Jun 00 Factory Floor
Volume Number: 16 (2000)
Issue Number: 6
Column Tag: From the Factory Floor
A Chat with Quinn "The Eskimo!"
By Quinn "The Eskimo!" and Richard Atwell, ©2000 by Metrowerks, Inc., all rights reserved
This months Factory Floor interview brings us back to Apple again for a visit with Quinn "The Eskimo!", co-creator of Internet Config. Quinn works in Developer Technical Support and took time out from helping us all write better code to sit down and talk about his world.
Richard: Who are you?
Quinn: Ah, a deeply metaphysical question, but for the moment I'll field it literally. I am Quinn "The Eskimo!" I do have a first name but I haven't used it for about 20 years. The epithet "The Eskimo!" is from an old Bob Dylan song, since covered by Manfred Mann amongst others. Listen to the words closely; it's a lot of fun.
Richard: Where are you from?
Quinn: Another deep question. I was born in Kenya and my folks are English. I grew up in Perth, Western Australia, and I'm now living in California. Perth is a big city (more than a million people) but by American standards it's very isolated. My friends and I often say that the definition of isolated is that if you travel for a thousand kilometers in any direction from Perth, the nearest big city is... Perth. Because of this isolation, the Mac developers are a close knit group.
I was very active in two areas of the community; namely, Mac developers and university support staff. There's not a lot of shrink-wrapped software development in Perth, so there was a large overlap between the groups.
Richard: What did you do after school?
Quinn: My first job after graduating was at the Computer Science department at The University of Western Australia <http://www.cs.uwa.edu.au/>. I spent a third of my time running the network, another third running the student Macintosh laboratory, and the rest programming. It was a great place to learn about the Macintosh. I handled both the user and developer side of things. And writing lab management tools is an excellent introduction to the underbelly of Mac OS!
Richard: What are your responsibilities in Developer Technical Support (DTS)?
Quinn: Since late 1995 I've been working at Apple in the Networking, Communications and Hardware group of DTS. During the day I answer developer questions, code samples and write technotes. I also review documentation, firefight, provide feedback to engineering, go to kitchens, and so on. It's a very diverse job, which keeps me from getting bored. I would find spending all day cooped up in my office coding for a single project a drag.
When I joined DTS I was cast into the role of "Open Transport guy", somewhat harrowing given that I had never programmed for either MacTCP or OT before. Hey, I knew people who had programmed for MacTCP, but I didn't actually do it myself! Since then I've supported OT through its (and Apple's) ups and downs. OT work is still a large chunk of my job, although I also support mass storage, external file systems, virtual memory, driver services, and so on.
Right now my day job is to support these existing technologies while gearing up for Mac OS X. In my spare time I'm working on a cool little tool called SystemFolderDiff <ftp://ftp.quinn.echidna.id.au/Others/Quinn/Development/>. Actually, I've been distracted from that by "MoreOSL", a C framework for implementing AppleScriptability in an application. It's a much harder job than I thought, so the code
will eventually end up as a DTS sample.
Richard: You're famous for being one of the co-creators of Internet Config.
Quinn: Mostly I'm famous for being friends with Peter N. Lewis. Peter, myself, and a mutual friend (Marcus Jager), were all friends in high school and then we went on to university together. After finishing our degrees, Peter started working on his shareware business (at first after-hours, and now full-time <http://www.stairways.com/>) and I helped him out with miscellanea like keeping his user interface honest, editing the documentation, and so on in my spare time.
The only major collaboration between Peter and I was through Internet Config <http://www.quinn.echidna.id.au/Quinn/Config/>. That project has a long history all to itself and I still have trouble groking that I've been working on it for over six years! It's great to make something that helps both users and developers. Another fortuitous outcome of the Internet Config project was that it raised my visibility sufficiently, so much so that Apple decided to hire me into DTS.
Richard: They say that necessity is the mother of invention. Was Internet Config conceived that way?
Quinn: Yes. It all started with a discussion on comp.sys.mac.comm sometime in mid-1994. Basically everyone was sick of having to enter the same Internet preferences into a gazillion (yes, that is the technical term!) Internet applications. This was particularly irksome for me because I didn't use Netscape as my web browser. In the days before Internet Explorer there was MacWeb and I got really tired of having to change the default preferences in every application that wanted to launch URLs, and then reapply the change on each new machine and also every time I reinstalled system software.
The newsgroup discussion quickly got out of hand (as they are apt to do) so Peter N. Lewis created a mailing list and he unilaterally subscribed all participants to the mailing list. The time we switched everyone over was midday, Oz time, so the US folks got a shock when they showed up to work the next day and found hundreds of messages from a mailing list to which they'd never subscribed! Fortunately we didn't get too many flame-o-grams.
