Revolution in The Valley [Paperback]
Page 20
The Macintosh software was finally coming together as the fall of 1983 wore on. The ROMs containing most of the system software were finished and were more or less holding up. Larry Kenyon devised a clever technique for fixing problems in the ROM by patching the nearest system trap to the problem. The patch code looked back on the stack for the ROM address of the caller, which allowed us to fix problems with tiny, surgical incisions, instead of replacing large chunks of the ROM with precious RAM, as we had originally envisioned.
MacPaint was already stable enough to ship by the middle of November 1983, even though it was always skirting on the edge of running out of memory if we changed something in the system. MacWrite still had lots of bugs, but its core functionality was ready as long as you weren’t pushing memory limits. The only vital application that could delay our target date in mid-January was the Finder, which was the shell application responsible for launching other applications and managing files.
Bruce Horn was the only programmer working on the Finder, and he was bogged down with a variety of problems, especially involving fixing file copying in low memory conditions. The Finder was built on top of Bruce’s Resource Manager using features that were barely finished before the ROM freeze. It probably should have been at least a two-man project, but Bruce was a brilliant, passionate, independent perfectionist who insisted he could get it done in time on his own.
Jerome Coonen, the Macintosh software manager, was afraid Bruce wasn’t going to make it and decided to assign someone else to help. Given Bruce’s perfectionism and temperament, it was going to be hard for someone else to dive in, especially this late in the game. Luckily, Steve Capps was the perfect man for the job.
Stockily built, with long brown hair and a bushy beard, Capps was an extraordinarily talented, creative, and prolific programmer. He had transferred over to the Mac team from the Lisa printing group in January 1983, giving us a new burst of energy just as we entered our sprint to the finish. Capps was one of the few people who Bruce respected enough to really listen to. And best of all, he possessed a cheerful, easygoing disposition that was the perfect complement to Bruce’s high-strung intensity.
To minimize distractions, which were mounting as Apple’s marketing machine kicked into high gear for the launch, Jerome arranged for Bruce and Capps to move into a small office a few blocks away, on Bubb Road. Capps dove in quickly and took over a bunch of tasks from Bruce. It wasn’t always easy, but by the end of December it looked as though the Finder was getting more or less back on track.
Everyone dressed up for Steve Capps day.
Jerome thought of a unique way for the software team to show our appreciation for Capps’ heroic effort. Like many hackers, Capps dressed idiosyncratically, almost always wearing a long-sleeved, white dress shirt with cut-off denim shorts, white socks, and a distinctive type of checkered sneakers called Vans. In fact, Capps had just given everybody on the team a pair of Vans as Christmas presents. Jerome had the idea to pay homage to Capps by declaring the next day “Steve Capps Day” and cooked up a scheme in which we’d all come in to work dressed exactly like Steve.
I had to go to Macy’s to buy myself a white dress shirt, but I managed to come up with a reasonable facsimile of Capps’ attire. It was hilarious to see everyone as we gathered in our “fishbowl” office in Bandley 3 the next morning, all dressed in our white shirts, denim shorts, and Vans. Even the French and German translators who were visiting us for the month joined in. The funniest sight, though, had to have been Patti Kenyon, who was over eight months pregnant at the time. Her extra large white shirt added just the right comic touch.
Once we had all assembled, Jerome went to get Capps and Bruce from their enclave, telling them there was an important meeting that required their attendance. We could hardly hold back the laughter as an unsuspecting Capps walked in. We all cracked up and gave him a round of applause when he realized what was going on.
A Mac for Mick
January 1984
We present a Mac to Mick Jagger
The last weeks before the Macintosh unveiling on January 24th were extremely hectic. The software still wasn’t finished, and it wasn’t clear if there was enough time left to get it into adequate shape. Meanwhile, the Apple PR machine was revved up to full speed, so there were also plenty of unusual diversions, such as being interviewed and photographed for the national press.
