Algorithms + Data Structures = Programs - Episode 40: Star Trek vs PowerPC (with Sean Parent)

Episode Date: August 27, 2021

In this episode, Sean Parent tells the story of the battle between Project Star Trek and PowerPC.About the Guest:Sean Parent is a principal scientist and software architect for Adobe Photoshop. Sean has been at Adobe since 1993 when he joined as a senior engineer working on Photoshop and later managed Adobe’s Software Technology Lab. In 2009 Sean spent a year at Google working on Chrome OS before returning to Adobe. From 1988 through 1993 Sean worked at Apple, where he was part of the system software team that developed the technologies allowing Apple’s successful transition to PowerPC.Show NotesDate Recorded: 2021-07-28Date Released: 2021-08-27Blue MeaniesTaligentMotorola 68000PentiumRISC (Reduced instruction set computer)NovellApple Star Trek projectMacHackPowerPCJohn SculleyGARY DAVIDIAN AND HIS 68000 EMULATORRhapsody (operating system)Epic Games v. AppleIntro Song InfoMiss You by Sarah Jansen https://soundcloud.com/sarahjansenmusicCreative Commons — Attribution 3.0 Unported — CC BY 3.0Free Download / Stream: http://bit.ly/l-miss-youMusic promoted by Audio Library https://youtu.be/iYYxnasvfx8

Roger gave this amazing little monologue where he said, hey, he said he's sick and tired of all these engineers complaining about people leaving. And they had to understand that Apple gets thousands of resumes every single day. And he said, if you want to see the impact you're going to leave leaving the company, he said, put your hand in a bucket of water and pull it back out and look at the hole you left. Welcome to ADSP, the podcast episode 40 recorded on July 28th, 2021. My name is Connor and today with my co-host Bryce, we interview Sean Parent, and he tells us the story of the battle between Project Star Trek and PowerPC. So a little background here. At Apple, when I joined Apple, I was working on QuickTry GX, and then I switched to a team which was known as the Blue Meanies, which were, the purpose of the Blue Meanies is they were the system integration architects. This was around Mac OS System 7, during Mac OS System 7 development.

So the Blue Meanies basically went around and figured out what all the various teams at Adobe were doing and tried to find areas where teams should be talking together more and coordinating more and just kind of making connections there. And they would also help with, you know, difficult to solve engineering problems that came about in kind of integrating components. Is this like meanies as in like M-E-A-n-i-e-s like someone who is mean to someone yeah you know the references to the uh the the the the blue meanies in the uh uh beatles uh uh what is it yellow submarine movie so oh okay so it's a it's a cultural reference that i'm out of the loop on yeah so so just search blue meanies and you'll see a little cartoon figure of what a blue meanie looks

like do you know what this is bryce uh no i think we are uh i think we're wrong generation for us to understand this connor yeah i do know who the beat are okay the disappointments in the back I guess yes that's a good thing yes so yeah so look up what a blue media is and they're cute little cartoon characters but you know Apple at the time was divided into pink and blue where pink eventually spun off to be Taligent. It was kind of future macOS thing, which never shipped. It ended up in a partnership with IBM as a company called Taligent. And blue was kind of traditional macOS.

And so the blue meanies, that's where the blue comes in. And so, yeah, so I was working on BlueMeanies and then part of what I was doing on BlueMeanies was I got tasked to start tracking a couple of projects that were shaping up the context here was Macs were sitting on Motorola 68000000 processors, and Intel had the first Pentiums coming soon. And even though the first Pentiums were a flaming hot, massive energy-cons consuming device, they were faster than anything that Motorola had in their pipeline in the 68k line. And Motorola was struggling with process manufacturing to build faster 68k'ss to kind of move to standard, to smaller manufacturing processes.

And so the answer seemed to be either to switch to switch max to Intel or switch max to risk architecture. And so somebody at Apple, a guy named Gary Davidian, wrote an emulator for the 88110, which was Motorola's risk processor. And he wrote a 68K emulator and showed that he was able to run Macintosh in kind of a fully emulated environment on a 88 110 with reasonable performance right right it it

wasn't it wasn't fast but but it didn't didn't suck and Apple established a partnership with Novell, which was a non-Microsoft company that had a competitor to DOS at the time. But, you know, Windows was kind of fairly new at this point. Novell didn't have an answer to Windows. And so Apple also entered into a partnership with Novell to investigate moving Macintosh onto Intel processors using Novell's underlying kernel.

