Previous Issue | Search TidBITS | TidBITS Home Page | Next Issue

TidBITS Logo


After a few days of holiday relaxation and a Thanksgiving dinner that couldn't be beat, we bring you another issue of TidBITS featuring Tonya's PageMill review and the second half of Adam's interview with Peter Lewis. Also check out the MailBITS this week, with news of Apple buying into AOL, Kaleida being closed down, yet another Netscape 2.0 beta, new versions of Fetch and NewsWatcher, and a suggestion for constructing a custom computer workstation.


Copyright 1995 TidBITS Electronic Publishing. All rights reserved.
Information: <> Comments: <>

This issue of TidBITS sponsored in part by:


Apple Buys a Piece of AOL -- Last week, Apple announced it exercised existing warrants with America Online (acquired as part of a 1992 agreement) to purchase a 5.1 percent stake in AOL. The purchase totals two million shares and cost Apple about $12.5 million, but has an estimated value of about $160 million. Although most analysts see this as a good investment decision, it does raise questions about Apple's future plans for eWorld. [GD]

Kaleida Closed Down; Taligent to Follow? On 17-Nov-95 Apple and IBM announced the closure of their joint venture Kaleida Labs, and plans to transfer the multimedia programming language ScriptX and other Kaleida technologies to Apple. Most Kaleida employees are being offered jobs with IBM or Apple, and both companies retain full licensing rights to Kaleida technologies. Apple is expected to roll ScriptX into tools for QuickTime and QuickDraw 3D. Neither company cited reasons for closing the venture, but reports have hinted at financial difficulties and problems due to Apple's and IBM's different technology priorities and corporate cultures.

Additionally, there are reports Apple, IBM, and Hewlett-Packard may be considering closing down Taligent, another joint-venture formed to design object-oriented operating systems for future computing platforms. These reports follow the resignation of Taligent president Joe Guglielmi in September, and the unfortunate death of acting president Richard Guarino of a heart attack in late October. [GD]

Netscape 2.0b3 Sneaks Out -- As of this writing, no official notice has appeared on Netscape's Web pages, but version 2.0b3 of Netscape Navigator appeared on Netscape's FTP sites in the middle of last week. This new beta expires 04-Feb-96 and is mostly a set of bug fixes and relatively minor changes, although Navigator's mail handling is improved and it (finally) loads text before displaying background images. The archive is about 2800K; if you plan to download this release, check out the release notes beforehand to see if you need it.

Of possible interest to developers, Netscape has also posted a brief SDK for creating Netscape Navigator plug-ins, including design specs, sample code and include files, plus three simple example plug-ins. [GD]

Fetch Updated -- Jim Matthews recently released Fetch 3.0, the latest version of the popular $25 shareware FTP client for the Mac. Fetch 3.0 retains the look of previous versions, but has changed to support multiple connections and bookmark lists. Other important new features include drag & drop, better AppleScript support, Internet Config support, better support for various firewall technologies and Open Transport, plus persistent directory list caching. Overall, it's an excellent and useful upgrade. [ACE]

NewsWatcher 2.1.1 Available -- John Norstad has released version 2.1.1 of his widely-favored Usenet news client NewsWatcher, which offers a number of bug fixes and memory improvements. The update also removes the previous limit of 16,000 groups in the full group list, and fixes a compiler bug that caused last week's version 2.1 to crash on 68000-based Macs (SE, Plus, etc.).

Brian Clark has also released version 2.1.2 of his Yet Another NewsWatcher, based on John Norstad's NewsWatcher code. YA-NewsWatcher features binary and anonymous remailer posting, article filtering via kill files, and article sorting. Brian doesn't plan to make further public releases of YA-NewsWatcher, but it has proved to be a useful and reliable program for people who need its special features. [GD]

James L. Ryan <> writes:
As an alternative to IKEA's Jerker desks, I'm looking into building the exact configuration I want by using the component system of shelving from InterMetro Industries. These finished metal components are of high quality and are exceedingly strong and easily put together. For slightly more than $300 (and about three hours work) I put together a professional-appearing 13' wide by 3' deep desk for my computer and related effluvia such as a printer and scanner. My setup includes six 3' wide and 1.5' deep shelves and six 4' wide by 1' deep shelves. The specs on the shelves say that they can support 300 pounds each. I am extremely happy with what I was able to put together with a minimum of time and effort. By the way, the only tool needed was a tubing cutter.