The absolute deadline for finishing the software was 6 A.M. on Monday, January 16th, eight days before the introduction. When I came into work on Friday, January 13th, I knew I would probably stay there all weekend, along with the rest of the team, working as hard as possible to shake out the remaining bugs before Monday. Steve Jobs, Mike Murray, Bob Belleville, and others were in New York City doing a press tour, so I thought we would be relatively free of distractions and able to focus on bug fixing.
Mick Jagger performing on stage at Byrne Arena during Rolling Stones concert.
I came into work later than usual, around noon, since I had been at Apple until 3 A.M. the previous evening, and I wanted to get one decent night’s sleep before the final push. As I went to sit down, I noticed that a handwritten note had been placed on my chair. It was from our software librarian, Patti King, who had taken a message from Steve Jobs’ secretary, Lynn Takahashi.
“Andy—Steve J. called—we can deliver a Mac to Mick Jagger tomorrow. You can fly out to meet them by tomorrow noon and bring lots of neat software. If you can come, make arrangements for the trip through Lynn. Steve will call back in a couple of hours, also, he’ll be at the Carlyle Hotel tomorrow.”
A chance to meet Mick Jagger was a once-in-a-lifetime opportunity, but we still had three more days before the deadline. If I flew to New York I would be absent for at least 30 hours, plus I knew I would be relatively useless from all that flying. I called Lynn to have her tell Steve I couldn’t make it. But of course I was dying to know Mick’s reaction, not to mention how all this had come about.
When Bill Atkinson returned from the East Coast on Sunday afternoon, he filled me in. I then got more details from Steve and Mike Murray a bit later. Apparently Steve had gone to a party on Thursday evening and was introduced to Andy Warhol. Andy got really excited about the Macintosh when Steve demoed it, and said, “You must show it to Mick.” Warhol then arranged for Steve and the Apple crew to go to Mick Jagger’s townhouse on Saturday afternoon to present him with a Macintosh.
Steve, Mike, and Bill then showed up at the address they were given and knocked on the door, but there was no response for several minutes. Finally, two huge guys opened it up, but they didn’t seem all that impressed to be face-to-face with the co-founder of Apple Computer and his entourage.
The Apple folks were led upstairs into an elegantly furnished room to wait for Mick. Bill set up the Mac, launched MacPaint, and started to fool around with it. Then, abruptly, Mick Jagger strode into the room, dressed casually in a t-shirt and blue jeans.
Mick was polite, but he didn’t seem to have heard of Apple Computer, Steve Jobs, or the Macintosh. Steve tried to strike up a conversation, but he wasn’t very successful. He later told me Mick couldn’t seem to put together a coherent sentence. “His speech was slurred and very slow,” Steve described it later. “I think he was on drugs. Either that or he’s brain-damaged.” After a few minutes, it was clear Mick had absolutely no interest whatsoever in Apple or the Macintosh, and an awkward silence ensued.
Fortunately, Mick’s 12-year-old daughter Jade had followed Mick into the room, and her eyes lit up when she saw MacPaint. Bill began to teach her how to use it, and pretty soon she was happily mousing away, fascinated by what she could do with MacPaint. Mick drifted off to another room, but the Apple contingent stayed with Jade for another half hour or so, showing off the Macintosh and answering her questions. They ended up leaving the machine behind. There was no way she was going to part with it.
Real Artists Ship
January 1984
The final push to finish the software
With the deadline for finishing the software less than a week away, it seemed obvious that there were still too many bugs to ship it. Late on Friday evening, we convinced ourselves we needed an extra week or two to fix the remaining problems. Steve Jobs was on the East Coast, along with Bob Belleville and Mike Murray, doing press for the introduction, so we arranged for a conference call early Sunday morning to tell him about the slip.
Jerome Coonen, our software manager, spoke for the team, as we sat around the speakerphone. We were exhausted and progress was slow. There were still bugs that we hadn’t gotten to the bottom of yet, and it didn’t seem possible that we could make it in the time remaining. Jerome proposed that we ship “demo” software to the dealers for the introduction, and update all the customers with final software a few weeks later. We thought Jerome was pretty persuasive as we held our breath waiting for Steve to respond.