And Apple kind of providing the GUI interface on top of that. And that was known as the Star Trek project. You can look these things up online now. They're fairly well documented. Well, the Star Trek project was this very high-profile partnership inside of Apple. And the team had, like, a tremendous amount of resources, and it was a large team, and they were, you know, this pirate team, you know, semi-super-secret.

They had their own floor of a building and, you know, pool tables and perks and lots of the best engineering talent, several of the Blue Meanies, kind of slid over into the Star Trek project. And I was monitoring the 88110 and the RISC project. And out of that, we created a little group to see if we could run with this idea of having some native code and some emulated code and bring up a machine in that environment using a RISC processor.

And so it started as just a bunch of committee meetings exploring it, and eventually we got to the point where things were coming pretty far along, and it was starting to look feasible. of all things. I ended up staying up all night kind of scribbling in a notebook of how mixed mode could actually work across for the seamless integration between 68K and PowerPC code, and neither code had to know about the other, and you could get existing Mac software to run even though everything in an existing 68K Mac was running in supervisory space, so you could have interrupt handlers and all of these things, you know, executing in user applications or system extensions or things of that, how I could make all of that work.

And, you know, even though I kind of have the aha moment, you know, this was not just me. There were a whole bunch of people who worked on this and were all listed on a patent for mixed mode. And my ideas were just putting together pieces that other people had. So in any case, I go back to Apple and I start like, oh, I got this figured out. This is how it all works and scribble it on the diagram

and everybody's like, okay. And from there, we kind of moved out of committees and formed a team, the PowerPC team. And one of the people who we brought in was an individual, Phil Koch, who had been a professor at Dartmouth. And he joined Apple to be engineering manager on this team.

And we ended up with several first interns and then later hires who had been his students at Dartmouth who kind of came in. But we were a very small team. It was just a handful of us literally working on this. And we had a small cluster of offices mixed up with a bunch of other teams and kind of, you know, no support. It was a secret Skunk Works project, but this was not like Star Trek's project.

We were not, like, well-funded. We didn't have perks. We were just kind of struggling along. And eventually, well, Scully came in at one point and said, oh, okay, so I cut this deal with IBM, and we're going with power architecture, so this will be a new processor, PowerPC, and honestly, we had been evaluating RISC chips, and like power architecture was, you know, like not even on the first page of candidates. But to Scully's credit, Scully was like, yeah, I looked at your list.

There aren't that many companies that can produce chips in the volume that we need. And, you know, IBM is one of them. And so it became this very complicated partnership with IBM and with Motorola, which is where PowerPC came from. So there was a three-way partnership around the processor chips. There was an extended partnership with IBM around Taligent, which was Apple's pink operating system, which would maybe be a future operating system running on PowerPC. And there was an effort to bring up Macintosh software on top of AIX, and that actually shipped at some point. AIX is IBM's flavor of Unix. And then there was our project which was secret even to IBM because IBM thought that when they were shipping these things that it would

be their AIX kernel sitting underneath Macintoshes that were PowerPC, and Apple wanted a kind of side bet on maybe we could do this just with all Apple software. And so that was the team that I was working on. And so we got to the point where we were, you know, first we prototyped everything. We, like, prototyped all of Mixed Mode running 68K to 68K, which sounds kind of weird, but it actually worked as a proof of concept, and kind of prototyped the architecture, and then started building things on power machines, which are not quite power machines, to get things kind of limping along. And then eventually we had PowerPC silicon and we built boards that you can plug into 68k max Quadrate hundreds and you know went through like how do you you know use logic analyzers to get to the point where you can boot these devices and

then we were able to get them up inside of an emulator booting on them based off of Gary Davidian's work, but now ported to PowerPC. And then finally we get to show mixed mode. And the first thing we moved over to mixed mode was there was an old Mac toolbox routine called setRect, which all it did was it took, you know, four coordinates of 16-bit integers, top left, bottom right, and returned a struct of those four integers that were passed as arguments. So an almost do-nothing routine, it's basically just copying four integers from the stack into a struct on the heap, or I'm sorry, into a struct on the stack. So it's just copying four integers. But the interesting thing about set-rect was it got called all over the place. It got called at

