
> > vendors effectively getting turned into unpaid Apple employees, just like with the iOS tightly controlled
> > app store? Oh, Apple very much does care, yes! And we will all cheer, because it's Apple.
> >
> > On the OS side, Apple is repeating Microsoft's mistake of forcing phone GUI metaphors
> > onto the desktop. Only this time, we will all cheer, because it's Apple.
> >
> >
> > Way back when Apple transitioned from PowerPC to Intel, I jokingly told some friends that
> > Intel's days are numbered, because Apple's processor makers all die sooner rather than
> > later. Turns out Apple was indeed prepared for that particular contingency. :-)
> >
> > But I am not a Wizard of the Dark Arts, so that curse is
> > a mere joke. And Apple will live, despite being their
> > own processor maker now. And we will all cheer Apple on. Even though we all should know better by now.
>
> I mostly agree with the core of your argument, but I evaluate things a bit differently.
>
> I for one consider being willing to change ISA/platform not a sign of weakness or a correction
> of an earlier error, but simply as good management, nobodies crystal ball actually works that well,
> and what might have been even an absolute optimal strategy/decision in the past might turn out
> to be a dead end in the near future. So flexibility and change per se are not bad things.
>
> But in this list even Maynard, who I classify as emotionally biased for Apple deference, argued
> that rosetta 2 is not intended to solve the hard ISA-emulation problem for good, but really just
> a helpful tool to allow to aintain the existing user base and give them peace of mind that they
> will not be forced to jettison all old code real soon now. (It being Apple, we all can assume that
> "not real soon now" does not equate "next to never", but that is the art of marketing).
>

I think a bigger point is that it's difficult to see what customers gain, ordinary or professional. Apple, if forces the issue, makes its ecosystem and customers go through a considerable disruption in exchange for nothing tangible other than higher in-house BOM content and even stronger gross margins for itself. It is difficult to see how exactly will performance or any 'user experience' be superior just because it's not a merchant but a captive design. Theoretical marketing arguments are easy to make. On the other hand of course, Apple customers are not exactly the paragon of self-interested agents, god bless their hearts. The previous ISA transition was, at least, beneficial to customers, largely because Apple waited so long a considerable performance gap opened up.