Dialog Box

By Matt Neuburg, letters@mactech.com

If They Build It, Will Anyone Come?

We thought that readers might be interested in these excerpts from a thread that occurred a while back (September, 1995) on the Apple internet providers mailing list (subscription info
at http://www.solutions.apple.com/apple-internet/) between ZiyaOz@aol.com (Ziya Oz) and jon@comvista.com (MacTech Magazine contributing editor Jon Wiederspan). Obviously the details of the situation have changed in the meantime on many of these products (please bear this in mind while reading!), but the point remains valid, the more so since, as this issue went to press, Apple was actively soliciting internet strategy advice from users.

Ziya: Are you bothered by the alarming absence of serious HTTP servers (as well as site-management, clients/browsers and visual authoring tools) for the MacOS? I’ve just begun to compile a list of all significant servers, clients and tools available for Windows NT and Unix, but not for the Mac. The objective is, first, to document the severity of the situation (so that perhaps Apple can put it on its radar screen) and, second, to pressure some of the companies that completely ignore the Mac, to pay more attention to it.

PRODUCT CLIENT/Browser SERVER DEV TOOL

OpenUI Q4 95 No No

WebObjects/Next-OpenStep No No No

Netscape/LiveWire Yes No No

Verity Topic Agent/Server Soon? No No

ATT Interchange No No No

Microsoft Network/Blackbird No No No

Hyper-G Soon? No No

Vermeer FrontPage Q1 96 No No

RAD PowerMedia Soon? No No

SapphireWeb No No No

ArchiText Yes No No

Sun Java Soon? NA Roaster

W3 Website Toolkit No No No

WebBase (ODBC) No No No

NaviSoft/NaviServer Yes No No

OpenMarket No No No

O’Reilly WebSite No No No

InContext Spider No No No

Microft Explorer No No No

Wollongong Emissary No No No

Jon: I’m a bit confused by the point of your list. Do you really mean that every server that runs on another platform should also run on a Mac?

Ziya: This is not just about HTTP servers. It’s also about developer tools, browsers, site managers, editors, etc. It’s really about what I called an alarming trend on the Mac. Here’s what I mean:

At the last MacWorld Expo in Boston, I asked myself: If Ceneca was not formed by ex-Taligent people, would they have developed Page/SiteMill for the MacOS first? Would they even have developed for the Mac at all? After seeing at the Fall Internet/Boston last week the preponderance of Windows NT and UNIX, and the widespread neglect of the Mac, my answers are: No and No again.

By now it’s futile to cling to an unreasonable hope that the Mac’s role in corporate “business” computing (databases, enterprise apps, servers, CAD, network management, etc.) won’t become less and less consequential. Instead of engaging in quasi-religious discourse on this, let’s move on. Is there a future for the Mac in other areas? Certainly, in education, 3D, digital video, graphics, DTP, interactive authoring and, I’d have thought, Internet.

I watched Spindler tell the audience at the launching of the PowerMacs that “Apple would do Internet right.” Two years later, Apple has precious little to show for it, with the possible exception of the promise (and no more) of Cyberdog. How else could it be when (as I’m told by the OpenDoc evangelists) there’s no person(s) in charge of Apple’s Internet strategy, product and marketing? Of course Cyberdog is a great concept, but so were PowerTalk, QuickDraw GX, GeoPort and a host of others Apple introduced in the recent past. Is Internet important? A lot of people think so. Is it important, or even vital, to Apple? How can you tell?

Internet is essentially a client/server architecture: browsers request and servers serve data of some kind. Apple is not in the browser business; that’s owned by Netscape (and soon to be shared by Microsoft). Cyberdog can’t really change that in a 90% non-Mac marketplace. Well then, is Apple in the HTTP server business? Not really. The only commercial HTTP server of any size is WebStar. (InterServe, Web Server 4D, Netwings, etc., may change that. We’ll see.)

It’s admittedly difficult to make a great and attractive server. When your share of the OS market is 9%, though, you cannot afford not to be really great. Currently, lack of true multitasking, multithreading, multihoming and myriad file and bus I/O issues cripple the Mac against many others in the NT and UNIX world. OpenTransport may alleviate some of these problems (and Copland others), but NT and UNIX don’t have these problems now, and they sure won’t remain stagnant either.

If you were a Windows user, say, two years ago and went to the Seybold show (for the digital graphics, DTP and prepress industry), you were alienated and felt virtually ignored. Sure, you had PageMaker and CorelDraw for Windows, but all the exciting technology came out on the Mac and developers concentrated on its advantages. Yet when I went through the Internet World last week, and stopped by nearly 100 booths, the overwhelming majority of them made me (a Mac developer) feel, well, irrelevant.