interrupt time, it got called from patches to the operating system, it got, you know, called from signal handlers, it was everywhere. And so we were able to take that routine native and show that, you know, native code could call it, native code could call it even if it had been patched with a 68k code. 68k code could call it regardless of the execution context and it would execute correctly and return correctly and keep the state of the machine and nothing fell apart, right? Everything still worked. And so that was our proof of concept. And after we got that running, the next day, we got it running kind of late in the evening. And the next morning, Phil Koch walked into my office and closed the door and said, I owe you an apology.

And I said, what for? And he said, this whole Mixmo thing thing I didn't think it was going to work and I said well you should have said something he's like oh no you explained it I would explain it to anybody who wanted to hear

walk up to the whiteboard and walk through the logic of how it worked through all the transitions he said I saw what you were doing. I couldn't poke a hole in it. He said, just my gut said this isn't going to work. And I said, well, you know, you don't owe me an apology for that, right? Thanks for letting us run it out.

And he said, no, no. He said, you don't understand. My reports to my bosses and the executives have been, this team's doing this, you know, wacky mix mode thing. I don't think it's going to work. He said, so you want to know why this team doesn't have the funding that Star Trek has? Why other teams aren't excited about porting their code to PowerPC, why this team has like zero support, why it's just a handful of people.

He said, because you're the side bet that I didn't think was going to work. Okay. So, you know, in reality, in retrospect, I think that keeping the team small, keeping the team focused, keeping all the outside distractions from pouring in was critical to actually getting this to ship. I give Scott Boyd, who was technical lead on this project for most of the duration of the project, a lot of credit for getting this thing shipped. And he coined the acronym NNTS, which was not necessary to ship. And kind of any time it was like, oh, I think we should be doing X, he would respond with an email of just NNTS, meaning like good idea, bad idea, I don't care, it's not necessary to ship.

So very much trying to keep the team focused. So that all happened and we shipped PowerPC and PowerPC launched and it was a big success. And PowerPC was known as V0 internally because there was a plan. Apple had a roadmap that was kind of V1, V2, V3 of what they were going to do with their new kernel work. And PowerPC, we scaled it back so much to get it out the door that we called it V0. And the team pitched a follow-on which was called Alberto, which is VO5, if anybody knows the hair product Alberto VO5. But yeah, so it was a half step between V0 and V1. But management at the time said no, we have to drive towards V1, and so that project

got canceled, and Roger Heinen, who was leading system software at the time, Apple was losing a lot of talent at this time, Apple was kind of bleeding cash and was scrambling, even with PowerPC's success. You know, Apple was looking to enable clones, and this is, like, soon after, kind of this unwinding is when Steve Jobs kind of did the reverse takeover thing and came back to Apple. So Apple had been bleeding talent and there was a meeting with Roger Heinen, who was head of system software, and the engineers were raising the concerns in the audience going,

you know, what are you going to do to keep talent here and keep people from leaving? leaving and Roger gave this amazing little monologue where he said he's sick and tired of all these engineers complaining about people leaving and they had to understand that Apple gets thousands of resumes every single day and he said if you want to see the impact you're going to leave leaving the company, he said, put your hand in a bucket of water and pull it back out and look at the hole you left. You know, basically meaning that like no engineer mattered. And one of the senior colonel engineers like walked up to the front of the little amphitheater we were in and handed Roger his badge and just kept walking. Morale was clearly high. Morale was clearly high.

We also had, in one of the hallways, we had tombstones, whereas each engineer left, we kind of drew a tombstone and put rip in their name on it so yeah morale was really high um when you said amazing speech i thought it was going to be this like uplifting you know it was like it was a lesson in like how not to manage is what this was. So yeah, so after the Alberto project got shut down, pretty much everybody who was on the PowerPC team left in about six months. And that's when I went to Adobe and joined the Photoshop team. So the end result was a good thing for me. And so now fast forward, you know, we kind of talked about I'm at Adobe and I'm Macintosh tech lead on Photoshop.