“No way, there’s no way we’re slipping!” Steve responded. The room let out a collective gasp. “You guys have been working on this stuff for months now, another couple weeks isn’t going to make that much of a difference. You may as well get it over with. Just make it as good as you can. You better get back to work!”
We did manage to wrangle an extra couple of days, by virtue of working the weekend and moving the deadline to 6 A.M. Monday morning, when the Macintosh factory (where the disks would be duplicated) opened, instead of Friday afternoon. We agreed to go home and rest before returning to work on Monday for the final push.
The final week was one of the most intense I ever experienced. Steve wanted Bill Atkinson and me to fly to New York to present a Mac to Mick Jagger, but I decided I needed to stay in Cupertino to help with the bug fixing (see “A Mac for Mick” on page 206). Some of us were pausing work to get photographed for magazines like Newsweek and Rolling Stone, which made others on the team feel terrible that they were being left out. At times, the atmosphere got pretty tense.
When Friday finally rolled around, it was clear that there were still too many bugs in both the Finder and MacWrite. Randy Wigginton brought in a gigantic bag of chocolate-covered espresso beans, which, along with medicinal quantities of caffeinated beverages, helped us forgo sleep entirely for the last couple of days. We started doing release cycles that were only a few hours apart, re-releasing every time we fixed a significant problem.
Top: Rony Sebok, Susan Kare. Middle: Andy Hertzfeld, Bill Atkinson, Owen Densmore. Bottom: Jerome Coonen, Bruce Horn, Steve Capps, Larry Kenyon. Front: Donn Denman, Tracy Kenyon, Patti Kenyon.
When a new release was ready, we would all grab it and start testing again. At one point, around 2 A.M. on Sunday night, I stumbled across a bug in the Clipboard code. I thought I knew what it might be, but I was too tired to deal with it. I tried to pretend I didn’t see the problem, but Steve Capps was watching my expression and knew there was something wrong. He grilled me about the problem and then helped me craft a fix since I was too tired to do it on my own.
Around 4 A.M., we had a release where everything seemed to go wrong—even MacPaint, which was usually rock solid, was crashing. But our final release around 5:30 A.M. seemed to be much better; the worst problems seemed to have receded and we thought we might actually have a decent release candidate.
That last half hour was devoted to testing the final release as much as we could. It looked pretty good, but then someone found a potential showstopper: the system seemed to hang when a blank disk was inserted while running MacWrite and the disk didn’t start formatting as it should. I realized that it was probably hung up waiting for an event, so I reached out and tapped on the space bar, and formatting commenced. Jerome thought the bug was bad enough to hold up the release, but he left to drive it to the factory anyway, figuring they needed to start duplication even if it was just going to be a demo release.
The sun had already risen and the software team finally began to scatter and go home to collapse. We weren’t sure if we were finished or not, and it felt really strange to have nothing to do after working so hard for so long. Instead of going home, Donn Denman and I sat on a couch in the lobby in a daze and watched the accounting and marketing people start trickling into work around 7:30 A.M. or so. We must have been quite a sight; everybody could tell we had been there all night (actually, I hadn’t been home or showered for three days).
Finally, around 8:30 Steve Jobs arrived, and as soon as he saw us he asked if we had made it. I explained the formatting bug to him, and he thought it wasn’t a showstopper, which meant we were actually finished. When I finally got home around 9 A.M., I collapsed on my bed, thinking I’d sleep for at least the next day or two.
Disk Swapper’s Elbow
January 1984
A last-minute bug causes some problems
contributed by Steve Capps
One of the more common afflictions of early Macintosh users was the dreaded “Disk Swapper’s Elbow,” which was caused by a disk-copying operation run amok. Disk swapping was a necessary evil caused by having 400 KB floppy disks, 128 KB of RAM, and a single floppy drive. If a user wanted to make a backup of a disk, she had to eject the source disk, insert a blank one, format it, and then drag the source disk over the new disk. The Finder would then copy data piece by piece with the necessary swapping.
