Years ago, starting in 1989, i worked at SoftQuad Inc. in Toronto. I got there because SoftQuad sold a text formatter based on troff (sqtroff). But they also sold an SGML structured editor, Author/Editor.
I was at a conference, ACM SIGIR, on information retrieval; I’d written a text retrieval package that was moderately popular at the time, lq-text.
In the last afternoon i got a call from the office saying I had to leave the conference early and go to support a sales effort somewhere or other. So I was ready to leave for the airport when there was an announcement.
We interrupt the schedule for this conference because an important program has been released, NCSA Mosaic, and its authors, two graduate students, are here to demonstrate it.
So I got to see Eric Bina and Marc Andreesen demonstrate Mosaic, the first cross-platform graphical Web browser.
I went back to the office afterwards, and what I’d seen changed the company.
A colleague, Jon Cummings, got Mosaic working using our 48Kbaud Internet connection. We were able to show Yuri Rubinksy, then president of SoftQuad and a leading proponent of SGML. It was exciting: the Web was based on HTML, which was in turn based on SGML. We could do this!
In the early Spring of 1994 Yuri and I went to Champaign, Illinois to meet with Joseph Hardin of NCSA. By that time we had a version of NCSA Mosaic with SoftQuad’s SGML parser in it. Yuri had also been discussing Web accessibility with people, and I’d been posting in Usenet newsgroups (people confused me with a Liam Quinn from Toronto who was also posting in the same groups!).
Yuri had recently co-authored a novel (with Marc Giacomelli) called Christopher Columbus Answers All Charges; it was the first book to be published in a regular trade edition, in large print, in spoken word, and in Braille, at the same day, made possible because of using SGML text markup.
At NCSA we met with Joseph Hardin. Yuri talked about accessibility needs; we discussed how Braille devices were narrow, getting a fixed number of characters per line, and how this meant that wide tables had to be turned round, or transposed, so they would fit.
There had been a heated discussion between Yuri and Dave Ragget about table headings in HTML. Dave had published an IETF draft specifying tables, but it did not include headings. Joseph came to agree, as a result of that meeting, that row and column heading markup was needed: what we have today as tr and th elements.
Although that part of our trip was a success, and perhaps was some small part of getting table headings into HTML (I later argued for them at IETF HTML Working Group meetings too), Joseph, it turned out, really wanted something else.
It turned out that Marc and Eric had just left NCSA and started a new company, Mosaic Communications Corporation, with the intent to sell the Mosaic browser.
As you can imagine, legal action ensued, and they renamed themselves Netscape.
Joseph’s worry was that Eric and Marc had been talking about moving away from HTML and using a word processor format called RTF instead. Forget the “view source principle”—files in the RTF format can be deciphered but are not really readable. Worse, RTF builds in formatting in a way that HTML does not. So this would be a huge backwards step for accessibility.
Joseph felt that if a reputable SGML company like SoftQuad were to make an HTML editor for the Web, the story that HTML was standards-based would be made stronger. RTF was an industry standard but SGML was an offiial ISO standard. The editor, he said, would need to be free.
The compromise was that we would build a commercial HTML editor and make a free (zero-dollar) downloadable version available that was fundamentally usable enough not just to be a marketing ploy.
Then Joseph asked whether Yuri could come to the first ever Web conference in Geneva in six weeks’ time. The hotel was oversold, the venus was full, but Yuri agreed to go.
*
Back in Toronto, Yuri Rubinsky, Joe Davidson, and others [who? Steve Downie maybe?] worked on a specification for the new editor. It was of course going to be a cut-down version of our $1200-per-seat Author/Editor.
I was head of our tiny customization/contracting division at the time, and soon had the enthusiastic support of the two programmers who reported to me full-time. The Web was exciting and they were super eager to work on the new program.
Soon the head of documentation, Rodney Boyd, joined in, and started writing a manual. The head of development warned that this was not an official project and anyone caught working on it in office hours would be fired. So people stayed late into the evening.
Yuri went off to Montreaux in Swizerland a week or two before the conference. HoTMetaL wasn’t finished but it was close, and suddenly we had interest from sales and marketing. He sent a fax from his hotel:
 |
Fax from Yuri Rubinsky to Liam Quin (then called Lee Quin) about the new (and as yet unnamed) HTML editor. Some annotations by Liam at the time are visible in red ink on the fax. |
Yuri, I think, came up with the name HoTMetaL, a pun based on the letters in HTML and on the fact that SoftQuad started out as an offshoot of Coach House Press in Toronto, a left-wing arts publisher that still occasionally operated a hot-metal typesetting machine made by Linotype.
We had a tight deadline, and the conference was approaching. There was still opposition from the head of development internally, but the Vancouver office had made a new version of Author/Editor that was called HoTMetaL and was very restricted, so people could not get the full $1200 product by breaking the restrictions: the code simply wasn’t there.
In the documentation group, Rodney was going on a much-needed on vacation. Before he left, he said that he was having problems documenting the URL editor that I had designed. At that time URLs were entirely new, so it had drop-downs for choosing http or ftp or whatever, and a place to enter the hostname, and so on. I said to him, can you document it in the way you think it ought to work? Then we’ll implement it that way. Rodney stared at me as if I’d said, point to all the chocolates you want in the shop and we’ll buy them for you. But that’s what happened, and when he came back from vacation two weeks later, HoTMetaL had shipped with a URL editor based on his writing.
*
When Yuri got to the conference in Geneva, HoTMetal was not yet ready for download. Some of us worked very late to try to make it ready, and in the end it was available I think three days after the end of the conference.
At the closing session of the conference, Yuri made the announcement. Actually, it was in a panel discussion (I was told afterwards: I was still in Toronto).
Someone had said that the strong anti-commercial attitude of the early Web needed to change. Yuri said, I agree, I have felt this attitude so strongly that I didn’eel able to make the announcement I came here to make, that SoftQuad, the leading SGML company, is making the world’s first commercial HTML editor available for free download right after this conference.
He got a standing ovation, partly for the announcement itself and partly for the clever way in which he made it. But that was Yuri all over.
Back in Toronto, we uploaded the software to a Web server and it was soon time to start on the commercial “Pro” version. At which point the head of development took over, seeing that HoTMetal was wildly successful. A lot of the early Web was like this: skunkworks projects that were so successful they got adopted officially.
HoTMetaL Pro shiped in time for the second Web conference in 1994, held in Chicago.
The head of development had been afraid it’d be a drain on his already-limited resources, but within a year HoTMetaL Pro was a third of the company’s revenue.
Within two years, Netscape included a free HTML editor, and sales started to plummet. But we were the first.
No comments:
Post a Comment