And I come up with the Cyan proposal now that Steve Jobs is back at Apple and they're doing a transition that was going to be Rhapsody. And then there's the Cyan proposal, and then we find out about OS X when I get a call from Steve Jobs and get to go over to Apple and learn about OS X. And so all of that happens, and then I get a phone call from Phil Koch at Apple, and Phil says, hey, I'm leaving Apple, and I just wanted to come and have lunch with you. And I'm like, oh, that's great. And so he comes over, and we're sitting at lunch, and yeah, he said, I'm going to retire and leave Apple. He said, but I wanted to let you know something that you're probably not aware of.

And he said, when this Cyan proposal happened, he said, this Apple executive who I won't name, but he's the person who had led to us cutting off negotiations around Rhapsody, who had written emails to John Warnock about both Greg Gilley and myself being impossible to work with. He said, yeah. So he said that at one point during that whole thing, the executive walked into my office and put a paper on my desk and said, I need you to write up why this won't work. And he said, I looked at the cover, he said, and it said, you know, the same proposal, Sean Parent. He said, and I handed it back and said it'll work and he said no he said the executive was like

no that it wasn't a question of will it or won't it work I need a report from you that says it won't work and and Phil said I flat out refused he said I won't do that he said I'll tell you right now.

He said, I can read the paper and tell you why it will work, or you can just take my word for it. It's going to work. And handed it back. And he said, so. So he said, I, you know, he said, after what happened with PowerPC, he said, I thought you deserved to know that. So I thought that was an interesting side story. So that's what the story that Tony was, was referencing.

So the question is, is that the genesis of what ended up, you know, the start of the, what was it, two-year path to Steve Jobs then calling you back and like, hey, we'd like you to come over to Apple. Yeah, you know, if I had to guess, I think the timeframe there was probably just before the meeting where I blew up at this executive, right in between the meeting with Greg Gilley and the meeting that I had. Because I mentioned before that that meeting was supposed to be with Apple explaining the cost of them doing the Cyan proposal and kind of walking through the business aspects of this whole thing. And Apple just came in and said, you know, it's impossible and we're not doing it. And that's what led me to get into the argument about saying, you know, you're full of crap. And, you know, if, if, you know,

if you want to build, you can hire me and I'll go build it. Right. Right. It's, it's, it's a hundred percent doable. And so, so I think like before that meeting, uh, this executive had gone to Phil looking for ammunition and, uh, do you, I, I, I doubt that you'd be able to share it, but do you still have a copy of the report? You know unfortunately not

the Cyan proposal I would love to have it just kind of as a historical artifact and I've kind of asked some friends at Apple to see if anybody kept it and yeah I keep a lot of old email, but like not anything that far back. And I kind of dug through and after, you know, leaving the company and coming back, I could not find the Cyan proposal.

So, yeah, if any of your listeners have an old copy lying about someplace. What year would it have been? So let's see this would have been i'm guessing 95 so if you worked at apple in 1995 on their operating system yeah you might have had a copy if you were working at uh at adobe too or if you're at adobe um I should ask I don't know that I've ever asked Russell Williams he worked on what was going to be the V1 kernel

at Apple and then I hired him to work on Photoshop and he's been at Adobe working on Photoshop ever since and one of the key people on the Photoshop team photoshop and he's been at adobe working on photoshop ever since and uh you know one of the key people on the photoshop team do you remember how long was the proposal was it just like a white paper was just a couple pages or was it like it was just a couple pages it was like two pages it was a short would you have been able to share it i feel like it's been

like a couple decades that that'd be yeah yeah you know i don't even think it was you know particularly interesting right it was it was uh uh uh you know basically it was an outline of what we had planned to do with uh with alberto and with v1 and so and what apple finally did with what was known as the blue box on Mac OS X, but it was basically saying that, yeah, you could have an environment where you could run existing apps that were written against the existing toolbox. And then you could also have an environment where those apps could call out to new Cocoa-based functionality, and you can mix and match the two. Technically, I don't think it was all that interesting. I don't think it was it was all that that interesting I don't think

it was like you know interesting and kind of the mixed mode space and I think anybody who seriously looked at the problem was like oh sure of course you could do that you know it was just Apple's reluctance Apple wanted kind of this clean move to to next step and to jettison the Mac baggage. And so it was much more the combined strength of, you know, Adobe and Microsoft at the time that basically refused to make that move that convinced them to go the other direction. And, you know, I've heard mixed reports from people in different parts of Apple as to how much of an impact the Cyan proposal had, anywhere from none, you know, it would have happened anyways, to it wouldn't have happened because Apple was so dead set against it and there wasn't any clear path in doing it to succeed, unless somebody wrote down and committed to say, look, we'll move Photoshop if you do this, right?

And so that was the real impetus. And, you know, we got a lot of flack from Apple then, because, like I said, you know, we found out about this, and then two weeks later we gave a demo on stage, but it hit at a really bad time in the Photoshop development cycle because you know we were kind of running 18 month product cycles and we were just finishing up a release and so we didn't really have time in finishing up that release to do the the Mac OS X support in that. And so we ran in the blue box, which had a bit of a performance penalty, right? Basically, it was an environment that let you run

old Mac apps that hadn't been ported yet for almost two years, right? It was almost two years until we shipped a quote-unquote native Photoshop. And, you know, since both machines were, you know, both the blue box and outside the blue box were, you know, power PC processors, it wasn't like we were running some big performance penalty but we didn't have the the new kind of lickable looking ui and and the app look dated and yeah i remember i bet the one time in my life when i used uh max was around that era right around the switch

when yeah yeah yeah yep yep so you know i somewhat think that uh uh that time period and the amount of pressure that adobe and microsoft could exert right to put it in perspective you know at there was a period there where Photoshop accounted for more than 10% of the total software revenue on Mac. Not Adobe's, just in general. Yeah, not of Adobe's general. Of all developers that were developing Mac software, Photoshop was taking a 10% cut of that pie. But has that changed? Because I know back during that era and even for like 10 years after that,

Macs were the platform that you'd go to if you were doing that sort of creative work. But it seems like it's changed a bit in recent years. Yeah, yeah. Well, yeah, I still think Macs tend to be more prevalent in kind of the creative

industry, but that segment of the market isn't as important for Apple anymore. Even Macs aren't as important for Apple anymore, right? I mean, they're completely dwarfed by iOS sales at this point yeah yeah i saw i saw this funny graphic the other day that um like the the apple's total business for airpods is like larger than the than the business of like a number of notable tech companies. Yep. Including, I think, both NVIDIA and Adobe, I'm sure.

Oh, yeah. Yeah. And, you know, I somewhat think the App Store rules are structured around trying to make sure that you don't have any companies that dominate. Interesting. Right. Why is that something that Apple wouldn't want? Yeah.

Because a company that dominates can exert control on Apple. Right, right. They don't want to have to deal with the Sean parents of the world. They don't want to have to deal with me. That's the whole thing. Interesting. Apple doesn't want to have to deal with me that's the whole thing it's interesting apple doesn't want to deal with sean that's why you guys all have to live with the app store you can put that in the podcast that makes a lot of well that's sort of the what was the twitter beef or whatever some company was it that they came out that they were breaking the rules and starting this feud because it was epic yeah um yeah yeah because i guess yeah as soon as that

happens then like you said they can sort of start a conversation or force their hand and be like we're not paying this 30 fee yeah is there still like a loss lawsuits happening around that i heard that's basically they they said that they weren't going to pay now the epic versus apple lawsuit and the discovery the discoveries in that lawsuit has just been hilarious the the the documents that were that were revealed during discovery have just been uh very hilarious because they've happened to incidentally reveal a bunch of like secret projects and plans that other companies were undertaking um yeah so you're gonna be there's gonna be lawsuits branching off of this lawsuit probably not but uh but yeah it's it's interesting i've been i've been reading the news about it a little bit

and i follow tim sweeney on the on twitter yeah oh yeah he said some unkind things about c++ a couple months ago uh because i remember on cpp cast they were like oh maybe we should get tim sweeney on and then like fast forward three weeks i see tim sweeneyey saying that C++ is, I don't know. I can't remember what he said, but it wasn't nice. What do you think will happen with this sort of stuff, Sean? Do you think that the app developers like Epic will eventually prevail in regaining some control of the ecosystem? Well, I think the interesting thing is now Microsoft is in a position where they're no longer dominant and they're the hungry player with still significant market share. You know, as far as desktop computers go, you know, they have a problem that they never made a successful transition to the mobile space but if you look they announced you know with Windows 11 very reasonable terms for their App Store which is basically