InterMetro Industries -- 717/825-2741 -- 800/433-2232

PageMill: Adventures in WYSIWYG Land

by Tonya Engst <>

PageMill 1.0, Adobe's much-anticipated WYSIWYG Web page creation tool, shipped a few weeks ago, and I wasted little time in trying it out. PageMill runs under any version of System 7, and Adobe recommends using it on a 68040- or PowerPC-based Mac with 6 MB of free application RAM, although you can scrape by with a minimum of 3 MB of application RAM. PageMill requires at least a 4-bit (16-color) monitor, so it will not run on a monochrome display.

PageMill works like an early version of PageMaker without paragraph styles. It also offers modern drag & drop features, provides plenty of optional keyboard shortcuts, and enables you to rapidly assemble a page from pre-existing parts. PageMill works well as a prototyping tool or as a way to quickly set up the overall look and feel of a new Web page or site.

PageMill's interface is easy and elegant, with the exception of the toolbar, which should have bigger buttons with more blank space between them. The hardest thing to figure out from the interface is how to create links, but if you take the time to skim the short manual, you should easily master all that PageMill can do. The manual, though professional and understandable, is written somewhat breathlessly, as though the writer tried to explain everything in ten minutes or less.

PageMill supports HTML 2.0 (basic options for a Web page that will reliably work in all Web browsers), and a smattering of Netscape extensions and proposed HTML 3.0 tags. PageMill shows high-ASCII characters as characters rather than named entities, making it usable for creating a Web page in European languages.

PageMill does not support tables, and this has disappointed some people enormously, perhaps because tables are difficult and tedious to hand code. It supports colored and tiled backgrounds, and can create the onscreen front end to a form. It also has a Raw HTML style which you can use for unsupported tags; unfortunately it has no macro or glossary features to speed up entering unsupported tags.

PageMill can directly use GIF and JPEG images or it can convert PICTs to GIFs. It can change a GIF's background color to transparent or add interlacing. PageMill offers a nice set of tools for making image maps, and makes it easy to associate different areas in a graphic with different URLs.

Still Raw in Places -- PageMill 1.0 may have been pushed out too fast. For instance, its font for body text is too small. The font looks like Times 12-point, and there's no way to change it, making PageMill inexplicably difficult to write in (though you can paste or drop text in from another source). PageMill has no text editing tools, not even a Find/Replace feature or a spelling checker. I've also heard too many reports about problems with PageMill crashing, though these reports usually end with a comment to the nature of "but PageMill is still my favorite program ever." PageMill suffers from a number of technical problems that have disappointed many serious HTML coders, and I'll talk a little more about those problems in a bit.

Mac users are sophisticated in their expectations of word processing and text editing programs, and we often transfer those expectations to HTML tools. These expectations are much of what brought on programs like Word 6, so I'm not too upset at PageMill for its lack of text formatting and manipulation tools. I am disappointed at PageMill's inability to act as an FTP client. It's easy to download, set up, and use an FTP client, but PageMill should hold new users' hands through this often traumatic step. At the very least, PageMill could interact with Anarchie or Fetch to ease the process.

First Date -- To find out if I could integrate PageMill into my HTML authoring tool set (I use Nisus Writer 4.1, which ships with excellent HTML macros), I tried using PageMill to make minor changes to the TidBITS home page, called default.html. The TidBITS site lives on King, a Power Macintosh 6150. The URL, in case you were wondering, is:

Maddeningly, US West hasn't yet set up a direct Internet connection to TidBITS's new location (in fact, after eight weeks they haven't even hooked up a second phone line), so I work remotely and use Anarchie to upload completed Web pages to King. I have a complete copy of the TidBITS Web site on my hard disk, so when I change a file on the site, I first change it locally and then upload it to King. The TidBITS Web site currently has 100 or so files, perhaps 15 on the main level, with the rest nested more deeply.