That to me means that those who develop for and profit from Internet are finding the Mac not so relevant. I’ve also talked to countless “corporate” types who are searching for tools to publish internal data and conduct commerce on a large scale. Uniformly, they did not consider the Mac as a viable platform for the server/backend operations. In the end, business people are pragmatic. If your product does not offer demonstrable superiority, they’ll ignore it. (Mac is still superior to Win 95, NT and UNIX in DTP and graphics, so it’s still favored.) Since the Mac does not offer sufficient unambiguous advantage for the Internet, unless something drastic happens soon, it will eventually be ignored.

When the browser used by 80% of Internet announces that it will not release its servers, visual HTML editors or scripting tools for the Mac, you have to wonder why and worry some. When a large number of people start developing with Blackbird and you cannot even use it on the Mac, you don’t need to wonder but you’ll have to worry a little.

If your business needs to access AT&T Interchange or Dow Jones Personal Journal on-line, or conduct sophisticated commerce with an Open Market server, or integrate your internal Oracle or Sybase databases with Web access via object-oriented tools such as WebObjects, or if you want to serve disk- and CD-ROM-based search engines like the Verity Topic Servers - well, you are out of luck if you are a Mac shop.

And if the Mac is not a serious player at this early stage when tools are simpler, what happens to more sophisticated Mac technologies like QuickDraw, QD 3D, QT VR, etc., in terms of future acceptance? You think it’ll get any easier? When increasingly more sophisticated plug-ins via Navigator talk to increasingly more capable and varied servers that don’t and won’t, apparently, exist on the Mac, how does that help the Mac user? If a developer of an HTTP browser or server plug-in has the impression that there are only a few servers on the Mac and, more importantly, throughput of the Mac servers is limited compared to, say, an NT box costing about the same as a PowerMac, do you think he will worry too much about that 9% Mac market share?

OpenDoc Solves the Wrong Problem

I’ve been worried about OpenDoc. My worry is that Apple and its allies are spending a lot of time working on a technology which is ultimately going to turn out to be irrelevant to users. A columnist in PC Magazine pointed out that compound-document technologies seem oriented towards enhancing the applications of today, rather than those of tomorrow. Then, last night, my misgivings crystallized at last, in a form that I think I can explain coherently.

Consider a sheet of paper. Do I hear someone ask, “Do you mean a word-processing sheet of paper, or a graphics sheet of paper?” Of course not! Stupid question. A sheet of paper is a sheet of paper. Contrast this with the modality of most current software: before you create a new document on your Mac, you must first decide whether it’s going to be a word-processing document or a graphics document. (ClarisWorks fans should not start smirking just yet.)

The essence of pre-OpenDoc software: an application is a mode. Remember, a mode is a state which is not quite the one you want to be in to do what you want to do next, and you have to consciously think about switching to the right state. Having different states for your software is not evil in itself - the secret is to match them to the problem at hand, and to make it as easy and natural as possible for the user to move between states, without even thinking about it. When a state becomes restrictive and unnatural to the problem at hand, then it becomes a mode.

Back to that sheet of paper. I pick up a pencil, and write some words in one part (a word-processing part!), draw a table with some numbers in another part (a spreadsheet part!) and put a graphic in another part (a graphics part!).

So far, so OpenDoc. Now, just for fun, I want to put a graphic into one of the table cells. Will OpenDoc allow me to do this - use the tools of one part to work on another part? No it won’t. Whereas, on a sheet of paper, every tool (pencil, pen, charcoal, eraser, paintbrush, typewriter, finger, whatever) works in some way on every part.

Hence, the essence of OpenDoc software: a part is a mode. Sure, a spreadsheet part could allow embedding of other parts within its cells, but you can appreciate that this only gets you a little closer to the modelessness of a sheet of paper.

So how should we approach the problem instead? The beauty of a sheet of paper is not just the variety of ways in which you can make marks on it, but that these different ways can interoperate - a mark made by a paintbrush can freely abut or even overlap a mark made by a pencil, unlike OpenDoc, with its demarcation of the document into parts. If you could make the granularity of parts fine enough, you might manage it, but I don’t think the part idea is set up to deal efficiently with a page containing hundreds of parts.

In short, the problem needs to be rethought in terms of the basic technology for putting marks on the computer screen. In order to achieve true, paper-style modelessness, different tools need to be able to operate on a common graphical representation of their data.

OpenDoc is a step in the right direction, in the sense that building a very high tower is a step towards getting into outer space - in short, it’s a step that I don’t think is worth taking. I think the effort would be better expended on solving a larger chunk of the problem, on the rockets that will take us closer to true modelessness.

Lawrence D’Oliveiro, ldo@waikato.ac.nz