The recipient list reads like a Who's Who of early Mac Internet types. With so much talent, there were bound to be some contentious issues.
Should the system support multiple users?
Should the system support layered preferences (e.g. a user layer above an application layer above a global layer)?
Should the system be an extension?
Peter and I had a strong opinion on all of these topics. We wanted to keep it simple while allowing for future expansion, so we answered no, no, and yes to the above questions. To cut short the discussion, we designed a strawman API and posted it to the list. After incorporating some feedback, we set off to implement the API.
The actual coding of the extension took very little time, probably a week or so of part-time work. The original plan was for Peter and I to write the extension while another developer did the user interface. After months of waiting for a UI, we eventually caved in and wrote that as well.
It's interesting to examine the core design decisions with six years of hindsight.
The lack of multiple user support was a problem, but I eventually retrofitted "sets" into IC 2.0 (Mac OS 8.5). Afterwards, Apple introduced Multiple Users (Mac OS 9.0) which provided another solution to this problem.
I still believe that layered preferences are too hard for users to understand and believe we made the right choice by not including them in IC.
I'm very happy that we shipped IC as an extension rather than a statically linked library. It has allowed us to update the system easily; a facility we've used extensively over the years.
I'm very disappointed by the fact that I failed to think clearly about international issues in the original IC design. For example, we should have had a clear policy regarding international-friendly text in preferences like the email signature.
For a further read about the rationale and the implementation details of IC 1.0, read my article in develop magazine.
Richard: How difficult was it to get authors of shareware networking programs to adopt IC?
Quinn: With a few exceptions, convincing developers to adopt IC was surprisingly easy. Mind you, we made numerous design decisions that smoothed the path. Our basic goal was to give developers no excuse to ignore the technology.
We started with a critical mass of Internet developers from the mailing list on our side.
IC was simple enough for both users and developers to understand.
IC was usable from all the popular development environments.
IC APIs could be used with or without having the Internet Config Extension installed.
IC supported System 6. In fact, the first version without System 6 support (IC 2.0) was released in June 1998!
We offered individual support to key players to promote adoption.
Richard: Why did you choose to release IC into the public domain?
Quinn: IC source code was put into the public domain so that, if Peter and I were killed in a 747 accident over the Pacific on the way to WWDC, another developer could pick up the development. This also helped to alleviate any fears of the technology being orphaned.
Richard: What have been the most recent Internet Config developments?
Quinn: The big news for Internet Config is its integration into Carbon. Universal Interfaces now includes "InternetConfig.h"! On traditional Mac OS, CarbonLib 1.0.2 provides glue that calls through to the existing Internet Config Extension. On Mac OS X, Apple will provide a re-implementation of the API based on CFPreferences. This gets me off the hook for future Internet Config development, which is a relief. It was fun while it lasted, but our mission was always to get Apple to adopt the technology.
Richard: Now that IC is part of Carbon, what will happen to the IC Programmer's Kit? Is there a new version planned?
Quinn: Before Apple decided to adopt IC as part of Carbon, I had finished the core code for IC 2.5, which included full Carbon support for traditional Mac OS (and some limited support on Mac OS X). Given Apple's decision to support IC in Carbon, I've had to revise my plan.
I do intend to ship a new IC that fits in with this new world order and will probably still call it 2.5. In the meantime, I would recommend that developers just use the interfaces and libraries from UI 3.3.1.
Richard: Can you tell us what about anything new regarding networking?
Quinn: In the networking space, Apple has made much progress over the last year but much of it is hard to see. Open Transport 2.5 includes many internal changes that should allow us to deliver new features more quickly.
The next year promises to be very interesting for Apple followers. It's time for Mac OS X to start delivering on its promises so it should be exciting to watch and participate in.
Richard: I know you're a staunch Pascal developer. How strong is the Pascal movement in the Mac developer community these days?
Quinn: I wish that I had some hard numbers on this, but alas I don't. Moreover, I'm not really at large in the Mac development community these days (DTS keeps me plenty busy with my real job!) so I don't even have a rough feeling for this. I can tell you that I'm regularly called in to consult on Pascal issues for Apple. For example, I've helped a number of significant developers move their large legacy Pascal source bases to Carbon.
Richard: How much longer can we expect Apple to provide support in their APIs?
Quinn: As a current employee, I can't comment on the future of the Pascal interfaces. I can say that the current situation for Pascal interfaces is pretty good. Apple creates the Universal Interfaces from an internal representation (something that looks like a C header file, but isn't) using a custom tool. This tool outputs C, Rez, Asm, and Pascal headers automatically. As long as this system is maintained, Pascal interfaces can be easily generated by flipping a switch.