I worked on a copy of default.html, and I was glad I did because I set my PageMill Local Root Folder preference after opening default.html, and PageMill incorrectly changed all the relative link paths in the document. Apparently, I should have set that preference before opening default.html. After solving that problem, I revelled in what PageMill does well by quickly making a few wording changes, adding an item to the bulleted list, and adding a few links.

No matter what the benefits of coding in HTML may be, HTML is an uptight, rigid system where every typo matters. PageMill is HTML in a t-shirt with her hair down. PageMill enables you quickly and visually play with ideas, and it helps you answer questions like: "what part of this paragraph should be emphasized?" and "should I stick a horizontal rule here?"

I experienced one oddity where linking information disappeared after a save, but I put the link back, saved again, and the link stuck around.

The Importance of MacWeb -- I checked my PageMill page in MacWeb, and found two instances of a space missing between two words. PageMill showed spaces (as did Netscape), so I hadn't noticed the problem. One missing space came directly after a <STRONG> tag; the other before an Anchor tag. I used Nisus Writer to look directly at the HTML, and - sure enough - the spaces were missing. It's always good to check Web pages in the less-error-tolerant MacWeb - you never know who might read your pages, or what browser they might use.

The Double-<BR> Problem -- Using PageMill was faster and more fun than using any HTML authoring system I have tried - and I've tried a great number of them. Before merrily posting my revised page to King, I took a close look at the HTML created by PageMill. This seemed prudent, because the PageMill-Talk list has had a flurry of complaints about PageMill altering existing HTML code without asking.

As I expected from paying attention to PageMill-Talk, PageMill removed all my <P> tags (tags for a new paragraph) and replaced each one with two <BR> tags (tags that start a new line). Apparently, the reasoning for this underhanded removal is that Netscape displays additional blank lines for each contiguous <BR> tag, but ignores extra contiguous <P> tags.

I can't say for sure about the wide world of Web browsers at large, but MacWeb and NCSA Mosaic both ignore extra contiguous <BR> tags. Besides the technical problems with PageMill's decision to replace <P> tags with <BR> tags, inexperienced Web authors my run afoul of this "feature" by unrealistically expecting that what they see in PageMill (white space created by pressing Return) will appear the same way in lots of different Web browsers.

Tag Confusion -- This leads to a problem that appears with increasing regularity on HTML-centric mailing lists. In their rush to learn HTML, many HTML authors haven't learned what tags go with HTML 2.0, and - as a result - have a murky understanding of how to make their pages look good everywhere. PageMill won't help Web authors sort out these issues; if anything, it makes them worse because knowing that a certain tag is a Netscape extension doesn't help you figure out how to implement (or avoid implementing) that tag in PageMill.

Can HTML be WYSIWYG? Although some people still prefer tags, the word processing industry made a fairly smooth transition to WYSIWYG word processors. PageMill, an early WYSIWYG HTML editor, makes an important step toward WYSIWYG, but HTML and the Web pose more complexities than the word processing transition to WYSIWYG did.

By and large, if you create a layout onscreen in a word processor, your printout will match that layout. If a word processor doesn't have a format (like columns), there's little or nothing you can do about it, no matter what printer you use. An experienced word processor user knows all the formats the program offers; an experienced (or even capable) user of a WYSIWYG Web authoring tool must know how the formats offered by the program interact with the larger feature set made available by various flavors of HTML.

This new wrinkle makes creating a useful and usable WYSIWYG Web authoring tool rather difficult. Some have suggested that PageMill's catering to Netscape users in its use of <BR> tags and certain supported extensions is fine, because Netscape is the most popular browser. I find this a tired argument, perhaps as tired as the argument that catering to Netscape users is a short-term effort to gain market share.

