AMosaic

~ the eulogy ~

Three people wrote AMosaic and all three were named Michael. Michael Fischer was at Stony Brook, Michael Meyer was at Berkeley, Michael Witbrock was at Carnegie Mellon. They started in 1993 on a port of NCSA Mosaic to the Amiga, running pre-releases through that autumn at a pace of roughly one a week, and shipped version 1.0 on 25 december 1993. It was the first graphical web browser the Amiga ever had. NCSA and the University of Illinois had nothing to do with distributing it.

The revision history is still archived and it reads like a diary of a very fast few months. Pre-release 3 to 4 on 3 october. 4 to 5 on the seventh, 5 to 6 on the thirteenth, 6 to 7 on the seventeenth. Prerelease 8 in late november, 9 on 18 december, 10a on the twenty-third, and then 1.0 two days later. Version 1.1 landed on 13 january 1994 and 1.2 on 18 june 1994, which is where the official page froze forever, still describing 1.2 as the current version and carrying a 1994 copyright line.

Commodore filed for bankruptcy in April 1994, four months after AMosaic 1.0. So the browser's entire useful life was spent on a platform whose manufacturer no longer existed, sustained by owners who were not going to buy another machine and developers who kept writing software for them anyway. That is a specific kind of scene and the Amiga had one of the most stubborn ever assembled.

I came online in 1998 on a PC with a modem, which means the Amiga question was already settled before I got a vote in it. Amigas had been around in Turkey earlier, mostly as game machines in the hands of people slightly older than me, and by the time I had a phone line to dial with, those machines were in cupboards. Getting AMosaic running took AmigaOS, a TCP stack, MUI and a working internet account in about 1994, and the number of people in the country who had all four at once would not have filled a room.

~ the dossier ~

~ what it could do that mosaic could not ~

The interesting claim is that the port was better than the original in specific places, because AmigaOS had subsystems Unix simply lacked. AMosaic was built on MUI, the Magic User Interface, which meant a user could change the fonts and the colors and the background patterns of the browser itself without touching the program. Nothing on X11 in 1994 offered that, and nothing on the web offered it either until CSS user stylesheets, years later.

Image handling worked through AmigaOS Datatypes. Instead of the browser needing to know how to decode a format, it asked the operating system, so adding support for a new image type meant installing a system plugin and restarting nothing. Workbench 3 users had to install the ZGIF DataType before inline GIFs would render, which is the trade-off in one sentence. The other feature was ARexx, the Amiga's inter-application scripting language, wired straight into the browser: you could write a script that told AMosaic to fetch a page and hand it back as plain text, or have AMosaic call out to an external bookmark manager. That is a scriptable browser with an automation API, in 1994.

The three Michaels wrote this up as a paper, User Extensibility in Amiga Mosaic, which Witbrock presented at the Second International World Wide Web Conference in Chicago in october 1994. The Amiga press caught up the following spring, when AMosaic took the cover of Amiga World for march 1995.

~ the machine underneath ~

Nothing about connecting an Amiga to the internet in 1994 was casual. AMosaic needed AmigaOS 2.0 at minimum and a TCP implementation, of which there were at least three incompatible ones in circulation, with AmiTCP 3.0 beta 2 and AS225 Release 2 the two you were most likely to be handed. Separate builds existed for different stacks. MUI 3.8 had to be installed. Full image support wanted Workbench 3. Every one of those was a download over a modem from an FTP site you had to already know about.

What you got at the end was a machine from 1985 rendering the World Wide Web. That is the whole appeal and it is why people did it. The Amiga had been designed years before anybody had a reason to want a browser, and it turned out to have exactly the datatypes and the scripting hooks that made a good one, which the people who built it could not possibly have planned for.

~ what it became ~

AMosaic never reached a finished 2.0. A pre-release exists, the source eventually turned up on Aminet as a preservation upload, and that is the end of the line. What happened instead is that the MUI browser idea got rewritten from nothing. IBrowse, described by its own history as the rewritten follow-on to Amiga Mosaic, released 1.0 on 28 july 1996 and immediately became the browser Amiga owners actually used.

By early 1997 the Amiga magazines had a browser comparison to run and AMosaic was not in it. IBrowse was called the king of Amiga web browsers, with AWeb-II and Voyager NG behind it, and the old Mosaic port got a sentence describing it as simplistic, capable of fonts and images and not much else. It had been three years. The web had grown tables and frames in that time and the first browser on the platform could not do either. Its reward for going first was to become the baseline everyone else was measured against and then dropped from the comparison.

@}-->--

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

« previous  ·  random  ·  next »