The Story of FrontierI wrote this piece as part of the business planning process for UserLand Software.
I thought it was interesting, a continuation of Outliners & Programming and wanted to get it on my personal site.
This is my story, told from my point of view...
An Apple product?
I started development of Frontier after leaving Symantec in the spring of 1988. It was the result of several meetings with Jean-Louis Gassee, then the top product guy at Apple. The goal was to come up with a scripting tool for the Macintosh OS, that would be similar in depth and power to the shell languages on Unix and the batch language on MS-DOS.
I designed, along with the scripting language, a new Finder, based on outlining. The scripting language was very simple, based on the syntax of Pascal, and deeply integrated into the new Finder. 2click on a script and a window opens in the new Finder. One layer, the scripting is deeply part of the OS.
I presented a prototype of the software six months later and negotiated with Apple for its acquisition. We couldn't agree on terms, so we decided to do two products, one by them, and another by UserLand, and we agreed to make them compatible.
Developing alongside Apple
I decided to go deeper and more powerful, believing that Apple would basically implement the idea I showed them. They had the source code to the Finder, I didn't.
I think it was sometime in 1990 that Doug Baron came on. For a while both of us worked on the Frontier kernel, but as we got closer to shipping Frontier 1.0, I concentrated on working in the environment, and Doug concentrated on developing the environment. That's why so much of the functionality of Frontier is implemented in scripts.
All the while, Apple was cancelling and restarting the AppleScript effort quite a few times. Eventually they settled down, and in January 1992, just as we were shipping Frontier 1.0, they held a one day developer's conference for developers on scripting in SF.
Frontier sold well for the first few months, we were building distribution (mostly mail order, but a few dealers carried the product) when in May 1992 they announced AppleScript publicly. Our sales plumetted.
We shipped Frontier 2.0, in September 1992 and Frontier 3.0 in December 1993. We tried a lot of angles to co-exist with Apple's at-first vaporous product, and by early 1994, shipping product. But we were dependent on working with other developers for much of our functionality, and for the most part the developers ignored Frontier in favor of AppleScript.
We did deliver compatibility with Apple's protocols, we actually supported them more completely than Apple did, but there was still work involved to glue our software to other products. That work fell to us or our community, but we were stretched very thin and not prospering, and our community consisted mostly of non-commercial programmers (not all though) so the results were mostly quirky and difficult to use.
In some cases we delivered superior results, as with THINK C and Quark XPress, but even then people in the Mac world of the early-mid 90s were still revolving around Apple. It was totally an uphill battle, and we couldn't capture any hilltops.
Out of gas
I ran out of gas, personally, in January 1994, and stopped working on Frontier. Dave Carlick was there at the meeting I had with Jack Russo and Doug Baron in early 1994, where I made it clear that I was finished, didn't want to fight with Apple anymore, and was leaving.
Jack and Doug and Carolyn Franz and John Baxter made a go of it, I'm not even sure what they did, I stayed far away. I appreciate the effort they made, but it was clear by May 1995 that it wasn't working. By then I had a new career as a web writer. DaveNet started in late 1994, I became a contributing editor at HotWired, one of the hottest early websites.
Sometime in the spring of 1995 Mason Hale sent me a challenging email. I don't remember exactly what he said, but it went something like this. "Someone told me I should take a look at Frontier because it's exactly what I'm looking for in a CGI scripting environment."
I barely understood what a CGI scripting environment was. Later I understood that he was right. In doing Frontier as a Mac scripting environment, we were putting together an ideal environment for Internet programming. The object database, native interpreter, multithreaded runtime, outliner, interapplication communication. All these were essential pieces, and they tied together.
At the time I was working on my second generation of website management software, AutoWeb. The experience is documented in DaveNet. I was more interested in applying some of what we learned in publishing systems built with Frontier and Quark to the new problem of website management.
At HotWired, we had a simple groupware system (the WebMonkey guys just documented it publicly on their website late last week). I wanted to do the same thing they were doing on Unix on the Mac. Lots more Macs out there, I figured!
Mason's email, the failure of the commercial effort, excitement over the Internet, and the success MacHTTP and Netscape were having, led me to the exciting turn we took in May 1995, we freed Frontier, renamed it Aretha, and uploaded the software to a website on hotwired.com.
We shipped some really buggy software! Tried a million ideas out. Web pages that contained scripts that ran on the client. A multi-user root, a BBS implemented in Frontier, a classified ads server, what else? A substantial stockpile of done or partially-done R&D.
I also did a third generation site manager, Clay Basket. The model was website development as a productivity app. But I hit a brick wall on this when I did a 1000-person site in February 1996, 24 Hours of Democracy. It's become kind of a classic website, still gets a lot of hits. It's a time capsule in a way.
Clay couldn't handle it. I realized then that website work is groupware. Then in March 1996 I figured it out. We had the perfect environment for website development, Frontier! It wasn't pretty like Clay, but it had everything I needed.
So I spent March-May doing the new framework in Frontier. In May 1996 I released Frontier 4.0. It was a big hit. I loved it. We attracted a new class of Frontier people to replace Mason and others who had gone onto other platforms.
This epiphany is also well documented in DaveNet. See three pieces, Watch This!, A Custom Cookie-Cutter and BBEdit.
I believed we had a hit, but for a variety of reasons, believed it couldn't be a commercial success on the Mac. So I hired Doug back in the summer of 1996, and then Bob Bierman in the fall, to port the Frontier environment to Windows. Brent came on in early 1997 and I hired Matt Neuburg to write two tutorials for Frontier 4. In March Matt went to O'Reilly to work on their Frontier book.
Frontier 5 will ship in late January 1998 for Mac and Windows. Whatever features we're committed to that don't ship in 5.0 will ship in 5.1 (June 1998). For the forseeable future we are committed to simultaneous Mac and Windows releases. Our core user base still uses Macs, but a transition is happening. How that goes will influence our direction.
Whew! That was a lot of writing!
PS: Here are two additional Clay Basket screen shots: #1 and #2. We miss the Clay interface and want to bring it back, maybe in 5.0, but definitely by 5.1.
PPS: Jack Russo has comments.
This page was last built on 3/29/98; 9:04:39 AM by David Winer. firstname.lastname@example.org.