Without several layers of additional sophistication, a WYSIWYG authoring tool cannot keep up with all the available HTML browsers and tags, nor can it make everyone happy. For example, by implementing the <BR> tags as it does, PageMill helps naive Web authors create white space. Of course, PageMill's big, still-untapped, market is naive Web authors who have neither time nor interest in using anything but Netscape.

In the long-term, it's possible that every double-<BR> put out by PageMill will have to be fixed (most likely with starting and ending <P> tags that surround each paragraph). Perhaps PageMill's designers figure they'll either ship PageMill 3.0 with an automatic converter or that future browsers will work okay when they encounter the incorrect but functional double-<BR> coding technique.

Will I Use It? I will use PageMill to prototype new pages and to experiment with design changes before implementing the HTML in Nisus Writer. Until PageMill leaves my <P> tags alone and give me more control over the underlying HTML, I cannot use it for final Web pages. What worries me is that SiteMill (due out soon), is based on PageMill but lets you manage the links and placement of files in a site. SiteMill's planned features sound wonderful, but if it forces me to put all my HTML documents into PageMill format, I won't use it. I don't want double-<BR> tags, and (based on comments written to PageMill-Talk) I'm concerned that PageMill will make other unwanted changes.

Professional typesetters were driven up the wall by people who used early versions of MacWrite to print newsletters to PostScript printers using Geneva for body text and New York for headlines, without anything resembling ligatures or kerning; experienced HTML authors may now be driven equally crazy by enthusiastic PageMill users. PageMill's energy and simplicity stand to make it enormously popular and influential.

PageMill-Talk -- In case you were wondering, you can chat about PageMill on the PageMill-Talk mailing list, and you can search back postings to the list.

You can also read about PageMill or order PageMill on Adobe's Web site:

Adobe -- 800/411-8657 -- 206/628-2749

InterviewBITS with Peter Lewis, Part 2

by Adam C. Engst <>

This week we conclude our interview from TidBITS-304 with Peter Lewis <>, one of the best-known Macintosh Internet programmers.

[Peter] I'm not particularly fond of either. They seem to announce new things they are going to roll into the browser every week. The last thing I want in a Web client is for it to be larger than a full Microsoft Office install! I'm also not sure who the target audience is for Web versions of VRML, Acrobat, etc. Most people I know are limited to modem speeds. I don't even let it download pictures; it's just too slow, and even text is too slow most of the time. Maybe in a decade or so we'll all have ISDN or a T-1 to our houses, but in the meantime?

On top of that, I don't want to see any one company have too much control over the commercial use of the net, and Netscape would certainly love to control the whole game.

Personally, I use MacWeb. It's the most Mac-like browser available. It's getting a bit old these days, but I've heard there is a new version that's very cool, so I'm looking forward to that.

[Peter] Do you think that's limited to the U.S.A.? We already have our state government rushing to enact legislation before the federal government does (our state and federal governments are involved in a bitter feud).

For myself, I'm opposed to any form of censorship. On top of that, any legislation is obviously going to be unenforceable, and I'm very much against unenforceable legislation - it puts too much power in the police force to choose who they want to pick on.

[Peter] It's a joke. Of course I have a copy of MacPGP (I don't use it much, my communications are very rarely secret enough to require that kind of messing around). I saw a great signature the other day, it was a three line chunk of Perl code that did RSA encryption (although, naturally, the code looked like it had already been encrypted). The whole thing seems to be caused by amazing egotism on the part of the U.S. government - do they really think only U.S. mathematicians can write encryption algorithms? What's even more insane is the fact that it's legal to export the algorithm, just not the code, so they obviously think there are no programmers outside the U.S.A. Weird.

[Peter] The normal C argument goes like this: "Everyone else is using C, so therefore it must be good." Every Mac user should recognize that statement in a slightly different form, "Everyone else is using PCs, so therefore they must be good."