A typical application on a 128K Mac had about 85K of memory available; the rest was used by the system, mostly for the bitmap display. A simple calculation shows that copying a 400K disk should have involved about 5 or 6 swaps. Five disk swaps was barely tolerable. However, as early Finder users will remember, it would occasionally take well over 20 disk swaps.
You’d start a disk copy and hold your breath during the fifth, and hopefully, final swap. If the Mac dutifully disgorged the floppy the sixth time, you’d convince yourself you miscounted, cross your fingers, and hope for the best. By the seventh swap you started cursing because you knew you were trapped and you started wondering about investing in an external drive.
Even though the whole Finder was only 46K of code and had about 10K of overhead, the remaining 30K of memory space was too small for practical copying. So, I had to break up the code into two chunks: the bare minimum for copying and all the rest. Then, I had to carefully flush out all data that was cached in memory, preload the small disk-copying chunk of code, and coalesce the balance of RAM. Usually, the Finder ended up with 75K of free memory and things worked as planned. But, sometimes the system would mysteriously reload the larger chunk of the Finder code, fragment the free memory, and cause another case of Disk Swapper’s Elbow.
It took me a long time to figure out what happened because we had rarely seen this in testing. There were a few bug reports of this problem that were never reproducible. The bug reports went like this: “Copied a disk, it took 20+ swaps! Tried a second time, it was fine.” The reason this was not reproducible was because we were all expert mouse users and usually skipped the crucial misstep.
When anybody first starts using a mouse, dragging is one of the more difficult things to do. It’s actually quite awkward to click down, move the mouse while holding down the button, and then release. Beginners very often accidentally release the mouse button while dragging. In the Finder, this means you could “drop” an icon you were dragging. You rarely thought about this (unless you happened to drop it over a folder and it disappeared); you’d immediately pick the icon up and continue the drag. It turned out if you dropped the disk icon during a disk copy, you’d induce the bug. Since all of the team members had been using the mouse for years by this time we rarely dropped icons, which is why we could never reproduce the problem.
To support the user’s spatial memory, the Finder always remembered where icons were located on the desktop. When you dropped the icon—even for a half a second—the Finder would dutifully record its location. The routine to save the icon’s location was, as you probably guessed, in the big portion of the Finder’s code. When this bug occurred, the Finder would carefully massage the memory for copying and then belatedly
discover the icon’s location hadn’t been flushed out. It would blindly call the routine to flush it and you now know what would happen....
I introduced this bug about 2 A.M. the morning we built the final disks. This bug was caused by a fix to a much more egregious bug, so it was definitely the lesser of two evils ... really!
It Sure Is Great to Get Out of That Bag!
January 1984
We need a demo for the intro
It took a monumental effort, fueled by inordinate amounts of chocolate-covered espresso beans (see “Real Artists Ship” on page 208), to finally finish the first release of the Macintosh software in time for the introduction. We finished with literally no time to spare, shipping the “golden master” of the Write/Paint disk to the factory at 6 A.M. on Monday morning January 16th, just a week before the introduction. By that point, most of the software team hadn’t slept for days, so we all went home to collapse.
I thought I would need to sleep for at least 24 hours, but I woke up after 6 hours with a desire to go back to Apple to see if the release held up and to see how everyone else was feeling. By 5 P.M., most of the software team had dragged themselves back for the same reason. We were lounging around in a tired daze, happy we had finally shipped the software but still not quite believing it, when Steve Jobs strode into the software area.
“Hey, pick yourselves up off the floor. You’re not done yet!”
Uh-oh, I thought. Someone must have found a showstopper in the release and we’re going to have to track it down. But that wasn’t what he meant.
“We need a demo for the intro! The Mac deserves to have a great demo for its first public showing. I want the Mac to play the theme from Chariots of Fire while it’s showing a slideshow of the apps. Plus lots of other cool stuff, whatever you can come up with. And it needs to be done by the weekend, to be ready for the rehearsals.”