tkWWW

~ the eulogy ~

Joseph Wang announced tkWWW to the www-talk mailing list on 25 july 1992, calling it a first pass at a browser for the World Wide Web using the Tk library, warning that it was very buggy, and suggesting people be careful. That message is still readable in the W3C archives. It landed a little over a year after the web itself was announced, which puts tkWWW among the first handful of graphical browsers anyone wrote, alongside ViolaWWW and well before Mosaic.

The work happened at MIT, under Project Athena and the Globewide Network Academy. The choice that defined the thing was Tcl and Tk: the network layer and the HTML parsing were compiled C sitting on the CERN library, but the entire user interface was an interpreted script that anyone could open and change. Wang wanted the browser to become a replacement for rn, the news reader, and eventually a swiss army knife for everything you did over a network. Word processing and mail reading were both on the roadmap.

It got further than most people remember. Version 0.5 in february 1993 added multiple fonts, including Asian ones. Version 0.7, on 1 may 1993, made tkWWW the first WYSIWYG HTML editor on X11, using code originally written by Nathan Torkington. By 0.9 it was in beta, handling character styling and partial image support, and the GNU Guile project extended it so it could be scripted in Scheme. Linux distributions of the period shipped it in the default set, which is how most of its users got it without ever choosing it.

The final release was 0.13 pre2, dated 2 april 1995. I was eleven years old in Ankara that spring and the nearest thing to a network in my life was a telephone. My first modem was three years away, and by the time I opened a browser at all the argument tkWWW had been part of was settled: browsers read, editors wrote, and nobody shipped both in one window.

~ the dossier ~

~ what it did first ~

The editor is the claim worth holding onto. In may 1993 you could open tkWWW, type into a page with headings and links behaving the way they would when rendered, and save it as HTML. Netscape did not attempt anything comparable until Navigator Gold in 1995, and Tim Berners-Lee's original browser on the NeXT had been an editor too but almost nobody outside CERN ever ran it. For about two years tkWWW was the practical answer to the question of how a normal person writes a web page.

The Tk foundation also let it do something no browser does now. With the htext widget you could embed Tk code directly into a document, so a page could carry interface elements that ran on arrival. This was years before JavaScript and it was completely unguarded, because Safe-Tcl did not exist yet. A page could execute whatever it liked with the privileges of whoever opened it. The feature that made tkWWW interesting is the same one that would have made it indefensible at scale.

~ the offspring ~

Two things grew out of it. At the University of Chicago, the Biological Sciences Division took tkWWW 0.9 in the summer of 1993 and built Phoenix, a browser and editor with Kerberos support and builds that ran outside Unix, on Windows and the Mac. Phoenix development stopped in may 1995, a month after its parent, because by then there were plenty of tools that did the same job.

The other was the TkWWW Robot, presented by Scott Spetka at the Mosaic and the Web conference in Chicago in october 1994 and built at the Air Force Rome Laboratory. It was one of the first web crawlers, and it worked by finding what Spetka called web neighborhoods, logically related pages near a starting point, returning the result as a list of bookmarks. It could only follow two links deep from the original homepages, which tells you roughly how big the web was that year.

~ what got buried with it ~

tkWWW was criticized at the time for real gaps. No HTML forms, and no stop button to interrupt a transfer, which on a slow line was not a small omission. Mosaic had those, had a company behind it, and had the look people wanted, so the audience went there and did not come back.

What went with it was the assumption that a browser is a two-way instrument. Every mainstream browser since has been a reader with a text box bolted on, and writing for the web moved into separate software owned by someone else. Thirty years later the tool most people use to publish a page is a hosted editor on a platform, which is a long way from a script file on your own disk that you could open and change while the program was still running.

@}-->--

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

« previous  ·  random  ·  next »