Basically, I continue to use Pascal because I'm more productive in it. I consider using Pascal to be a strategic advantage, doubly so when compared to C++. I've been reading a C++ book recently (know thy enemy), and every time I turn the page I see new ways to make tiny errors that are catastrophic and impossible to debug. I'm amazed that anyone can produce a working C++ program.

However, programming in Pascal does cause occasional problems. The Apple interfaces tend to be quite broken. I wanted to try out QuickDraw GX, and it took a year of new versions before they finally got one that I could hack to work with Pascal. By then I'd given up on GX. In some ways this is actually a good thing for me: I have way too many projects and not enough time to do them all, so not being able to work with GX or OpenDoc is helpful for limiting my options.

[Peter] I've never really had a favorite model. I quite like my current 7100/66 (the best price/performance ever - Apple gave it to me for free). I have a theory that people become fond of the models they own. I don't pine for a new machine, I want more disk space more than anything else at the moment.

[Peter] Absolutely. It would be interesting to survey the eleven people who won last year's award and see how their Macs affected their development. My 7100 is now the center point of my home office - without it I would probably have to have stayed working for Curtin for several more months before I could set up my office. I'm sure that's exactly what Apple had in mind when they made those awards.

[Peter] In future models I'd like to see more colours; grey and beige are so boring. Apple needs to sell computers the same way the car industry sells cars. Have you noticed how lots of Apple's new Macs have started looking like PCs (OK, not quite that bad). The Mac is just a silly box to hold the monitor. I like some of Apple's innovations in the CPUs with built-in monitors, but I prefer the monitor separate (since they're unlikely to build one-piece computers with large monitors). Apple should have more fun with their designs - Macs are fun to use, they should be fun to look at.

[Peter] From what I've seen, the clones are even more boring, most of them look like PCs. Maybe in a few years when (if?) Apple loosens up a bit on their clone deals we'll see some small companies producing designer Macs.

I can't buy a clone Mac yet, since they haven't set up in Australia as far as I know. I'd also be concerned about their ability to support remote locations like Perth at this early stage. I think I'll wait and let them build up some infrastructure first. That said, I had the opportunity to visit the Power Computing folks in Austin earlier this year. They seem to be well in control of the situation, and very much into providing good service and value for money. I recently asked several smart people about their Power Computing machines, and they all said that the machines were as good as Apple's. So if you live a bit closer to the action than I do, I think you should definitely look in to it and compare the prices.

[Peter] Well, my personal Internet connection is a 28.8 Kbps modem, permanently connected to the Western Australian hub. I have an Ethernet running around my house, and a 386 running Linux as my Ethernet-to-PPP gateway (another interesting project that might have been funded by higher shareware response is a Mac-based Ethernet-to-PPP router).

Of course, it doesn't much help having a 28.8 Kbps link while Australia's international link is completely saturated. Right now it's running at 20 bytes/second, which is pretty painful.

[Peter] The Australian backbone charges at a rate of about U.S. $0.70 per megabyte. I compared that against an international phone call at 28.8 Kbps, which comes out around $2 per megabyte. So it's about three times more expensive, but on the other hand, a 1 MB file might take six minutes instead of 14 hours. I've been thinking of going back to the bad old days of using an email-to-FTP gateway to retrieve files across the link. Fortunately most FTP sites are mirrored in Australia so it's not an impossible situation, and eventually Telstra will get around to improving the situation.

[Several people wrote in last week to point out the URL we published last week for Peter's Web site was broken (the FTP URL worked just fine). Problems with that server seem to have been resolved and both links are currently working correctly. -Geoff]

Non-profit, non-commercial publications and Web sites may reprint or link to articles if full credit is given. Others please contact us. We do not guarantee accuracy of articles. Caveat lector. Publication, product, and company names may be registered trademarks of their companies. TidBITS ISSN 1090-7017.

Previous Issue | Search TidBITS | TidBITS Home Page | Next Issue