you know you put things in the App Store and if they handle the financial transaction then they take a cut which is like a 5% cut which is you know more in line with what you would expect for kind of a little co-marketing and a and you know a credit card fee basically, for running the financial transaction for you. And they've even said if they don't handle the financial transaction, then they won't take their cut. And so I think that's going to change the dynamics a bit. You know, Apple is so big at this point, I don't know how much pressure that puts on them.

But I think it will make a... I think it will certainly be good for the Microsoft ecosystem and for kind of expanding the Microsoft store offerings. And there is something nice about having kind of a centralized store and knowing that your software is up to date and that at least somebody is looking at it to make sure it's certified, right? You know, you feel confident like buying an Adobe product

or a major product over the web, but a lot of times if it's a little utility or something, I'll see you be like, do I want to download this from this sketchy looking website that hasn't been updated since the 90s? Or maybe there's something that will work that's in the store where I at least know, you know, there's a little bit of a gating factor into what I'm installing and running on my machine. So, so, yeah, so, so, so I think, you know, Microsoft might successfully start to change the dynamics and I think there's an opportunity for Google there if Google wants to more differentiate the Android space

from the iOS space to kind of follow suit with what Microsoft is doing. And in the case of Google, I think that's a reasonable thing because I'm sure they make a lot of money, but they don't make Apple-sized money out of the Android App Store. But they, you know, and most of their money generating on the platform is advertising revenue. That's their main business. And so I think that they might be able to improve their platform to net benefit for them by actually reducing their cost structure. Whether or not they'll do it, I have no insight.

100% pure conjecture here. But that, again, might start to put competitive pressure on Apple to cave. Or somebody like Epic might just win a lawsuit in this space and open it up. Or, you know, the right to repair folks might win some lawsuits and establish a better right to repair. It's related. Certainly an interesting time for these legal issues in tech because there's a bunch of these cases coming to a head right now. Yeah. There's going to be another massive one, the Blizzard stuff that's coming out. It's pretty now. Yeah. There's going to be another massive one.

The Blizzard stuff that's coming out. It's pretty awful. Blizzard does not sound like a nice place to work. It does not sound like a nice place to work. Yep. Send your resumes to Adobe. We're a nice place to work.

Yeah, let's think of the software technology lab as a restart. It's an exciting time we're a nice place yeah let's think of the the st the software technology lab is uh has restarted and that's gonna be an exciting and it's an exciting time to to go go work at adobe i bet it is it is and uh uh yeah and i think you know adobe if you look you know consistently ranks in the in the in the list for for one if not the top places to work. And I think kind of the way we've been managing the pandemic and now kind of getting people starting to migrate back to work

and the amount of flexibility the company has shown in that space, I think has been very good and enlightening. And the company has been listening to their employees and trying to be very flexible. Not insisting that everybody go back to the office like Apple and Google. The company has been listening to their employees and trying to be very flexible. Not insisting that everybody go back to the office like Apple and Google. Yes, not insisting that everybody goes back to the office. The company has basically come out and said,

we're going to have flexible time as the default for people. I didn't know that Apple, I heard that Facebook and Twitter both decided like a year ago, they were like, oh, this seems to work fine. Feel free to stay forever working remotely. But Apple and Google both are mandating people have to come back? At least somewhat, yeah. Yeah, at least somewhat.

They're coming up with different terms. I haven't been following all the news in and out, but I was reading something this morning where basically I think the gist of it was that Apple's policy was either you register as a remote worker, you know, or you come into the office on a full time. And if you're a remote worker, it means you're not in the office on a regular basis.

And that comes with, you know, ramifications for pay and benefits and things of that nature. Some of that is unavoidable, right? Some of it is just kind of baked into tax laws and employment laws and things like that. But Adobe is basically dividing it into you can be a remote worker or you can be a flex worker. And a flex worker is everything from in the office all the time to just in the office on some regularly scheduled pattern.

Thanks for listening. We hope you enjoyed and have a great day.