Copland

~ the obit ~

in 1994 i was ten years old in ankara. dial-up was still four years away for me and mac hardware was not a category anyone i knew touched. copland was happening in cupertino board rooms i had no way to see. what i learned about it came a decade later from mac history retrospectives that used the project as the case study for what happens when a legacy code base tries to catch up in one release.

copland began in march 1994 as the codename for the next major version of the mac operating system. the plan was to fix the biggest weakness of system 7, which was that a single misbehaving application could take down the whole machine. copland would deliver preemptive multitasking and memory protection. it also planned a new object-oriented driver model and an appearance manager that would give mac os its first real ui skinning system.

by late 1995 the project was already slipping. every group inside apple had asked for a feature and gil amelio's team had said yes to almost all of them. the november 1995 developer release went out to around fifty macintosh developers as a driver development kit rather than a full system. what shipped was too incomplete to build against and the promised summer 1996 general release quietly moved into 1997 in press coverage.

this part i was not around for. in july 1996 amelio hired ellen hancock away from national semiconductor to run engineering. within a few weeks she concluded that copland was not a shippable operating system. it was, per her later interviews, a pile of good ideas that had never been integrated into a working whole. apple cancelled the project in august 1996 and pulled its useful pieces into planned dot-releases of mac os 8. the search for an outside replacement started immediately and ended with the acquisition of next in december 1996 for around four hundred million dollars.

~ the rap sheet ~

~ what copland was going to be ~

the pitch inside apple was that copland would erase the twelve-year gap between mac os and the windows nt kernel philosophy. windows nt already had preemptive multitasking and memory protection. classic mac os was still running cooperative multitasking on top of a memory model that dated to 1984. copland was supposed to close that gap in a single release while keeping backwards compatibility with every 68k and powerpc mac application in the field. the ambition was the reason it never shipped.

the appearance manager was the piece that leaked into shipping mac os first. it eventually landed as part of mac os 8 in july 1997, giving users the platinum look that replaced the flat gray system 7 windows. that was the entire consumer-visible legacy of copland, a facelift for the desktop nine months after apple had walked away from the operating system underneath.

~ the cracks ~

the project failure post-mortems all point to the same root cause. copland had no product manager empowered to say no. every apple team had a wish-list feature they wanted in. the operating system group tried to implement all of them, missed integration deadlines, then discovered that pieces built by different teams did not compose into a working machine. by mid-1996 the project had over five hundred engineers assigned to it and no shippable build.

ellen hancock's arrival made the diagnosis official. she was an ibm veteran who had built and shipped mainframe operating systems for two decades. her conclusion after six weeks was that copland was not one project running late. it was several projects running independently under the same codename, and no reasonable amount of additional time or money would finish it.

~ what replaced it ~

copland's cancellation is the reason apple bought next instead of shipping its own next-generation os. that decision brought steve jobs back to apple as an advisor in early 1997 and as ceo by mid-year. the operating system apple actually shipped, mac os x public beta in september 2000, was built on the openstep foundation from next with a nod to freebsd underneath. the copland engineering effort mostly evaporated into the reorganizations that followed.

copland sits alongside kaleida labs and taligent as the third failed apple software project of the mid-90s. all three were bets that apple could build its way out of the platform crisis using engineering alone. the escape route ended up being an acquisition of a company steve jobs had built while he was outside the building.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »